13 Q. WHY DO YOU BELIEVE IT IS IMPORTANT TO USE MORE THAN ONE

Transcription

1 1 discussed throughout my Direct Testimony, that selection must be based on a 2 comprehensive review of relevant data and information, and does not necessarily lend 3 itself to a strict mathematical solution. Consequently, the key consideration in 4 determining the Cost of Equity is to ensure that the methodologies employed reasonably 5 reflect investors' view of the financial markets in general, and the subject company (in 6 the context of the proxy group) in particular. 7 Q. WHAT METHODS DID YOU USE TO ESTIMATE THE COMPANY'S COST OF 8 EQUITY? 9 A. I used the constant growth Discounted Cash Flow (DCF) model as my initial approach, 10 and considered the results of the Capital Asset Pricing Model (CAPM), and Bond Yield 11 Plus Risk Premium approach in assessing the reasonableness of the DCF results and my 12 ROE recommendation. 13 Q. WHY DO YOU BELIEVE IT IS IMPORTANT TO USE MORE THAN ONE 14 ANALYTICAL APPROACH? 15 A. Because the Cost of Equity is not directly observable, it must be estimated based on both 16 quantitative and qualitative information. As a result, a number of models have been 17 developed to estimate the Cost of Equity. As a practical matter, however, all of the 18 models available for estimating the Cost of Equity are subject to limiting assumptions or 19 other methodological constraints., Consequently, many finance texts recommend using 20 multiple approaches when estimating the Cost of Equity. When faced with the task of 21 estimating the Cost of Equity, analysts and investors are inclined to gather and evaluate 22 as much relevant data as reasonably can be analyzed and, therefore, are inclined to rely 23 on multiple analytical approaches. PUC DOCKET NO ROBERT B. HEVERT 937

2 1 In essence, practitioners and academics recognize that financial models simply are 2 tools to be used in the ROE estimation process, and that the strict adherence to any single c 3 approach, or to the specific results of any single approach, can lead to flawed or 4 misleading conclusions. That position is consistent with the Hope and Bluefield finding 5 that it is the analytical result, as opposed to the methodology, that is controlling in 6 arriving at ROE determinations. Thus, a reasonable ROE estimate appropriately 7, considers alternate methodologies and the reasonableness of their individual and 8 collective results. 9 Consequently, I believe it is both prudent and appropriate to use multiple 10 methodologies in order to mitigate the effects of assumptions and inputs associated with 11 relying exclusively on any single approach. Such use, however, must be tempered with 12 due caution as to the results generated by each individual approach. As such, I have 13 relied primarily on the Constant Growth DCF model, and used the CAPM and Bond 14 Yield Plus Risk Premium approaches as corroborating methodologies Constant Growth DCF Model 17 Q. ARE DCF MODELS WIDELY USED IN REGULATORY PROCEEDINGS? 18 A. Yes, in my experience the Constant Growth DCF model is widely recognized in 19 regulatory proceedings. Nonetheless, neither the DCF model nor any other model should 20 be applied without considerable judgment in the selection of data and the interpretation of 21 results. PUC DOCKET NO ROBERT B. HEVERT 938

3 1 Q. PLEASE DESCRIBE THE DCF APPROACH. 2 A. The DCF approach is based on the theory that a stock's current price represents the 3 present value of all expected future cash flows. In its simplest form, the DCF model 4 expresses the Cost of Equity as the sum of the expected dividend yield and long-term 5 growth rate, and is expressed as follows: 6 P= D= ' DZ '-' D^` [1] (1+ k) ' ii=x)= 7 where P represents the current stock price, Di... Do, represent expected future dividends, 8 and k is the discount rate, or required ROE. Equation [1] is a standard present value 9 calculation that can be simplified and rearranged into the familiar form: 10 k= D, (1.: s) ;.9' [2] 11 Equation [2] often is referred to as the "Constant Growth DCF" model, in which the first 12 term is the expected dividend yield and the second term is the expected long-term growth 13 rate. 14 Q. WHAT ASSUMPTIONS ARE REQUIRED FOR THE CONSTANT GROWTH DCF 15 MODEL? 16 A. The Constant Growth DCF model requires the following assumptions: (1) a constant 17 average growth rate for earnings and dividends; (2) a stable dividend payout ratio; (3.) a 18 constant price-to-earnings multiple; and (4) a discount rate greater than the expected 19 growth rate. PUC DOCKET NO ROBERT B. HEVERT 939

4 I Q. WHAT MARKET DATA DID YOU USE TO CALCULATE THE DIVIDEND YIELD 2 COMPONENT OF YOUR DCF MODEL? 3 A. The dividend yield is based on the proxy companies' current annualized dividend, and 4 average closing stock prices over the 30, 90, and 180-trading days as of May 31, Q. WHY DID YOU USE THREE AVERAGING PERIODS TO CALCULATE THE 6 AVERAGE STOCK PRICE? 7 A. I did so to ensure that the model's results are not skewed by anomalous events that may 8 affect stock prices on any given trading day. At the same time, the averaging period 9 should be reasonably representative of expected capital market conditions over the long 10 term. In my view, the use of the 30-, 90- and 180-day averaging periods reasonably 11 balances those concerns. 12 Q. DID YOU MAKE ANY ADJUSTMENTS TO THE DIVIDEND YIELD TO ACCOUNT 13 FOR PERIODIC GROWTH IN DIVIDENDS? 14 A. Yes, I did. Since utility companies tend to increase their quarterly dividends at different 15 times throughout the year, it is reasonable to assume that dividend increases will be 16 evenly distributed over calendar quarters. Given that assumption, it is appropriate to 17 calculate the expected dividend yield by applying one-half of the long-term growth rate 18 to the current dividend yield.16 That adjustment ensures that the expected dividend yield 19 is, on average, representative of the coming twelve-month period, and does not overstate 20 the dividends to be paid during that time. 16 See Exhibit RBH-1. PUC DOCKET NO ROBERT B. HEVERT 940

5 ^ 1 Q. IS IT IMPORTANT TO SELECT APPROPRIATE MEASURES OF LONG-TERM 2 GROWTH IN APPLYING THE DCF MODEL? 3 A. Yes. In its Constant Growth form, the DCF model (i.e., as presented in Equation [2] 4 above) assumes a single growth estimate in perpetuity. In order to reduce the long-term 5 growth rate to a single measure, one must assume a constant payout ratio, and that 6 earnings per share, dividends per share and book value per share all grow at the same 7 constant rate. Over the long term, however, "dividend growth can only be sustained by 8 earnings growth. Consequently, it is important to incorporate a variety of measures of 9 long-term earnings growth into the Constant Growth DCF model. 10 Q. PLEASE SUMMARIZE YOUR INPUTS TO THE CONSTANT GROWTH DCF 11 MODEL. 12 A. I applied the DCF model to the proxy group of integrated electric utility companies using 13 the following inputs for the price and dividend terms: The average daily closing prices for the 30-trading days, 90-trading days, and trading days ended May 31, 2012 for the term Po; and The annualized dividend per share as of May 31, 2012 for the term Do. 17 I then calculated the DCF results using each of the following growth terms: The Zacks consensus long-term earnings growth estimates; The First Call consensus long-term earnings growth estimates; and The Value Line long-term earnings growth estimates. 21 Zacks, First Call, and Value Line are all commonly accepted sources of earnings 22 growth rates. Furthermore, Zacks and First Call are consensus estimates ensuring that no 23 single analyst estimate unduly influences the DCF results. PUC DOCKET NO ROBERT B. HEVERT 941

6 I Q. HOW DID YOU CALCULATE THE HIGH AND LOW DCF RESULTS? 2 A. I calculated the proxy group mean high DCF result using the maximum Earnings per 3 Share growth rate as reported by Value Line, Zack's, and First Call for each proxy group 4 company in combination with the dividend yield for each of the proxy group companies. 5 The average mean high result then reflects the average maximum DCF result for the 6 proxy group as a whole. Likewise, the median high result reflects the median maximum 7 DCF result for the proxy group as a whole. I used a similar approach to calculate the 8 proxy group mean low results, using instead the minimum growth rate as reported by 9 Value Line, Zack's, and First Call for each proxy group company. 10 Q. WHAT ARE THE RESULTS OF YOUR DCF ANALYSIS? I I A. My Constant Growth DCF results are summarized in Table 4, below (see also Exhibit 12 RBH-1). 13 Table 4: DCF Results Low Growth Rate Mean Results Mean Growth Rate High Growth Rate 30-Day Average 9.06% 10.79% 13.43% 90-Day Average 9.07% 10.80% 13.44% 180-Day Average 9.13% 10.86% 13.50% Median Results 30-Day Average 9.40% 9.99% 11.22% 90-Day Average 9.07% 10.04% 11.29% 180-Day Average 9.53% 10.13% 11.43% 14 Q. DID YOU UNDERTAKE ANY ADDITIONAL ANALYSES TO SUPPORT YOUR 15 DCF MODEL RESULTS? PUC DOCKET NO ROBERT B. HEVERT 942

7 I A. Yes. As noted earlier, I also used the CAPM and Bond Yield Plus Risk Premium 2 approach as a means of testing the reasonableness of my DCF results. 3 4 CAPMAnalysis 5 Q. PLEASE BRIEFLY DESCRIBE THE GENERAL FORM OF THE CAPM ANALYSIS. 6 A. The CAPM analysis is a risk premium approach that estimates the Cost of Equity for a 7 given security as a function of a risk-free return plus a risk premium (to compensate 8 investors for the non-diversifiable or "systematic" risk of that security). As shown in 9 Equation [3], the CAPM is defined by four components, each of which theoretically must 10 be a forward-looking estimate: 11 k = rf z P ( rira -rf ^ where: 13 k = the required market ROE; 14 = Beta coefficient of an individual security; 15 rf= the risk-free rate of return; and 16 r,n = the required return on the market as a whole. 17 In Equation [3], the term (r,n - rf) represents the Market Risk Premium. 18 According to the theory underlying the CAPM, since unsystematic risk can be diversified 19 away by adding securities to their investment portfolio, investors should be concerned 20 only with systematic or non-diversifiable risk. Non-diversifiable risk is measured by the 21 Beta coefficient, which is defined as: 22 ^'J = Q x Pl,axi [4] ^ PUC DOCKET NO ROBERT B. HEVERT 943

8 1 where 6} is the standard deviation of returns for company "j'; o,,, is the standard 2' deviation of returns for the broad market (as measured, for example, by the S&P Index), and pj;,,, is the correlation of returns in between company j and the broad market. 4 Thus, the Beta coefficient represents both relative volatility (i.e., the standard deviation) 5 of returns, and the correlation in returns between the subject company and the overall 6 market. 7 Q. HAS THE CAPM BEEN AFFECTED BY RECENT ECONOMIC CONDITIONS? 8 A. Yes, it has. First, as noted above, the risk free rate, "rj', in the CAPM formula is 9 represented by the interest rate on long-term U.S. Treasury securities. As discussed in 10 Section VIII (below), during periods of increased equity market volatility investors tend ^ 11 to seek the relative safety of low-risk securities such as Treasury bonds. In addition, 12 since the 2008 Lehman Brothers bankruptcy filing, Federal policy has focused on 13 maintaining low long-term interest rates. As a result, the first term in the model (i.e., the 14 risk-free rate) is lower than it would have been absent such events. As also discussed in 15 Section VIII, the increasing trend of correlations between electric utility stocks and the 16 broad market has put upward pressure on Beta coefficients. 17 Finally, as a result of the extraordinary loss in equity values during 2008, the 18 Market Risk Premium, when measured on a historical basis, actually decreased from the 19 prior year, even though other measures of risk sentiment, in particular market volatility, 20 indicated extremely high levels of risk aversion. That result is, of course, counter- 21 intuitive. While the subsequent market rally resulted in a somewhat higher historical 22 average Market Risk Premium, it still remains below its pre-financial crisis level. PUC DOCKET NO ROBERT B. HEVERT. 944

9 1 Q. WITH THOSE OBSERVATIONS IN MIND, WHAT ASSUMPTIONS DID YOU 2 INCLUDE IN YOUR CAPM ANALYSIS? 3 A. Since utility assets represent long-term investments, I used two different estimates of the 4 risk-free rate: (1) the current 30-day average yield on 30-year Treasury bonds (i.e., percent); and (2) the projected 30-year Treasury yield (i.e., 3.58 percent).17 6 Q. WHAT MARKET RISK PREMIUM DID YOU USE IN YOUR CAPM ANALYSIS? 7 A. For the reasons discussed above, I did not use a historical average; rather, I developed 8 two forward-looking (ex-ante) estimates of the Market Risk Premium. 9 Q. PLEASE DESCRIBE YOUR FIRST EX-ANTE APPROACH TO ESTIMATING THE 10 MARKET RISK PREMIUM. I1 A. The first approach is based on the market required return, less the current 30-year 12 Treasury bond yield. To estimate the market required return, I calculated the market 13 capitalization weighted average ROE of the S&P 500 based on the Constant Growth DCF 14 model. To do so, I relied on data from two sources: (1) Bloomberg; and (2) Capital IQ 15 which are both widely accepted sources of market information. For both Bloomberg and 16 Capital IQ, I calculated the expected dividend yield (using the same one-half growth rate 17 assumption described earlier), and combined that amount with the projected earnings 18 growth rate to arrive at the market capitalization weighted average DCF result. I 19 performed that calculation for each of the companies for which data was available, and 20 then summed the individual company results to derive the estimated market capitalization 21 weighted average return. I then subtracted the current 30-year Treasury yield from that 22 amount to arrive at the market DCF-derived ex-ante MRP estimate. The results of those 23 calculations are provided in Exhibit RBH Blue Chip Financial Forecasts, Vol. 31, No. 5, May 1, 2012, at 2. PUC DOCKET NO ROBERT B. HEVERT 945

10 1 Q. PLEASE NOW DESCRIBE THE SECOND EX-ANTE APPROACH. 2 A. The second approach assumes a constant Sharpe Ratio, which is the ratio of the risk 3 premium relative to the risk, or standard deviation of a given security or index of 4 securities. The Sharpe Ratio is relied upon by financial professionals to assess the 5 incremental return received for holding a risky (i.e., more volatile) asset rather than a 6 risk-free (i.e., less volatile) asset. The formula for calculating the Sharpe Ratio is 7 expressed as follows: 8 S = trx- R I-1 X [5] 'ax 9 where: 10 Sx = Sharpe Ratio for security "x"; 11 RX = the average return ofx; 12 Rf= the rate of return of a risk-free security; and 13 dx= the standard deviation of rx 14 As shown in Exhibit RBH-2, the constant Sharpe Ratio is the ratio of the 15 historical market risk premium of 6.60 percent (the numerator of Equation [5] above) and 16 the historical market volatility of percent (the denominator of Equation [5]). The 17 expected market risk premium is then calculated as the product of the Sharpe Ratio and 18 the expected market volatility. For the purpose of that calculation, I used the thirty-day 19 average of the Chicago Board Options Exchange (CBOE) three-month volatility index 20 (i.e., the VXV) and the average of settlement prices over the same thirty-day period of The standard deviation is calculated from data provided by Morningstar in its annual Valuation Yearbook. (See, Morningstar Inc., Ibbotson SBBI 2012 Valuation Yearbook, Large Company Stocks: Total Returns Table B-1, at PUC DOCKET NO ROBERT B. HEVERT 946

11 1 futures on the CBOE's one-month volatility index (i.e., the VIX) for September through 2 November, Q. HOW DID YOU APPLY YOUR EXPECTED MARKET RISK PREMIUM AND RISK 4 FREE RATE ESTIMATES? 5 A. I relied on each of the ex-ante Market Risk Premia discussed above, together with the 6 current and near-term projected 30-year Treasury bond yields as inputs to my CAPM 7 analyses. 8- Q. WHAT BETA COEFFICIENTS DID YOU USE IN YOUR CAPM MODEL? 9 A. I considered two methods of calculation. My first approach simply employs the reported 10 Beta coefficient from Bloomberg and Value Line for each of the proxy group companies. 11 While both of those services adjust their calculated (or "raw) Beta coefficients to reflect 12 the tendency of the Beta coefficient to regress to the market mean of 1.00, Value Line 13 calculates the Beta coefficient over a five-year period, while Bloomberg's calculation is 14 based on two years of data. I also calculated Beta coefficients calculated over a more 15 recent time period to provide a more current view as to investors' perspectives with 16 respect to the systematic risk represented by the proxy group companies. 17 Q. PLEASE DESCRIBE HOW YOU CALCULATED THE MEAN ADJUSTED BETA 18 COEFFICIENT FOR YOUR PROXY GROUP. 19 A. As shown in Equation [4], the Beta coefficient is calculated as the ratio of the standard 20 deviation of returns for the subject company and the market, respectively, multiplied by 21 the correlation of returns between the two. I therefore calculated the "raw" Beta 22 coefficient for each member of the proxy group, based on Equation [4], and adjusted 19 I recognize that the VIX forward settlement prices are liquid for approximately six to eight months; nonetheless, that data represents a market-based measure of expected volatility that should be considered in estimating the ex-ante Market Risk Premium. PUC DOCKET NO ROBERT B. HEVERT 947

12 1 those raw Beta coefficients to address the tendency to regress toward the market Beta 2 coefficient of unity. For the purpose of that calculation, I used weekly returns, and 3 calculated the standard deviation and correlations over the twelve month period ended 4 May 31, Averaging those results produces an adjusted Beta coefficient of Q. HOW AND WHY DID YOU ADJUST THE RAW BETA COEFFICIENT? 6 A. I adjusted my raw Beta coefficient consistent with the methodology used by Bloomberg, 7 which multiplies the raw Beta coefficient by 0.67, and adds 0.33 to that product. The 8 purpose of that adjustment is to reflect the results of substantial academic research 9 indicating that, over time, raw Beta coefficients tend to regress to the market mean of ^ 11 Q. PLEASE EXPLAIN WHY YOU RELIED ON A TWELVE-MONTH ESTIMATE OF 12 THE PROXY GROUP MEAN ADJUSTED BETA COEFFICIENT. 13 A. As noted in Section VIII, while volatility in the broad market and the utility sector 14 recently has moderated, the correlation in returns has been trending upward in recent 15 years. And, as discussed above, the Market Risk Premium tends to change over time. In 16 my view, the use of Beta coefficients calculated over shorter periods is consistent with 17 the notion that market conditions, and the risk premium required by investors in response 18 to those conditions, also may change over shorter periods.21 In any case, by relying on 19 both Value Line and Bloomberg, my CAPM analysis reflects Beta coefficients calculated 20 over longer periods, as well. 20 The regression tendency of Beta coefficients to converge to 1.0 over time is well known and widely discussed in financial literature. (See, e.g., Blume, Marshall E., On the Assessment of Risk, The Journal of Finance, Vol. 26, No. 1, March 1971, at 1-10). Please note that Value Line uses a similar adjustment methodology. 21 See Felicia Marston, Robert Harris, Peter Crawford, Risk and Return in Equity Markets: Evidence Using Financial Analysts' Forecasts, in J. Guerard and M. Gultekin (eds) Handbook of Security Analysts Forecasting and Asset Allocation, JAI Press, at 199. PUC DOCKET NO ' ROBERT B. HEVERT 948

13 I Q. IS YOUR CALCULATED BETA COEFFICIENT REASONABLE RELATIVE TO 2 THOSE CALCULATED BY VALUE LINE AND BLOOMBERG? 3 A. Yes, it is. As shown in Exhibit RBH-3, the proxy group average Value Line, Bloomberg, 4 and Calculated Beta Coefficients are 0.72, 0.72, and 0.76, respectively. In light of the 5 market dynamics noted earlier, the calculated Beta coefficient reasonably reflects current 6 conditions, although it is not materially different than those provided by Value Line and 7 Bloomberg. 8 Q. WHAT ARE THE RESULTS OF YOUR CAPM ANALYSES? 9 A. The results of my CAPM analysis are summarized in Table 5, below (see also Exhibit 10 RBH-4). 11 Table 5: Summary of CAPM Results Sharpe Ratio Bloomberg Derived Capital IQ Derived Derived Market Market Risk Market Risk Risk Premium Premium Premium Calculated Beta Coefficient Current 30-Year Treasury (2.98%) 9.35% 10.98% 10.92% Near Term Projected 30-Year Treasury (3.58 %) o/ / /o Bloomberg Beta Coefficient Current 30-Year Treasury (2.98%) 8.99% 10.53% 10.47% Near Term Projected 30-Year Treasury (3.58 %) 9.59% % % Value Line Beta Coefficient 12 Current 30-Year Treasury (2.98%) 9.01% 10.56% 10.50% Near Term Projected 30-Year Treasury (3.58 %) 9.62% 11.16% % PUC DOCKET NO ROBERT B. HEVERT 949

14 I Q.- DOES YOUR RECOMMENDATION SUBSTANTIALLY RELY ON THE CAPM 2 RESULTS? 3 A. No, it does not. While I have calculated the CAPM using the approaches and 4 assumptions discussed above, I did not give any specific weight to those results. Rather, 5 1 used the CAPM results to assess the DCF results discussed earlier. 6 7 Bond Yield Plus Risk Prenziuzn Approach 8 Q. PLEASE GENERALLY DESCRIBE THE BOND YIELD PLUS RISK PREMIUM 9 APPROACH. 10 A. In general terms, this approach is based on the fundamental principle that equity investors 11 bear the residual risk associated with ownership and therefore require a premium over the 12 return they would have earned as a bondholder. That is, since returns to equity holders 13 are more risky than returns to bondholders, equity investors must be compensated for 14 bearing that risk. Risk premium approaches, therefore, estimate the cost of equity as the 15 sum of the Equity Risk Premium and the yield on a particular class of bonds. As noted in., 16 my discussion of the CAPM, since the Equity Risk Premium is not directly observable, it 17 typically is estimated using a variety of approaches, some of which incorporate ex-ante, 18 or forward-looking estimates of the cost of equity, and others that consider historical, or 19 ex post, estimates. An alternative approach is to use actual authorized returns for electric 20 utilities to estimate the Equity Risk Premium. PUC DOCKET NO ROBERT B. HEVERT 950

15 1 Q. PLEASE NOW EXPLAIN HOW YOU PERFORMED YOUR BOND YIELD PLUS, 2 RISK PREMIUM ANALYSIS. 3 A. As suggested above, I first defined the Risk Premium as the difference between the 4 authorized ROE and the then-prevailing level of the long-term (i.e., 30-year) Treasury 5 yield. I then gathered data for approximately 1,346 electric utility rate proceedings 6 between January 1980 and May In addition to the authorized ROE, I also 7 calculated the average period between the filing of the case and the date of the final order 8 (the "lag period). In order to reflect the prevailing level of interest rates during the 9 pendency of the proceedings, I calculated the average 30-year Treasury yield over the 10 average lag period (approximately 202 days). 11 Because the data cover a number of economic cycles,22 the analysis also may be 12 used to assess the stability of the Equity Risk Premium. Prior research, for example, has 13 shown that the Equity Risk Premium is inversely related to the level of interest rates23 14 That analysis is particularly relevant given the historically low level of current Treasury 15 yields. 16 Q. - HOW DID YOU MODEL THE RELATIONSHIP BETWEEN INTEREST RATES 17 AND THE EQUITY RISK PREMIUM? 18 A. The basic method used was regression analysis, in which the observed Equity Risk 19 Premium is the dependent variable, and the average 30-year Treasury yield is the 20 independent variable. Because the analytical period includes interest rates and authorized 22 National Bureau of Economic Research, U.S. Business Cycle Expansion and Contractions. 23 See, for example, Robert S. Harris and Felicia C. Marston, Estimating Shareholder Risk Premia Using Analysts' Growth Forecasts, Financial Management, Summer 1992, at 63-70; Eugene F. Brigham, Dilip K. Shome, and Steve R. Vinson, The Risk Premium Approach to Measiuring a Utility's Cost of Equity, Financial Mana g ement, Spring 1985, at 33-45; and Farris M. Maddox, Donna T. Pippert, and Rodney N. Sullivan, An Empirical Study of Ex Ante Risk Premiums for the Electric Utility Industry, Financial Management, Autumn 1995, at PUC DOCKET NO ROBERT B. HEVERT 951

16 1 ROEs that during one period (i.e., the 1980's) are quite high and another (the 'post- 2 Lehman bankruptcy period) that are quite low relative to the long-term historical average, 3 1 used the semi-log regression, in which the Equity Risk Premium is expresses as a 4 function of the natural log of the 30-year Treasury yield:.rp = a t fl(ln((tsa]) [5] 5 6 As shown on Chart 1 (below), the semi-log form is useful when measuring an 7 absolute change in the dependent variable (in this case, the Risk Premium) relative to a 8 proportional change in the independent variable (the 30-year Treasury yield). 9 Chart 1: Equity Risk Premium As Chart 1 demonstrates, over time there has been a statistically significant, negative relationship between the 30-year Treasury yield and the Equity Risk Premium. Consequently, simply applying the long-term average Equity Risk Premium of 4.31 percent (see Exhibit RBH-5) would significantly under-state the Cost of Equity; assuming the current projected 30-year Treasury yield of 3.58 percent, for example, the simple average Equity Risk Premium would suggest an ROE of 7.89 percent. That, of PUC DOCKET NO ROBERT B. HEVERT ^^*A

17 I course, is well below any reasonable estimate. Based on the regression coefficients in 2 Chart 1, however, the implied ROE is percent (see Exhibit RBH-5). In any event, 3 the analysis demonstrates that there has been a significant inverse relationship between 4 the 30-year Treasury yield and the Equity Risk Premium. 5 6 VII. BUSINESS RISKS 7 Q. DO THE MEAN DCF AND CAPM RESULTS FOR THE PROXY GROUP PROVIDE 8 AN APPROPRIATE ESTIMATE OF THE COST OF EQUITY FOR THE COMPANY? 9 A. No, the mean results do not necessarily provide an appropriate estimate of the Company's 10 Cost of Equity. In my view, there are additional factors that must be taken into 11 consideration when determining where the Company's Cost of Equity falls within the 12 range of results, in particular the incremental risks associated with the Company's need to 13 fund substantial capital expenditures associated with environmental regulation 14 compliance, and the effect of prudently incurred flotation costs Effect of Environmental Regulation on Coal-Fired Generation 17 Q. PLEASE BRIEFLY DESCRIBE THE RISKS ASSOCIATED WITH THE 18 OWNERSHIP OF COAL-FIRED GENERATING RESOURCES. 19 A. In general, capital-intensive base load generation assets, such as coal-fired plants, face 20 risks associated with capital recovery in the event of market structure changes or plant 21 failure, or replacement cost recovery in the event of extended or unplanned outages. To 22 that point, in a recent report regarding the effect of environmental regulations on electric 23 utilities, Fitch Ratings stated that electric utilities will face increasing operating and PUC DOCKET NO ROBERT B. HEVERT 953

18 1 capital costs, both of which would be a "credit negative". And while Fitch believes the 2 risks may be "manageable", the increasing number of coal unit retirements will inject "a 3 measure of risk for investors."24 As to the effect of environmental regulations on coal unit 4 retirements, Fitch recently estimated that some 80 gigawatts of coal-fired generating 5 capacity are at risk of early retirement. For this reason, it is important for SWEPCO to 6 maintain access to capital markets as needed to respond to environmental regulations. 7 Q. DOES THE COMPANY'S GENERATION RESOURCE PORTFOLIO INCLUDE 8 COAL-FIRED ASSETS? 9 A. Yes, SWEPCO's coal-fired generating assets represent approximately percent of its 10 generating capacity25 and more than percent of its 2011 net generation Q. IS THERE EVIDENCE THAT DEMONSTRATES THE EFFECT OF STRICTER 12 ENVIRONMENTAL REGULATIONS ON SWEPCO? 13 A. Yes, as a result of the increased likelihood of carbon emissions regulations, as well as 14 other emissions regulations, investors see coal generation as taking on even greater risk. 15 AEP recently notified regional reliability organizations that it plans to retire 4,600 MW of 16 coal-fired capacity, primarily to comply with the EPA's regulations. This includes 17 SWEPCO's 528 MW Welsh Plant Unit 2. AEP also stated that another 13,000 MW of 18 capacity will require capital investment in emission control systems?^ 19 Given the increasing regulatory and legislative focus on environmental 20 compliance for companies that are dependent on coal-fired generation, it is important to. 24 SNL Financial, Fitch: Coal retirements a credit risk, but one that can be managed, March 2, American Electric Power Co., Inc., 2011 Fact Book, 46'h EEI Financial Conference, at SNL Financial. 27 American Electric Power Co., Inc., AEP Notifies Reliability Organizations of Planned Plant Retirements, Press Release, March 22, PUC DOCKET NO ROBERT B. HEVERT 954

19 1 acknowledge that the projected costs related to the'compliance programs represent an 2 additional risk for SWEPCO. 3 Q. PLEASE DESCRIBE SOME OF THE ENVIRONMENTAL REGULATIONS THAT 4 COULD CAUSE AN INCREMENTAL EFFECT ON THE COMPANY'S COAL- 5 FIRED ASSETS. 6 A. Federal environmental regulations creating emissions control requirements have been 7 issued in recent years. Compliance with new regulations can require substantial capital 8 investment, or add operational costs. For instance, the Cross-State Air Pollution Rule 9 (CSAPR) would replace the Clean Air Interstate Rule (CAIR) and impose stringent 10 emissions limits on SO2 and NOx. While enforcement of CSAPR will not begin while 11 the rule remains under judicial review, 28 the stringency of the required emissions 12 reductions creates substantial challenges for AEP that would require additional spending 13 to meet emissions limits In addition; the Federal Environmental Protection Agency (EPA) proposed a rule 15 in June 2010 to regulate fly ash and bottom ash generated at coal-fired electric generating 16 units. The rule is not yet final,. but the proposed rule has the potential to restrict certain 17 uses of coal ash currently employed as well as require upgrades of current waste 18 facilities. Since the rule is not yet final, the potential costs are not known, but the 19 Company expects the costs to be "significant."30 28 Recently, the US Court of Appeals for the District of Columbia Circuit reaffirmed the Federal Environmental Protection Agency's ("EPA") authority to regulate greenhouse gases. See for example SNL Financial. Court deals major blow to industry's challenge of greenhouse gas rules. June 26, American Electric Power Co., Inc., 2011 Annual Report: Appendix A to the Proxy Statement, at S. 30 Ibid., at 9 and 10. PUC DOCKET NO ROBERT B. HEVERT 955

20 I Q. PLEASE BRIEFLY SUMMARIZE THE COMPANY'S CURRENT CAPITAL 2 INVESTMENT PLAN FOR ENVIRONMENTAL EXPENDITURES. 3 A. SWEPCO's environmental construction plan for 2012 to 2014 includes $594.3 million in 4 capital spending and is based on current expectations of environmental regulation.31 To 5 the extent regulatory constraints, or environmental regulations change, further spending 6 could be required. 7 Q. DO CREDIT RATING AGENCIES RECOGNIZE THE RISKS TO CREDIT QUALITY 8 ASSOCIATED WITH INCREASED CAPITAL EXPENDITURES DUE TO 9 ENVIRONMENTAL REGULATIONS?, 10 A. Yes, they do. In a recent report, Standard and Poor's described the increased level of 11 capital expenditures in the utility industry and the particular challenges related to coal 12 plants. S&P noted: ' 13 Coal currently fuels about 45% of all electricity generated in the U.S., and 14 is the most-used fuel for regulated utilities. Utilities have filed.plans with 15 regulators to retire, retrofit, and rebuild their generation portfolios. Some 16 utilities are either retiring or idling regulated units, while others have 17 proposed construction of new natural gas units to replace the shuttered 18 coal capacity. Of the roughly 310 gigawatts (GW) of U.S. coal-fired 19 capacity, less than one-quarter has all the pollution equipment necessary to 20 comply with EPA rules. Given that regulated utilities own about three- 21 quarters of U.S. coal generation, the cost of installing pollution controls 22 and the speed with which utilities can fully recover those costs through 23 rate adjustments could affect credit quality American Electric Power Company, Inc., SEC Form 10-K for the fiscal year ended December 31, 2011, at 10 and Standard & Poor's, U.S. Utilities' Capital Spending Is Rising, And Cost-Recovery Is Vital, RatingsDirect, May 14, 2012, at 5. PUC DOCKET NO ROBERT B. HEVERT 956

21 i 1 Q. HOW DO CAPITAL EXPENDITURES AFFECT A COMPANY'S CREDIT 2 QUALITY? 3 A. From a credit perspective, the additional pressure on cash flows associated with high 4 levels of capital expenditures exerts corresponding pressure on credit metrics and, 5 therefore, credit ratings. Standard and Poor's has noted several long-term challenges for 6 utilities' financial health including: heavy construction programs to address demand 7 growth; declining capacity margins; and aging infrastructure and regulatory 8 responsiveness to mounting requests for rate increases. S&P further noted that: 9 To sustain their current credit quality in the face of these long-lived 10 challenges, utilities need to have established-and be able to maintain-a I1 firm credit foundation. This will require a strong and effective working 12 relationship among management, regulators, and increasingly legislators 13 and governors, in the planning and execution of strategies. A 14 comprehensive vetting and understanding of the risks associated with the 15 regulatory mechanisms under which the utility will recover its investment, 16 which could include a cash return during construction and timely 17 recognition of volatile costs, will be paramount in preserving 18 creditworthiness As to the Company in particular, speaking to the importance of maintaining 20 strong credit metrics to support the Company's capital investment plan, S&P noted: 21 SWEPCO's financial risk profile reflects AEP's consolidated, financial 22 measures that are in line with the rating. This assessment reflects large 23 capital expenditures mostly for environmental-compliance programs and 24 for new generation and transmission. The elevated spending and dividend 25 payments could result in negative discretionary cash flow for several 26 years, and will require vigilant cost recovery to maintain cash flow 27 measures The rating agency views noted above also are consistent with certain observations 29 made earlier in my Direct Testimony: (1) the benefits of maintaining a strong financial Standard & Poor's, Industry Report Card: Utility Sectors In the Americas Remain Stable, While Challenges Beset European, Australian, and New Zealand Counterparts, RatingsDirect, June 27, 2008, at 4. Standard & Poor's, Summary: Southwestern Electric Power Co., Standard & Poor's Research, June 19, 2012, at 3. PUC DOCKET NO ROBERT B. HEVERT 957

22 r I profile are significant when capital access is required, and become particularly acute 2 during periods of market instability; and (2) the Commission's decision in this 3 proceeding will have a direct bearing the Company's credit profile, and its ability to 4 access the capital needed to fund its investments. Maintaining the Company's credit 5 profile is of particular importance, given SWEPCO's current BBB and Baa3 credit 6 ratings by Standard and Poor's and Moody's, respectively. ` 7 Q. ARE EQUITY INVESTORS ALSO CONCERNED WITH COMPARATIVELY HIGH 8 LEVELS OF CAPITAL EXPENDITURES? 9 A. Yes, equity investors also recognize the pressure on cash flows associated with relatively 10 high levels of capital expenditures. For example, KeyBanc Capital Markets (KeyBanc) 11 conducts a quarterly review of the electric utility industry. In a recent report, KeyBanc 12 noted that: 13 Credit and liquidity concerns have driven many companies to revisit 14 capital spending plans and reassess operational efficiencies. The primary 15 response has generally been to delay projects, as opposed to outright 16 cancellation. Initially, reductions in capital programs were a function of 17 lower growth, which eliminated the need for growth-related capital 18 spending on items such as line extensions and new substations. However, 19 as difficult economic conditions persist, the cuts have grown more 20 extensive, with deferrals in non-core maintenance spending, reevaluating 21 the cost-effectiveness of running' older inefficient power plants and 22 pursuing company restructurings or mergers More recently, KeyBanc noted that: 24 While recent prices may have come off of their earlier highs due to the 25 global economic crisis slowing construction demand, we believe the long- 26 term trend of rising construction materials costs could resume as the 27 global economy rebounds. The cost of building new generation remains a 28 moving target, as worldwide demand for' construction materials 29 commodities (steel, concrete and copper), labor and components (turbines 30 " and boilers) would remain fundamentally strong, driven by a rebound in 31 the U.S. and Chinese economies and required compliance with future U.S. 35 KeyBanc Capital Markets Inc., Electric Utilities Quarterly 4Q10, March 2011, at 7. PUC DOCKET NO ROBERT B. HEVERT 958

23 I environmental regulations. We believe this presents challenges to both 2 unregulated and regulated investment in new generation plants. In 3 particular, on the regulated side, there exists a chicken-and-egg problem. in 4 that securing pricing without a regulatory buy-in is as difficult as receiving 5 regulatory pre-approval without firm pricing.36 6 Q. WHAT ARE YOUR CONCLUSIONS REGARDING THE EFFECT OF THE 7 COMPANY'S CAPITAL SPENDING PLANS AND ENVIRONMENTAL 8 REGULATIONS ON ITS RISK PROFILE? 9 A. It is clear that the Company's capital expenditure program is substantial, and further 10 spending may be required to comply with environmental regulations, especially with its 11 significant proportion of coal-fired generating assets. It also is clear that the financial 12 community recognizes the additional risks associated with capital expenditures and that 13 those risks are reflected in market valuation multiples. In my view, these factors suggest 14 a comparatively high level of risk Flotation Cost Adjustment 17 Q. WHAT ARE FLOTATION COSTS? 18 A. Flotation costs are the costs associated with the sale of new issues of common stock. 19 These costs include out-of-pocket expenditures for preparation, filing, underwriting, and 20 other issuance costs of common stock. 21 Q. WHY IS IT IMPORTANT TO RECOGNIZE FLOTATION COSTS IN THE 22 ALLOWED ROE? 23 A. In order to attract and retain new investors, a regulated utility must have the opportunity 24 to earn a return that is both competitive and compensatory. To the extent that a company 25 is denied the opportunity to recover prudently incurred flotation costs, actual returns will 36 KeyBanc Capital Markets Inc., Electric Utilities Quarterly 3Q11December 2011, at 17. PUC DOCKET NO ROBERT B. HEVERT 959

24 I fall short of expected (or required) returns, thereby diminishing its ability to attract 2 adequate capital on reasonable terms. 3 Q. ARE FLOTATION COSTS PART OF THE UTILITY'S INVESTED COSTS OR PART 4 OF THE UTILITY'S EXPENSES? 5 A. Flotation costs are part of the invested costs of the utility, which are properly reflected on 6 the balance sheet under "paid in capital." They-are not current expenses, and therefore 7 are not reflected on the income statement. Rather, like investments in rate base or the 8 issuance costs of long-term debt, flotation costs are incurred over time. As a result, the 9 great majority of a utility's flotation cost is incurred prior to the test year, but remains 10 part of the cost structure that exists during the test year and beyond, and as such, should 11 be recognized for ratemaking purposes. Therefore, recovery of flotation costs is 12 appropriate even if no new issuances are planned in the near future because failure to 13 allow such cost recovery may deny SWEPCO the opportunity to earn its required rate of 14 return in the future. 15 Q. IS THE NEED TO CONSIDER FLOTATION COSTS ELIMINATED BECAUSE 16 SWEPCO IS A WHOLLY-OWNED SUBSIDIARY OF AMERICAN ELECTRIC 17 POWER COMPANY? 18 A. No. Although the Company is a wholly-owned subsidiary of American Electric Power 19 Company, it is appropriate to consider flotation costs because wholly-owned subsidiaries 20 receive equity capital from their parents and provide returns on the capital that roll up to 21 the parent, which is designated to attract and raise capital based on the returns of those 22 subsidiaries. To deny recovery of issuance costs associated with the capital that is 23 invested in the subsidiaries ultimately will penalize the investors that fund the utility PUC DOCKET NO ROBERT B. HEVERT 960

25 1 operations and will inhibit the utility's ability to obtain new equity capital at a reasonable 2 cost. This is important for companies such as SWEPCO that are planning continued 3 capital expenditures in the near term, and for which access to capital (at reasonable cost 4 rates) to fund such required expenditures will be critical. 5 Q. DO THE DCF AND CAPM MODELS ALREADY INCORPORATE INVESTOR 6 EXPECTATIONS OF A RETURN IN ORDER TO COMPENSATE FOR FLOTATION 7 COSTS? 8 A. No. All the models used to estimate the appropriate ROE assume no "friction" or 9 transaction costs, as these costs are not reflected in the market price (in the case of the 10 DCF model) or risk premium (in the case of the CAPM). Therefore, it is appropriate to 11 consider flotation costs when determining where within the range of reasonable results 12 SWEPCO's return should fall. 13 Q. IS THE NEED TO CONSIDER FLOTATION COSTS RECOGNIZED BY THE 14 ACADEMIC AND FINANCIAL COMMUNITIES? 15 A. Yes. The need to reimburse for equity issuance costs is justified by the academic and 16 financial communities in the same spirit that investors are reimbursed for the costs of 17 issuing debt. This treatment is consistent with the philosophy of a fair rate of return. 18 According to Dr. Shannon Pratt: 19 Flotation costs occur when a company issues new stock. The business 20 usually incurs several kinds of flotation or transaction costs, which reduce 21 the actual proceeds received by the business. Some of these are direct out- 22 of-pocket outlays, such as fees paid to underwriters, legal expenses, and 23 prospectus preparation costs. Because of this reduction in proceeds, the 24 business's required returns must be greater to compensate for the 25 additional costs. Flotation costs can be accounted for either by amortizing 26 the cost, thus reducing the net cash flow to discount, or by incorporating 27 the cost into the cost of equity capital. Since flotation costs typically are PUC DOCKET NO ROBERT B. HEVERT 961

26 1 not applied to operating cash flow, they must be incorporated into the cost 2 of equity capital.37 3 Q. HAVE YOU CALCULATED THE EFFECT OF FLOTATION COSTS ON THE.4 RETURN ON EQUITY? 5 A. Yes; I have. I modified the DCF calculation to derive the dividend yield that would 6 reimburse investors for direct issuance costs. Based on the weighted average issuance 7 costs shown in Exhibit RBH-6, I believe that a reasonable estimate of flotation costs is 8 approximately 0.15 percent (15 basis points) VIII. CAPITAL MARKET ENVIRONMENT 11 - Q. DO ECONOMIC CONDITIONS INFLUENCE THE REQUIRED COST OF CAPITAL 12 AND REQUIRED RETURN ON COMMON EQUITY? 13 A. Yes. As discussed in Section VI, the models used to estimate the Cost of Equity are 14 meant to reflect, and therefore are influence by, current and expected capital market 15 conditions. 16 Q. HAVE YOU REVIEWED ANY SPECIFIC INDICES TO -ASSESS THE 17 RELATIONSHIP BETWEEN CURRENT MARKET CONDITIONS AND INVESTOR 18 RETURN REQUIREMENTS? 19 A. Yes, I considered several measures of capital market risk, including: (1) incremental 20 credit spreads on investment grade utility debt; (2) the relationship between electric 21 utility dividend yields and long-term Treasury yields; and (3) equity market volatility and 22 correlations. As discussed below, each of those measures provide information that is 37 Shannon P. Pratt, Roger J. Grabowski,'Cost of Capital: Applications and Examples. 4th ed. (John Wiley & Sons, Inc., 2010), at 586. PUC.DOCKET NO e 42 ROBERT B. HEVERT 962

27 V relevant to the implementation of models used to estimate the Cost of Equity, and in the 2 interpretation of the model results. 3 4 Incremental Credit Spreads 5 Q. HOW HAVE CREDIT SPREADS BEEN AFFECTED BY CURRENT MARKET 6 CONDITIONS? 7 A. The "credit spread" is the return required by debt investors to take on the default risk 8 associated with securities of differing credit quality. For a given credit rating, the credit 9 spread is measured by reference to a Treasury security of similar tenure. That is, the 10 credit spread on A-rated utility bonds may be measured by reference to the 30-year 11 Treasury Bond yield; the same would be true of Baa-rated securities. Because lower 12 credit ratings reflect higher levels of risk, credit spreads typically are higher for lower- 13 rated securities. In that regard, the incremental credit spread (e.g., the difference between 14 the credit spreads associated with A and Baa-rated securities, respectively) is an 15 indication of incremental return required by investors to take on additional levels of risk. 16 As Chart 2 demonstrates, since the beginning of 2010, the Moody's Utility Bond Index 17 Baa/A credit spread has steadily increased, indicating that debt investors have increased is their marginal return requirements., PUC DOCKET NO ROBERT B. HEVERT 963

28 I Chart 2: Moody's Utility Bond Index Baa-A Credit Spread It also is interesting to note that the incremental credit spread has increased as long-term Treasury yields have decreased. In fact, as Chart 3 demonstrates, even since 5 January 2010, changes in the incremental spread are negatively correlated with changes. 6 in the 30-year Treasury yield. PUC DOCKET NO ROBERT B. HEVERT 964

29 I Chart 3: Incremental Credit Spread and 30-Year Treasury Yield 2 3 Q. 4 5 A. 6 WHAT ARE THE IMPLICATIONS OF THOSE FINDINGS IN ASSESSING THE COMPANY'S COST OF EQUITY? The implications are twofold. First, the recent decline in long-term Treasury yields has been accompanied by an increase in the premium required by investors to accept 7 incremental levels of credit risk. That is, the incremental credit spread has increased as 8 9 the level of Treasury yields have decreased. While that inverse relationship applies to the cost of debt, prior academic research has demonstrated that the equity risk premium 10 likewise is inversely related to interest rates.38 Consequently, neither the Cost of Equity 11 nor the-cost of debt has decreased in lock step with Treasury yields. 38 Robert S. Harris and Felicia C. Marston, Estimating Shareholder Risk Premia Using Analysis' Growth Forecasts, Financial Management, Summer 1992, at 63-70; Eugene F. Brigham, Dilip K. Shome, and Steve R. Vinson, The Risk Premium Approach to Measuring a Utility 's Cost of Equity, Financial Mana gement, Spring 1985, at 33-45; and Farris M. Maddox, Donna T. Pippert, and Rodney N. Sullivan, An Empirical Study of Ex Ante Risk Premiums for the Electric Utility Industry, Financial Management, Autumn 1995, at PUC DOCKET NO ROBERT B. HEVERT 965

30 I Those results also demonstrate the importance of maintaining a financial and 2 credit profile that supports the Company's current BBB rating. Because incremental 3 credit spreads have steadily increased, the benefit of maintaining a BBB-rating is greater 4 in the current market than it has been, even over the past two years. That conclusion is 5 consistent with recent findings by FitchRatings (Fitch), which noted that: 6 While it appears that the credit spread differential between the rating 7 categories has a relatively small impact during times of economic stability, 8 during recent periods of economic stress, a higher credit rating produces a 9 meaningful difference in credit spreads... and provides more assured 10 access to capital Since regulatory actions affect credit ratings in several, often significant ways, the 12 Commission's decision in this proceeding will directly affect the Company's credit 13 profile and influence its ability to maintain a credit profile that enables continued access 14 to capital at reasonable costs. Given the Company's capital investment plans and 15 external funding needs, the benefits of reliable and cost-effective capital access are 16 significant Yield Spreads 19 Q. PLEASE BRIEFLY DEFINE THE TERM "YIELD SPREAD", AND EXPLAIN ITS 20 MEANING IN ASSESSING CAPITAL MARKET CONDITIONS A. The "yield spread" is the difference between the yield on long-term Treasury securities on the one hand, and dividend yields on the other. Investors often consider yield spreads 23 in their assessment of security valuation and capital market conditions. As explained below, to the extent that yield spreads materially deviate from long-term relationships, it may be an indication of continuing dislocations within the capital market. 39 FitchRatings, Fitch's Review of Utility ROE Trends, March 22, 2010, at 3. PUC DOCKET NO ROBERT B. HEVERT 41-11