w n w v.”- *- su‘ .. -.~ . ,. ‘Wu. "”“ .. .g“ o—n-vJ‘ coo "w. - . II I II” '....... N. w.-... ., ' - v Mo 55 555. 5 5555; 5555 5-. 5.5.55 5. ' [‘I'HIIE'II ”55",“ l' ”55‘ '5 8g;5§7I..551'$I'." n 'I’II’I’ ”I" 'I-I’I‘I'III -5‘ .5..I"'I’5'5"".5' " I’I" I’I’J (5 3555'! )4 I my} II” ’EJ’ii'I'j’l‘» '4. 1'; ‘-.. I 5'5 5W"? 5: 5" 1532:; 35555553..» 5 ”1’5!" 5JIHI'II" I“??? : III—LIL} ' :_.'“1‘: . f L" 1:... ‘35 5 M55555!“ ' 5 ‘5 ' ‘55 I'“F5555 5 I .5 5" 5 ' ,- .. 55555555 5,55”. N- .. . u. .H......._ _. ’ “~. 4.. ,. ‘5 %. .{k— L. ‘ *0. . - . v wk:— -. r... n ”M- I ‘vzr-v- -.-. um, ‘1’». I..- ‘ 4......” , H‘.‘ .5.»- “‘ ,‘M -- 37.5... M M“" .o... - _ . .. . .- ' ' , ‘ w- “ ' - c ~37... , ' . ,. : , - . ' ' I 5. . 5 ‘ . .. ' 5 - . .. . . .r.':5- ~55 - .. -mk" “*«lr-- _ . L... L‘. I ‘H’ I‘IIJ...‘ I'I 5‘5 i'5. .5”. .:-:-5‘ « 55 55 15' 5-5555 .. . 5.55:: . 5:5 5 5'}. ’5 " 1"31’ "}o"""o).‘ :‘c. .... 555‘ '5’5’555'5’5-5‘5‘553 I’. :5: 55.2I 5 "- - ~ 3‘ ‘ ‘5 51.85555 .5 55 .I5 5. 55 5 -. I In -'-, 5. "'5§II51I'I.i"II’I’I "5’1" I 5‘5"“r 'I.. ‘.':Z".. . "5' c. 5-:5‘) 55 ._ ‘5 ‘_ I" . " :\.'.,5‘ 51'55 555‘5 5 .1 5'2' ‘1555355': iII’i515‘II:?"' ’75 "(MU-I5 5' 555555 5 5. '- 5. * 5 ' .I 5 ‘Izi”3"'Ifi‘I> 'I5”’:15"::"I" ' ". ' "5 I" ' "' I' ' ’ ‘ 55 11‘5” .' ‘ ~ I‘ II' 1"?55- 55‘"'1-' " 5' 5'.I " . “ 5 ‘, 55’ :‘ II 'wI- 55*..' I .’::. ’5 -l5 5 55 3555' "It!" . . ‘ -1 "’5 I 5-:5I- ‘5’5'. .5 5. ....’ II" 5 5. .‘IL' ,-: '"II ' .I 55. II 555 55.5.55- :55... ~ . I: 5.~ 55'55I'.‘5‘I;555'551.‘.I5-. “I." . 5'" 5', . '5 '1:-=';‘.': '51:";’5'“'jI’I."I""""'II"'.‘ .5 ‘55- "335 ‘ , 5. .5'5'" “:‘9 ’5'5' _I "I" 1:” '5 .5 ‘5‘; ‘IIII‘. "’I' "II" '5'I :1 II :. .55:.~I .- ." I‘II’5‘II'I ’55 .5'5"5I:i5- 5".”'..:“' :‘I555. I .2' "HF II:. 5I' '5 I‘ III ’anIII 55” 5.5’!" .‘-I '55 - ' ’II’IIJ' HIV-”II "”""‘5'”" 'I .II " H 5' " I. ‘5 " ”"L’ ’5' 5 5.". '1" “’5',{.‘"51'I':""§"’I"" “ '5" 5' .- "' NIH" v‘-':'.'fi"‘I"II " ' "" "'“:'5;: ”I 5'1"". I, " '5£’5' 'fif’I" ’l.:.” II ' "i': I‘;|":l II! ""I [’I’Iil" H" ‘1’ '5'“ I 'I " ." I 3' ’5’5555.‘ 5’III5IIII'I’IIIII-I-II“5"5' .55.- '5 '5 5 5 5 5 5.55 55’."‘I'. 5. I 5 5‘ ”I" I5 .I III.- .5; 55.55.5555 K 5 - I ‘. I. I 5 5 5- 55 _ I |. I: ! 5|_‘.|‘... -I'I'I‘55"'I’ 'I 55 "I I ‘ "' 5;;5 'I;I. I'I’TI-Izn '.;'.,' z I ".15" II“ 'I’:’II"' ”I5"- ‘ 5‘55”. '5 5-“5 '0":$';. "'1?qu I" u, "’ Rim I ¢ , .,I_-5‘5I5‘ ’I’II'II ’II'I’II-I5I55 III I' 5 55' I' 5-5555 55-55 .‘ 5"" 5: II" ’55,. 555’ IL’I’” .5-5' 'I" III’ ’” 'III’II "II-"I 555’555 .I " '5 ’ ' "I". 5"”5'! ' {it'll I 5. I I 'l I ' ' .‘. 5 ’-I "V f I, 5 " l‘ .‘1 ,I, I551I"}.’535~5I55?" 5 “55""_',' . '5’” " l5 .5! 5'IIII 5 5555.555“ ’5IIII5555.155' ’II 5‘"';I'" I (J. (j .I. MM“ HH"I"'5II """'~‘ """' ' 5'IIIII’.'.I'5’I'I ' " "I’m MI 'I I"""’ I W5“ I’II'II""""'I’I’ IIIII ”Iii" ! ' '5” "’1'" ' III “I III III’I5IIII5 II 5555 5555” I "III 5.’I.’III"' II II' ”"II’ 5 I ' II ’5 . 5”" III’I”'I '5’I5'II"_’I "'IHIIII 5.," ———--,_.._ .3. m 2;— -——- THhL—BIS “$.33: 23;. \\\\t\\x\\;ii\\sz\\wttx\\\\ Uni Vfl‘s! ty EMPIRICAL INVESTIGATION OF ALLOWANCE FOR FUNDS USED DURING CONSTRUCTION IN THE ELECTRIC UTILITY INDUSTRY presented by A. James Ifflander has been accepted towards fulfillment of the requirements for Ph. D. degree in Business Administration . \ A ,7 % .:5{7”M ajor pro essor //———— DateXf/f/ "‘321 MS U is an Affirmative Action/Equal Opportunity Institution 0- 12771 MSU LIBRARIES M \' RETURNING MATERIALS: PIace in book drop to remove this checkout from your record. FINES wiII be charged if book is returned after the date stamped be10w. ii a 1 A iii.; 0-. '23» . I? n k "“W/ée 3‘ S I..." EMPIRICAL INVESTIGATION OF ALLOWANCE FOR FUNDS USED DURING CONSTRUCTION IN THE ELECTRIC UTILITY INDUSTRY BY Albert James Ifflander A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Finance 1982 \ ERIC—l anvvf fin 4th.“: w. “‘Avu Lory" h" “‘ “‘1- ">~5 \ "'hhuvuu' . . :‘vfi‘.‘; fin ”'"bhdu up!“ ~L N"“¢a 4 ES ‘1‘ b\ 0.. “‘c. I '-.. ~‘b "‘4‘. I In “I 't 1 ". It! (I. no It .5.” \:n ABSTRACT EMPIRICAL INVESTIGATION OF ALLOWANCE FOR FUNDS USED DURING CONSTRUCTION IN THE ELECTRIC UTILITY INDUSTRY BY Albert James Ifflander Allowance for Funds used during construction (AFUDC) represents capitalized interest on funds devoted to con- struction projects. This capitalized interest flows through to the income statement and is considered other utility income in the year of capitalization. The problem relates to whether the allowance for funds is the same quality as the operating earnings of the utility. A major controversy has arisen among followers of the electric utility industry about AFUDC, whether it is really repre- senting lower quality earnings and whether the declines in market valuation over the last fifteen years are the result of increasing percentages of AFUDC. The research performed here focused on the information content contained in the disclosure of AFUDC and the immediate market reaction to information concerning AFUDC contained in the financial statements. The test design follows the methodology first developed by Gonedes. The basic premise is to form portfolios so that the relative risks as defined by Beta are equivalent. If the relevant risks are equivalent then the expected returns should also be equivalent. Both univariate and multivariate T-tests are performed to examine the equivalence of monthly stock .50 E‘f . v ' {-H‘ any f g - ADI. \ u-VOU ' 5 - Hilre 0‘ J otib“ n‘. ..‘.E :3 . .. OI (I. v — noo‘d‘: b e..- 3"} '3‘”. '""‘ b» 5‘. \M‘. . I I Luv} ls ';E A 'R In \ « ‘1’“. .1 J ‘i‘.‘. J ‘(Vu:g‘: . ‘9’»; ’7 "‘C" “ PA”. ‘V‘Cv- ...I.. h“ M“ 1" 3 5 Ln (n c r U) returns. Equivalence of monthly stock returns around the disclosure date implies no information content to the dis- closure of AFUDC earnings in the financial statements. The major findings of the study may be summarized as follows remembering that method I companies exclude while method II companies include construction work in the rate base. First, there is no significant difference in monthly stock returns between portfolios of method I companies when AFUDC is used as the information variable. Second, there is no significant difference in monthly stock returns for portfolios of method I firms compared to portfolios of method II firms. The consludion is that all information is known previous to the disclosure date, allowing perfect forecasts of AFUDC in relation to operating earnings. Therefore, no significant information is contained in AFUDC disclosure that may be used by investors to assess stock returns. Ii. ! I at n In 1.1.. h 2v TABLE OF CONTENTS List of Figures . . . . . . List of Tables. . . . . . . Introduction. . . . . . Purpose of the Study. . Outline of the Study. . Methodology . . . . . . I. Description of Allowance for Funds Used During Construction. . . . Definition of Allowance During Construction. . Background on AFUDC . . Regulatory Process. . . Rate Base . . .'. . . Rate of Return. . . . Expenses Allowed. . Regulatory Lag. . . . . for Funds Used II. Regulation and Allowance for Funds. Brief History of AFUDC. . . . . . . Determination of Capitalizing Rates Accounting for the Allowance. . . . Taxes and the Disclosure of AFUDC . . Differences Between Method I and Method II. . . . . . . . . . . . . . The Concept of Interest on Interest . Mathematical Treatment of Differences Between Method I and Method II . . . Summary . . . . . . . . . . . . . . . Defenses and Criticisms of Allowance for Funds Used During Construction . . . . . Defenses. . . . . . . . . . . . . . . Separation Argument . . . . . . . . . Valid Asset Argument. . . . . . . . . Additional Defenses . . . . . . . . . Criticisms. . . . . . . . . . . . . . Manipulation. . . . . Quality of Earnings . . . . . . . . . Unstable Earnings . . Comparison of AFUDC with Other Income Depreciation. . . . . . . . . . . . . Deferred Taxes. . . . . . . . . . . . Summary . . . . . . . . . . . . . . . ii page O‘ubI-‘H ll 19 21 23 25 25 26 28 28 30 35 41 43 48 54 59 59 69 71 72 72 73 74 75 76 78 80 , u {on .w v . P UH o I Ab M; e . . .c “ Paw HI- N. 0| .9 - ‘u iii page III. Literature Review . . . . . . . . . . . . . . . 83 Price Earnings Ratio. . . . . . . . . . . . 83 Market- to—Book-Ratio. . . . . . . . . . . . 89 Interest Coverage Ratio . . . . . . . . . 98 Other Empirical Tests of AFUDC. . . . . . . 100 IV. Methodology: Theoretical Development and Empirical Support. . . . . . . . . . . . . . . 105 Information Content . . . . . . . . . . . . 106 Market Efficiency . . . . . . . . . . . . . 108 Capital Asset Pricing . . . . . . . . . . 110 Distributional Properties and Security Returns. . . . . . . . . . . . . . . . . . 113 Test Design . . . . . . . . . . . . . . . . 114 Statistical Tests: Tests of Means, , , , , 121 Specific Hypotheses: Percent AFUDC . . . . 129 Change in the Percent AFUDC . . . . . . 134 Method I Versus Method II Firms . . . . 140 Percentage AFUDC . . . . . . . . . . . 143 Test Design: 'Quarterly Reports . . . . 147 Data Selection and Validation . . . . . 150 Summary . . . . . . . . . . . . . . . . 154 V. Empirical Results: Discussion of Statistical Tests. . . . . . . . . . . . . . . . . . . . . 150 Summary of Null and Alternative Hypotheses . . . . . . . . . . . . . . . . 151 Tests of Percentage AFUDC: Case A. . . . . 152 Tests of Percentage Change in AFUDC: Case A . . . . . . . . . . . . . . . . . . 155 Percentage AFUDC: Case B . . . . . . . . . 159 Percentage Change in AFUDC: Case B . . . . 17o Method I Low Percentage Versus Method II. . 173 Method I High Percentage Versus Method II. . . . . . . . . . . . . . . . . 175 Method I Positive Percentage Change AFUDC Versus Method II . . . . . . . . . . . . 178 . Method I Negative Percentage Change Versus Method II . . . . . . . . . . . . . 131 Tests of Quarterly Earnings Announcements . 182' Summary . . . . . . . . . . . . . . . . . . 186 :u-c :- v"! V ., ghoul. d' A, ‘Y 7:595. Q". U‘u- . fin. EH“ 3 " DOA. q q .1 .- “Rina-dd OO ‘I‘U‘II 1- .- 01V»... .. ““a*1 “~5vsn to ”L a»; u‘\ I I". 1 -O~...‘,: iv VI. Summary, Conclusions and Areas for Future Research . . . . . . . . . . . . . Summary . . . . . . . . . . . . Conclusions: Method I. . . . . Method I Versus Method II . . Limitations of Research . . . Future Research . . . . . . . Appendices Appendix A: Empirical Tests of Price Earnings Ratios and Market-to-Book Ratios. . . Appendix B: Earnings Changes and Effects on Stock Returns . . . . . . . . . . . . Bibliography . . . . . . . . . . . . . . page 188 188 193 194 197 198 202 217 224 (’1 w; Icy LIST OF FIGURES Overview of Rate-Making Process. . . Accounting Entries for the Allowance for Funds 0 O O O O 0 O O O O O O 0 Simplified Example of a Hypothetical Income Statement for a Utility, Showing the Effects of AFUDC on Reported Income. . . . . . . . . . Illustration of the Rate-Making Process. 0 O O. O O O O O O O O O 0 Example of Timing Differences for Tax Purposes of AFUDC. . . . . . . Example of Interest on Interest Timing Differences . . . . . . . . Time Line Showing the Time Period From the Start of Construction to End of the Plant Life . . . . . Present Value of the Cash Flows From $1 of Allowance Income. . . . The Separation Argument for Allowance for Funds. . . . . . . . 33 36 38 46 49 51 67 .0 ‘- . gob-b .\ ‘1’. Table 10 ll 12 LIST OF TABLES Distribution of Companies by Year and by Percent of Income Repre- sented by AFUDC. . . . . . . . . . . Mean and Standard Deviation of AFUDC as a Percentage of Reported Income Available for Common: 1970-1977 . . A Comparison of Market Value to Book Value Ratios Between Industrials and Utilities for the Years 1960-1974. A Comparison of P/E Ratios Between Industrials and Utilities for the Years 1960-1977. . . . . . . . . . . Method I Companies . . . . . . . . . . Method II Companies. . . . . . . . . . Percentage AFUDC: Method I Firms High EPS Forecast Error Versus Method I Firms Low EPS Forecast Error: Case A . . . . . . . . . . . Percentage Change AFUDC: Method I Firms High EPS Forecast Error Versus Method I Firms Low EPS Forecast Error: Case A . . . . . . Percentage AFUDC: Method I Firms Matched on EPS Forecast Error: case B I O O O O O O O O O O O O O 0 Percentage Change AFUDC: Method I Firms Matched on EPS Forecast Error: case B O I I O O O O O O I O O O O 0 Low Percent Method I Versus Method II. High Percent Method I Versus Method II vi Page 12 14 90 92 152 153 164 166 171 ' 172 175 177 M V. 1. Vii Table Page 13 Positive Percent Change: Method I versus MethOd II 0 O O O O O O O 0 O O O 18 O 14 Negative Percent Change: Method I Versus Method II . . . . . . . . . . . . 183 15 lst Quarter Results. . . . . . . . . . . . 184 16 Market-to-Book Ratios for Method I Fims. O O I O O O O O C O O O O O O O O 210 l7 Market-to-Book Ratios for Method II Firms. I O O O O O O O O O O O O O O O O 211 18 P/E Ratios for Method I Firms. . . . . . . 212 19 P/E Ratios for Method II Firms . . . . . . 213 20 Market-to-Book, P/E Ratios and Standard Deviation of Monthly Returns. . 214 21 Calculations for P/E Ratios and Market-to-Book Ratios. . . . . . . . . . 215 22 Percentage Change in Annual Earnings: A Comparison Between Method I and Method II Firms. . . . . . . . . . . . . 219 a“: u cor-in ‘..,. 1:11.." ( Digiu. a "'m *0 ‘0... V ‘ e 9:,“ _. .. ' \ ‘...d‘~ ~vm g. ACKNOWLEDGMENTS I would like to thank my committee members Dr. Alden Olson and Dr. Richard Simonds for their guidance and insight during the months of this study. I would especially like to thank my chairman Dr. John O'Donnell for his numerous helpful comments and perserverance in the prepara- tion of this dissertation. Thanks must also go to the Public Utility Institute at Michigan State University for partial funding of this work and to Kirk Butler for his very able assistance in computer programming. We r~ ‘v on y;& ' ' I w-ouap 5.1,. ‘ Liv-3:5 hue. A"... ‘ '0':¢:e 0‘ ’:'-‘-=“nr ~-.5_‘ by; I - Q ".-—:‘Q q“ ~0UQ...‘.‘ “ n 'IV’AT‘AV ‘ \ ~“HcUuu‘ C“ 1 INTRODUCTION The purpose of the present study is to assess whether there is any information contained in the dis- closure of "allowance for funds used during construction" (hereafter, AFUDC), in the financial statements. This information may be unanticipated and therefore of use to investors in properly adjusting to changes in common stock values. In the study presented here, changes in common stock values can be tested by examining the immediate changes in market valuation, thatis the distribution of stock returns, for electric utilities. Justification for this study is fourfold: 1) Previous research has not been conclusive with respect to capital markets reactions to AFUDC earnings; 2) Previous research has concentrated on the total valuation of firms with AFUDC earnings. The previous tests have used different .accounting numbers and ratios, such as cost of equity capital, price-earnings ratios, market-to-book ratios and (Ithers, as surrogates for the true market valuation of these firms. To the author's knowledge, there have not been any tests focusing on the immediate market valuation or change in market valuation due to the disclosure of AFUDC earnings, which is the topic this research addresses ; '...oq~ ”null I 8"”? “v: “V. . q wan: I6 v coo—v U ...‘..'5 . I!- -...'..§ -'nun \ n... lab-Al, : I-.. ~ ‘Ol:’ 'n... .‘ . ~u... a . “Q. N. u.» “ ti 5 ‘;,. ~' Wu . :I:“‘: 7‘1““ Q - H: In: ‘5 En“ TI , ‘1 I1.‘ “:3. a V ' H n.‘ J '. :F rt"‘ \ \ l I V“ .H“ ~ "4 . u 6 N‘ v- .“ 5., I ..' a. I U‘ u)‘ . .. 'n 5‘ \“. :1- ”a 1., 3) Previous empirical tests of the differences in market valuation between method I and II firms is extremely limited. Briefly stated, method II firms include construc- tion work in progress in the rate base while method I firms do not. This research addresses the change in market valuation and any differences between these methods; 4) This design is a substantial refinement compared to other studies. The research design has been used to examine other accounting policies and the immediate capital market reaction to them, but never in the case of AFUDC. AFUDC is examined for any information content that might affect the distribution of stock returns for elec- tric utilities. Disclosure of both the amount of AFUDC and the magnitude of AFUDC as a percentage of total earnings is examined for some impact on the investors' assessment of the distribution of stock returns for electric utilities. There should be some perceived differences in the risk associated with each component of total earnings that will be reflected in common stock prices. The argument concerning AFUDC implies that AFUDC earnings are different quality earnings when compared to earnings from operations and therefore, of differing risks. The concept of "qual- ity of earnings” is discussed at length in Chapter One. Briefly stated, the quality of AFUDC earnings is different because of the relationships of these earnings to cash flows, the timing of the cash flows and the variability of AFUDC earnings. Given the premise that the quality u :- vd' - IDA! .‘ h “w in 5.- 1". RQ. c l Iggy- Iob' to a ‘0-‘1‘ ‘9... . tbb;“ "-5.. ““N- \ "C'U. lg...‘ ""qu \, I. n.“ .. .1. sl ‘. u“..- N .s of AFUDC versus operating earnings is indeed different, then there should be some perceived difference in the market price of equity securities which can be tested. Any differences in the risk of equity securities for electric utilities may affect the ability of electric utilities to raise more capital through common stock offerings. This eventually will impact the amount the customers of these utilities will pay for electric needs. AFUDC has been a long-standing and important area of contention among followers of the electric utility in- dustry. Accounting techniques for recording AFUDC have existed since the early 1900's. However, beginning in the mid 1960's, the debate has intensified. This is due to the fact that the percentage of total earnings repre- sented by AFUDC increased on average, five to six times. This large and growing percentage of total earnings available to common stockholders, that is represented by AFUDC, has been accompanied by increasing skepticism in 1 The reason for this skepticism the investment community. is the perceived lower quality of electric utilities' earnings and consequently the ability of utilities to pay cash dividends from earnings. If the percentage of total earnings represented by AFUDC is higher than the dividend payout and remains high, then the utility may be forced to sell debt or equity to pay their cash divi- dends. an! gu (I) who " In. -- \vu 'P.F'A tub... . 2" up " >4. 5q.‘ u.‘: i “a. 5": p i In As a result, utilities may be forced to sell more securities than the market is willing to bear, thus increasing the securities' cost and thereby hindering capital formation in the electric utility industry. This problem will continue to be an area of contention as construction periods lengthen, inflation increases, interest rates continue to increase and remain volatile, and more militant consumers accerbate the problem. This study is divided in the following manner. The first chapter outlines a description of AFUDC and a discussion of the nature of the problem in detail. The chapter continues with a presentation of the regulatory process and the use of AFUDC in the context of this process. A mathematical framework for AFUDC is developed and the differences between the two specific applications of AFUDC are also formulated and described. The two applications of AFUDC are defined as method I and method II. Chapter Two reviews both the major defenses and criticisms given for AFUDC as these concepts have been developed in the pertinent financial literature. Chapter Three is the literature review, including a discussion of time empirical tests performed in this field. Emphasis is pflaced on the results of these tests and some of the statistical problems which have biased these results. The appendix to Chapter Three contains empirical tests of price-earnings (P/E) and market-to-book (M/B) ratios, ‘1'. N .uwai 4' 1 0A --:-' .r cu.- vi. u . co ""“V;P‘_‘ p... 9%“! mung. ' LC. ...e u v 1 I .1," m .‘b., I I . '2"-~: 'n.‘.“‘ at ‘_ A bu A. n o VA 5. A I. .¥‘-: t a“. - Q g ._‘D: c ft.” . . ,n: and shows how they differ between method I and method II firms. Much of the work presented in the first three chapters provides evidence suggesting AFUDC earnings are different and more risky than earnings from Operations. Nevertheless, it seems the controversy has intensified instead of lessened. ,Therefore, as the use of AFUDC has increased, the topic has come under increased scrutiny. In fact, one recent gubernatorial race in New Hampshire focused on the different methods of accounting for AFUDC as something more than a minor difference between the two candidates.2 Chapter Four is a short review of the four aspects of relevant financial theory pertaining to this study. These aspects are integrated into the general methodology employed in this research. The final section of this chapter contains the specific application of this metho- dology and a detailed description of the specific hypo- theses that are tested. As already stated, the primary question is whether the disclosure of AFUDC is useful to investors in assessing stock returns. Therefore, the time period surrounding the month of disclosure is used to examine tflue monthly stock returns of two portfolios. One port- folio is identified as the treatment and the other portfolio is identified as the control. The classifi— cation of firms into the treatment and control portfolios is discussed in Chapter Four. Two basic strategies are . c ; we ”2' fl anv18u ‘ v I ‘layo'fit . 1 VA“ . .3- "grab . II n vuanpnay:$ .u’vv'LIvv.u~ b n ' 014.. fin - ugvibot-‘J I I Q 0 IA I ‘9...th I. nfi UQI¥V I . Qa-und T zo-cvu & ‘ s " ‘IA 7‘- '- but y: ‘0‘ II. - ’:,“Rv\1 "Hives.- 0 bin- ‘75 - -' .~2 'hp'. . 'v--..=e . n 9- A _ _v-:“ ti..".‘ . I. A C -. “:a ‘I ‘ .‘ A . e. ':~:Q «,‘~ ‘ '. ‘5'. v: Q U... . . ‘h‘ 'I U 1 .“ \ ”A. . u s -'s F n‘.‘ a .Q ‘~“-. ..“ 1 O . (h (I) employed for the formation of the treatment and control portfolios. First, treatment portfolios are formed to incorporate firms using the method I technique for re- flecting AFUDC while the control portfolios are composed of method II firms. The second strategy employs only method I firms. The first are classified into portfolios on the basis of what percentage of total earnings is represented by AFUDC. High percentage firms will comprise the treatment portfolios and low percentage firms will comprise control portfolios. Betas, representing the relevant risks of each firm, are then calculated and weighted within each portfolio so as to ensure each pair of treatment and control portfolios has equivalent risk as measured by the beta of the portfolio. The hypothesis states that the information contained in the disclosure of AFUDC earnings has not been adequately anticiapted by the market and therefore has not been impounded in the measure of relevant risk, beta. This hypothesis can be tested by examining the monthly stock returns of port- folios with equivalent betas. The significance of the difference between the nunsthly stock returns of the treatment and control portfolios is evaluated using Hotelling's T2 test, which is the multivariate analog to the simple T-test between means. Also, in a number of cases, it is apprOpriate to use the simple t-test between means as the statistical test. 1‘ '0' ' oat Ve' itacter S; "I C 4... U qua6‘. a. 4.6» ufi¥es n . " ARQQ I ‘1‘ 1\ I? “UV‘UUB‘ u ‘ .‘ -’Hl .‘va “=V.OS g. \- “1:. mm; fl "3- wa.L _ . “1“] fix '- .':" “ o . v . ‘A”:.H‘ “V. - :bs bk..“i‘ _ . C‘ Gnu \ In. C " "‘5 MQK. " L531 any 5 D 3:31“! b4 :1le, a: p h {5' a“; 3 "ya ls: \ :-.'.:ehce Chapter Five contains a detailed discussion of all the tests performed and the results of these tests. Chapter Six offers the conclusions, limitations and areas for future research. This study will hopefully shed some light on the practices of AFUDC.. The market's reaction to the disclosure of AFUDC and any information content this disclosure might have is important to investors and to the public utility commissions responsible for the regulation of electric utilities. Any empirical evidence pertaining to AFUDC will help the public service commis- sions make better judgments concerning the use of AFUDC. Also, any evidence presented here should enable investors to make better judgments concerning changes in AFUDC policy, and the future effects this might have on the firm's valuation. Finally, the study will provide some evidence on the question of semi-strong market efficiency. Footnotes 1This argument could be inferred from a number of authors. Among them Charles Tatham, "Interest During Construction and Price-Earnings Ratios," Public Utilities Fortnightly($eptember 27, 1973), pp. 32-36, Thomas G. Marx, Market-to-Book Return on Equity Correlation," Public Utilities Fortnightly (December 4, 1975), pp. 28- 31. Also a number of references at the end of Chapter 3. 2Donald D. Holt, "The Nuke That Became a Lethal Political Weapon," Fortune (January 15, 1979), pp. 74-77. CHAPTER 1 DESCRIPTION OF ALLOWANCE FOR FUNDS USED DURING CONSTRUCTION This chapter contains three sections. The first section develops the concept of allowance for funds used during construction. The regulatory process and how AFUDC is used within this proceSs is presented. The second section develops the differences between method I and method II. The last section is a mathematical and theo- retical development of the earnings stream due to AFUDC. The cash flows of this earnings stream is developed and shown to be less than the income reported. 1.1 Definition of Allowancewfor Funds USedjguringConstruction AFUDC is the implicit cost of equity and the explicit cost of.debt a utility will pay for its construction program. It represents the capitalized interest on a construction program, where the term "interest" includes both debt and equity costs. If“ r .51 O" 1 5 to ahow a: to e r: " "‘9: w" 3.. is.) \ £113 to cc: ':«:b (L «ii I. 5‘ 9'1"“ n Isc‘ly.‘ .‘ n ‘ ' ‘1 "O- ‘. “A ":.D ‘ULJ :1". “is“ l 1“ m 10 "Theoretically, the purpose of capitalizing AFUDC is to allow utilities to recover the capital costs Of, and to earn return on, funds devoted to construction."1 In other words, during the construction period the utility adds to construction work in progress the costs of capital used to finance that construction. Later, when the con- struction is completed and placed in service, the construction costs including the costs of capital are depreciated over the life of that plant. These additional financing costs can then be recovered through additional depreciation charges which are considered a cost of service expense. Basically, AFUDC represents the income a utility would have earned on the funds devoted to a construction project if those funds had been used for some other pur- pose. In the case of AFUDC, even though reported as income, the income is not actually earned and there is no immediate associated cash flow. It is an accounting entry reflecting future cash flows that will result when the completed plant is allowed in the rate base. For clarity of future discussion, a definition of earnings must be given here. Reported book earnings will consist of two components. The first component is "real" earnings signifying earnings that actually give rise to cash flows. Another way of expressing this is to call these earnings "Operating earnings." The author realizes that perhaps not all operating earnings give rise to immediate cash flows, but let's assume this is true to RI 1'0 .9' ‘9‘ ”II ”no my. :H ’“. n.. 5‘! 11 prevent any misunderstanding of what these earnings represent as Opposed to AFUDC earnings. The second component of re- ported book earnings is AFUDC earnings. Future sections contain extended discussions and examples why AFUDC earnings may not be considered "real" earnings by investors. In all future discussions, "real" earnings will signify earnings from operations and other earnings will signify AFUDC earnings. 1.2 Background on AFUDC AFUDC is an established part of utility accounting originating in the early 1900's. It has always been an area of contention in the regulatory field. But, starting in the mid-1960's, the debate intensified among people familiar with and responsible for electric Utility regu- lations. This is due to the fact that AFUDC as a percentage Of net income available to stockholders has grown dramatically from the mid-1960's until the present time. An examination of Table 1 clearly shows the trend towards much higher amounts of AFUDC as.a percentage of reported book net in- come. In a sample of approximately 130 electric utilities taken from.the compustat tapes, the average AFUDC as a percentage of net income was 4.7% in 1966, rising steadily tr) 24.2% in 1975 before declining slightly in 1976-77. The continuance of double-digit inflation in 1978-79 and the increased cost of building new power plants have only 12 .mommu umumsmeoo Eonm Honusm an pomoHo>oo "mousom OOHcmmEoo mo HOQEDZe O.HN O.NN N.¢N 0.0N 0.0H OH m.MH 0.0H N.O H.O >.¢ m .O>< NOOION N N O N m H mmhioe O O m H NONIOO O N N H H NOOIOO N O O m H H H NOOIOO h O O O H H wmmIOO n N m O m N H H N NOOIOO 009mm O O O m N m N wmqaos an omu O OH O N O O 4 wownmm ucmmmudmu OH N O HH HH O m m H wmmaom cOEEoo on NH NH OH HH OH HH O O H NONION OHQMHHSVo OH NH O O OH O NH O N H NONION oEoocH OH NH OH OH OH OH OH OH O m H NONIOH m0 O OH NH OH OH OH OH. NN HN OH HH v NOHIOH OH OH OH OH OH ON NN ON Hv NO mm wOHIm OH OH OH OH HN ON ON mm NO mm *OO mm IO ONOH OOOH OOOH mbOH NOOH HOOH ONOH OOOH OOOH OOOH OOOH mummy UODMd an concomoumom mEoocH mo unmouom Ob pom “mow On OOHcmmEOU mo OOHuanuumHo H «Hams l3 served to continue this upward trend. Duff and Phelps, specialists in utility security analysis, report 59% as the average in 1979.2 Even more important is the increase in the number of firms with large amounts of AFUDC. In 1966, only two firms had AFUDC percentages greater than 15% of reported net book income. In 1971 there were forty-one firms greater than 25%. While in 1976, thirty—five firms had over 35% of the total reported earnings represented by AFUDC . The companies that comprise the sample for this study appear to have a slightly higher averages of AFUDC when expressed as a percentage of total earnings. Table 2 gives the average percent by year and the standard devia- tion for both the method I and method II firms included in this study.3 There are three observations one can make from this table: 1) The average percent varies from year to year among the firms that comprise both groups. 2) There is a substantial difference in the average percent each year between method I and method II firms for most of the years presented here. 3) The standard deviation is quite large with respect to the mean for each year and for both groups of firms. This implies the percent of total earnings repre- sented by AFUDC varies substantially from firm to firm within a particular year. Utilities are typically classified as income stocks because they pay large percentages of their earnings out 14 Table 2 Mean and Standard Deviation of AFUDC as a Percentage of Reported Income Available for Common .1970 - 1977 Method I Method II Standard Standard Year Mean Deviation Mean Deviation 1970 12.85% 10.32% 11.40% 6.24% 1971 17.78 14.08 19.06 9.97 1972 21.91 15.77 27.57 15.57 1973 24.63 18.15 32.07 17.75 1974 27.33 20.35 33.67 17.32 1975 32.81 21.93 37.38 14.59 1976 34.54 23.00 32.14 15.93 1977 30.21 17.99 38.22 17.73 Source: Developed by author from Compustat data. mg: I 04‘ 15 in dividends. Dividend payouts of 60-70% are not at all uncommon in the utility industry. For the majority of firms studied (method I firms), there is no immediate cash flows associated with AFUDC earnings. Therefore, in later years of the 1970's, a number Of electric utilities were paying more in cash dividends than they had available in cash from earnings. Security analysts have expressed concern over the growing size of AFUDC earnings because they are not current cash flow earnings. These earnings are only claims on future revenues. A question of the firm's financial status arises when sub- stantial amounts of the firm's earnings do not give rise to cash flows and therefore impair the firm's ability to pay cash dividends. Mr. Richard Walker, managing partner of the Regulated Industries Division of Arthur Andersen and Company, testified before the Florida Public Service Commission on June 18, 1973. He states, Investors apparently do not value a dollar of earnings from the allowance for funds charged to construction as much as they do a dollar of earnings from Operations, for in this sense they view it as being of lesser quality than cash flow.4 Many others have reached similar conclusions. Mr. W. T. Hyde, economist and public utility consultant, tes- tified before the North Carolina Utilities Commission, Investors can hardly be expected to give much value to earnings so heavily dependent on the credit for interest charged to construction which results from nothing more than an arbi- trary credit and an assumption that the plant under construction will produce sufficient a) 0hr? r—rmg. (32.1 i w .W: “- «in ”.11 I ('1 u» . "a... “a. i“: f I ’1 no '0- .‘ ‘3. 16 earnings to offset the decline in this credit when the plant is placed in service. This is a highly speculative and problematical assump- tion.5 The recording of an allowance for the use of funds results in a significant increase in net income without a commensurate increase in cash flow.6 Clearly a number of people, including many experts in the field, consider AFUDC to be of inferior "quality" when compared to earnings from operation. The above comments suggest that this is primarily the result of the cash flow implications, both the timing and amount Of those cash flows. Two questions immediately arise: l) Exactly why are AFUDC Considered inferior "quality" earnings, where quality is defined and discussed below; 2) If those earnings are inferior quality, do market agents value or react to those earnings differently than to earnings from operations? In other words, does the dis- closure of AFUDC earnings produce some signaling that the market uses to assess the return distribution of that stock? "Quality" of earnings is a nebulous term that first must be defined adequately to facilitate an understanding of AFUDC. "Quality" of earnings in the context of AFUDC can be defined as the value inherent in each dollar of AFUDC earnings. This concept of lower "quality" earnings as represented by AFUDC is based on two distinct components. First, it is the relationship between reported earnings and cash flows. Numerous authors argue that investors are concerned with cash flows rather than accounting earnings 17 for several reasons. One is the ability to pay cash dividends as a return to common equity investors. Another aspect Of this cash flow problem is the reinvestment im- plications. Any cash not paid out in dividends can be reinvested in the firm to earn more cash and therefore pay additional dividends in the future. The investment community might be skeptical of earnings, such as AFUDC, that do not give rise to cash flows, in the same manner as earnings from Operations. This question of cash flows is compounded by the timing differences between AFUDC earnings and earnings from Operations. Also, by the dis- parity between the timing of cash inflows and outflows. (The question of the timing and the mathematical relation- ship Of AFUDC earnings to cash flows will be discussed further in later sections Of this chapter). The cash out- flows to pay interest payments on debt and the dividends on preferred and common equity are both current and certain. AFUDC earnings, which is the income "earned" during the construction period, do not give rise to immediate cash flows. They are postponed until the plant is completed and goes into service and then they are recovered over the depreciable life of the plant. AFUDC earnings are really claims to future cash flows. Thus, because AFUDC earnings are further in the future, they are more uncertain and therefore more risky than operating earnings. The second component of "quality of earnings" is the 7 variability of those earnings. Two-earnings streams may 18 be thought to be of different quality because their vari- ability as measured by their standard deviations are different. In this context, the variability of AFUDC earnings may be different than the variability Of earnings from Operations. This study addresses the more fundamental question of the timing of AFUDC earnings and a judgment on the variability of AFUDC is not necessary at this time. In summary, AFUDC earnings are considered of inferior quality primarily due to the timing of the associated cash flows. Even so, there are three different opinions among industry Observers concerning the quality of AFUDC earnings. The three groups are: 1) Those who believe AFUDC earnings are as good as earnings from Operations; 2) Those who believe AFUDC earnings are worthless; 3) Those who believe AFUDC earnings are of "lower quality." The first two groups are the minority, while the last group, which holds the intermediate position, seems to have the largest following. The majority of Chapter 2 contains the defenses and criticisms of AFUDC expoused by these three groups. It must be emphasized that it is rare to.find industry observers who consider AFUDC earnings the same as earnings from Operations. Most observers con- sider AFUDC to be of different quality and therefore in- vestors should value AFUDC earnings differently. In a major piece of empirical work on this quality question, Bowen8 has recently found evidence to substantiate the claim of the third group, that is; AFUDC earnings are of inferior clv ' a o-q,‘ l“'. :- n25 . ‘h. "H ‘e. §... 5.- O. ‘0" . \:I‘ . 19 quality. There is some contradictory evidence on this quality question, which will be discussed in Chapter 3. 1.3 Regulatory Process Public utilities are natural monopolies, therefore, the government has seen it in the best interest of the public to regulate these companies. This is accomplished through several levels of government control. At the federal level, there are two principal regulatory agencies. The Federal Energy Regulating Commission and the Federal Communications Commission provide the framework for account- ing regulations, reporting requirements and general regu- lation of utilities engaged in interstate. The primary regulatory body resides at the state level. Here, the public service commission in each state imposes the specific regulations on each company which supplies services in that state. Regulation tries to accomplish the same goals for public utilities that competition achieves for unregulated competitive enterprises. This implies that the regulatory agencies must allow a rate of return sufficient to ensure the financial integrity and continuing existence of the utility. Balanced against this is the fact that utilities are monopolies and therefore the agencies must ensure that consumers are not forced to pay monopoly profits. Essen- tially, regulation means the regulation Of earnings. Of mnmse, in practice regulation means much more than the shufle regulation of earnings. Virtually every aspect 20 of the utility's Operations are regulated in some way. For purposes of this research, we are principally con— cerned with the earnings Of an electric utility. There- fore, it is convenient to think of regulation as the regu- lation of earnings. The major tenents of regulation were set forth in the Hope_case in which the U.S. Supreme Court stated, It is important that there be enough revenues not only for Operating expense but also for the capital costs of the business. These in- clude service on the debt and dividends on the stock...By that standard the return to equity owner should be commensurate with risks on investments in other enterprises having corresponding risks.. The return, moreover, should be sufficient to assure confidence in the financial integrity Of the enterprise, so as to maintain its credit and to attract capital.9 But, a qualifying view was also noted in an earlier Supreme Court case. In the Natural Gas Pipeline case the court stated, . . .the utility gets its return. . .by rates sufficient, having in view the character of the business, to secure a fair return upon the rate, provided the business is capable of earning it. But regulation does not insure that the business shall produce new revenues...10 In other words, regulation provides a utility with the Opportunity to earn a fair return, but the return is not a guaranteed return. In summary, the Supreme Court cases state that a utility is allowed to earn a fair return on assets enabling it.to cover its operating expenses and capital costs. This return should be equivalent to a return earned by a up (lb 1 2"? I...' H 21 competitive enterprise with similar risks. The capital costs implied are the interest on debt and the dividends paid to both preferred and common stockholders.r The rates are set in a formal hearing before the public service com- mission where the rate of return, the rate base and a schedule of rates is approved. In the illustration of this computation that follows, assume the following facts: Rate base 10,000,000 Cost of service expenses including taxes and depreciation, but ex- cluding the return to capital, that is, divi- dends and interest charges 1,000,000 Allowed rate of return 12% The commission would allow a rate schedule to be approved which should allow a total of 2,200,000 in revenues. This would be a cost of service expense of $1,000,000 plus the allowed return to capital of 1,200,000 (12% x 10,000,000). An excellent overview of the entire rate-making process is provided in Figure l. A brief discussion of the major points of contention involved in regulatory proceedings follows. 1.4 Rate Base Obviously, from an examination of the exhibit a criti- cal question in the regulation Of electric utilities con- cerns the determination of the approPriate rate base. Two Emincipal issues effect this rate base. First, what items should be included or excluded. The rate base is not the 22 MA MB Wham Todflm Tomi Rash-am . Alb-«Rem 93W . fiDeh fiCemnerquuy D 7051““? X X x o .. Alla-dam PM mu Underline 00qu D Dividends IwuneOu' Wfifinflbu uwufieo: namanuh {Debt officiated dignity Afia'l'ua O r Tana I Mince dictum ‘ o X WM: A Befon'l'ue L halal: . ' PL mm :[ MW , o ”C"*‘ lamudhdh mm Ind'l'ana O [ Opauiang I Toullkveeue W Source: Figure l Overview of Rate-Making Process P-JC MPaUni: on . D Tod Rm Rum" w Output In Unit: ofScme' iifiuhuUdt Dennis T. Officer, “Issues in Public Utility Regu- lation," Arizona Business, January, 1981, p. 19. O-': n. 23 same as total assets. All public service commissions agree that the plant in service belongs in the rate base, but the agreement stops there. In the context Of this study an important point must be made here. Most com- missions do not allow construction work in progress (CWIP) in the rate base. That is, plants under construction but not yet completed are excluded from the rate base. The commissions have not felt it proper to force present customers to pay for future electric supplies. Second, the dollar value of the rate base is also important. This is concerned with whether a firm is allowed to use fair values or replacement values in the calculation of the rate base. Most firms are forced to use historical costs, which is original cost minus de- preciation. In an era of high inflation the historical COStS are substantially less than fair values. Therefore, increasing the rate base by using fair values would in- crease the prices customers would have to pay. This is a very controversial topic and it is beyond the scope Of this research to answer this question. Therefore, there will be no further discussion of fair values versus his- torical costs. 1.5 Rate of Return In theory the allowed rate of return equal the com- pany's cost of capital. Therefore, theoretically, utility rates are set so as to cover all expenses including capital costs. But, there is a basic problem in that most utility 24 commissions have ignored current financial theory relating to the cost of capital calculations. This problem has two aspects. First, the regulatory agencies have traditionally allowed the utility sufficient revenues to cover only the historical or embedded interest costs on long-term debt. The agencies do include the current cost of debt into the calculation of the cost of debt capital. This is the average cost of debt. However, in recent years, current interest rates are much greater than past interest rates. In other words, the current cost of debt is greater than the average cost of debt. When a utility finances a con- struction project, it muSt borrow at current rates, not at these historical or embedded rates. Therefore, there does appear to be a major difference between the rates a utility is required to pay on the debt funds used for con- struction and the revenues they are allowed to earn on this debt component. Whether this is inequitable in view of the ongoing nature of utilities and the way in which the appropriate rate of return is determined is perhaps debatable, but not of concern here. Second, in the determination of the appropriate rate of return to equity, commissions tend to use the "comparable earnings" standard. For this standard, the commission computes the rate of return earned on the book value of the equity for both regulated and non-regulated industries. The utility is then allowed to earn a rate of return on 25 the book value of its equity based on the rates other firms with similar risks earn.ll There are two major problems with this approach. First, there is no straight forward or unambiguous method for determining equality of risks. Second, accounting methods are quite different between regulated and unregulated firms. Thus, rates of return based on book values makes comparisons very dif- ficult. Therefore, due to the combination of these factors, it is only by chance that the allowed rate of return equals the company's incremental cost of capital.12 1.6 Expenses Allowed The public service commissions also regulate the cost of service expenses which can be recouped through revenues. The basic expenses allowed are Operating, 13 There are certain maintenance, depreciation and taxes. expenses such as advertising, research and prerequisites to executives which may not be allowed. These expenses are not pertinent for purposes of this study and will not be discussed further. 1.7 Regulatory Lag When discussing rate of return a concept that cannot be ignored is regulatory lag. Regulatory lag refers to the lag in time from the date the utilities present their case to the commission until the rates are actually billed to customers. Due to the increased militancy of consumers 26 and therefore, longer commission hearings, the problem of regulatory lag has become more important. During this time lag, the rate base, the cost Of service expenses, and the cost of capital may change and result in a dif— ference between expected and actual income. A few state commissions have attempted to alleviate this problem through automatic adjustment clauses fOr expenses and by projecting changes in the rate base and the cost of capital.14 Yet, the majority of commissions do not follow this procedure and the problem of regulatory lag is now compounded by rapidly fluctuating interest rates. 1.8 Regulation and Allowance for Funds Construction work in progress (CWIP) was discussed in an earlier section. It is an extremely controversial topic among regulators. In fact, most commissions do not allow plants not yet completed (CWIP) in the rate base. Commissions felt it is unfair for current users of utility services to pay a return on a utility plant that will benefit future customers.15 The amount of other income reported as AFUDC in any year is a function of CWIP and the allOwed rate of return. Therefore, the inclusion of CWIP in the rate base gives rise to a crucial difference in AFUDC among utilities. Those firms that include CWIP in the rate base are known as method II firms, while the majority of firms that don't allow CWIP in the rate base are method I firms. This dichotomization will serve as the basis for a number of empirical tests in later chapters. 27 In the regulatory framework there are two principal arguments given for the use of AFUDC. These are the separation and matching arguments which are discussed at length in chapter two. Briefly, the separation argument states that the real purpose of AFUDC is to separate the current Operations from the effects of a construction program. The matching argument is a basic premise of accounting theory. Revenues should be matched with the expenses that generated those revenues. The recording of an allowance defers the cost Of financing construction activities is premised on the basic regulatory principal that current customers should pay a return only on those assets that are currently performing a useful service.16 The utility industry is then faced with a dilemma. It is thought unfair to charge current customers for future services, but at the same time to properly match the costs and benefits of this service potential it must recognize the costs of construction. The best solution in the view Of public service commissions is AFUDC. AFUDC allows a utility to report as income the capital costs of the con- struction program. This accomplishes two purposes. One, the incOme from and the effects of the construction program are separated from the income for current Operations. Second, the costs of the construction are properly matched with the revenues from the construction program. 28 1.9 Brief Historygof AFUDC The concept is part of the basic philosophy Of regu- lation for utilities. The states of New York and Wisconsin provided for the capitalizing of interest paid on funds used to finance construction starting in 1909. This was concerned with just debt funds. The cost of capital was not recognized until 1914 when the Interstate Commerce Commission's system Of accounts provided for both debt and equity to be capitalized.17 The Federal Power Com- mission adopted a similar position, which was called interest during construction (IDC), starting in 1922. Even though the concept of recognizing all the costs of capital was developed early, the term allowance for funds used during construction (AFUDC) was not in common use until the late 1960's.18 1.10 Determination of Capitalizinquates ____v._ A very serious practical problem arises when trying to determine the appropriate rate for the capitalization of allowance funds. That is, identification of the specific funds, which are used to finance a particular 19 In fact, financial theory suggests that this project; identification if possible, would not be correct. The Separation Theorem indicates that the financing decision must be separate from the investment decision. Therefore, the determination of the proportion of a project cost pro- vided by debt or equity is a moot point. The rate should be the firm's marginal cost of capital according to financial 29 theory. In practice the commissions allow a rate that varies considerably from year to year and among utilities in different states. 20 The construction work in progress account is multiplied by some rate, which is then considered the allowance for funds used during construction for that particular year. The_rate is not set so as to capitalize a certain amount of interest, preferred dividends and certain return on common equity. In fact, utilities have different ways Of calculating this rate, but for the most part, the rate is not related to the firm's marginal cost of capital. Some idea of the divergence among utilities is presented: In a Generally, it has been the practice of utility companies to credit Interest Charged Construc- tion at a lower rate than the allowed rate of return; usually the rate used has been the pre- vailing short-term interest rate.21 different study, where the author examined the dis- closure of the rate, he stated: It appears that most companies are not using a rate that is based directly on the cost of funds required for construction, but rather one that is considered representative of the industry. . .footnote disclosures in many cases described the rate as based upon, but less than the cost of capital employed to finance the construction program.2 Another author in analyzing the trends in the allowance rates notes: In analyzing the upward revision it is quite obvious that.these rates did not cover the rapidly rising capital costs. For example, in 1968, when the cost of senior capital to electric utilities hovered at 7% and the cost of equity capital was considerably higher, most firms were capitalizing AFC at less than 3O 7%. Likewise, for 1969 and 1970, when debt costs alone soared to ranges of from 8.5 to over 10 percent, the corresponding capital- izations rates continued to lag considerablyy23 Finally, in a 1973 Federal Power Commission survey, it was found that only one commission (Puerto Rico) allowed the allowance rate to be the same as the allowed rate Of return. Twelve of the fourteen other interpretable responses indicated that the rate was geared to the interest rate on debt.24 But there may be a considerable difference between this rate and the cost of debt. An examination of current annual reports, reveals 8-9% is not uncommon when the cost Of debt hovers around 12-l4%. It appears the rate used to capitalize these capital costs of con- struction has very little theoretical validity. 1.11 Accountingfor the Allowance A discussion of the actual accounting entry to record AFUDC will perhaps help clarify the situation. The account- ing entry is a debit to the asset account, construction work in progress and a credit to the allowance for funds used during construction. The later account is then closed to Income Summary. The entry follows: Construction work in progress XXX Allowance for funds used during construction XXX An example of the accounting ledgers is given in Figure 2. The first part of the exhibit shows the actual accounting entries, a debit to the asset account (CWIP) and a credit to the revenue account (AFUDC). The next section of the 31 Year 1-5 Accounting Entries CWIP 8,000 AFUDC 8,000 CWIP Begin 100,000 AFUDC 8,000 ' 8,000 yr 1 8,000 8,000 yr 2 8,000 8,000 yr 3 8,000 8,000 yr 4 8,000 8,000 yr 5 Total I40,000 Income Summary AFUDC Year 1 8,000 ' 8,000 Year 2 8,000 8,000 Year 3 8,000 8,000 Year 4 8,000 8,000 Year 5 8,000 8,000 40,000 40,000 Depreciation equals 28,000/year, given a total asset value of 140,000 and the use of straight-line depreciation. Figure 2 Accounting Entries for the Allowance for Funds 32 exhibit is an example of the actual ledgers over the five year period. The actual number are discussed in the explanation Of Figure 3. The last section Of Figure 2 shows the closing of the AFUDC account to the Income Summary account each year. This 8,000 is then included as income in each year. Once an annual amount is added to the CWIP account, it cannot be distinguished from other amounts in the CWIP account. The AFUDC is then closed to income summary. Starting in 1977 a slightly different procedure was used to report this income. It now appears as two separate line items on the utility's income state- ment. The portion of AFUDC attributed to debt funds appears as an offset (reduction) to interest expense. The portion attributed to equity funds appears under the section "Other Income." Even though this does separate the two components, the effect on the total income re- ported is the same as previous to 1977. ‘ Figure 3 is an attempt at recreating a simplified income statement using the data presented in Figure 2. The first three years are left out of the exhibit for purposes of clarity since they are identical to years 19x4 and 19x5. The tax treatment of AFUDC may be different due to timing differences, but again for purposes of clarity a separate example is presented later in this chapter to exemplify the tax timing differences. The example can be divided into two five year sub-periods. Years 19x1 through 19x5 are the construction period. During these years AFUDC 313 oeoucm neuuoeox so un=h< uo caveman ecu wcHzozm .OOHHHOD e no» acoEoOeum osoocu HeuHuoruochz O OO oHeEexm OOHOHHOEHm m OHOOHm .osem orb OH O=OEOOOOO osoocH one so uuoOOo one use once ob OHbmusteuue ucocoaaoo orb ”mucoEwoo oau OusH peacuweom 30: OH oa=h< one .soHueuHHeEHe s OH erh .OuHsco ob OHbmustuuue assess may OOO.OOH.H . OOO.OOH + AOO0.0 x no + OO0.000.H oxOH OH OOOO cram .uuo» co>Hm New cH on=m< Oo uc=OEc use chu by com: eHsh .muwoa m uo>o coHueHooueov I OxOH .H .seO so OOH>qu sH vooeHc useHm .NO I mHzo so assume no can» vosoHH< NoH I mean mush co unseen no Ouch poaoHH< HOOOOO «Om NNH an area Nam an mocecHh ooc.ooH w I OHIO sH ucaaueo>cH OOO.OOO.HO . OOOO OOHOOHuxo «man «One OHso eoHceeEoo H corner. "msoHueesoe< .Oomcoaxu :oHucHoouaop HucoHuHcpm one cu one economm ooo.mNm one OHIO munch :H ewesoexmv .aoxau ovsHesH noncoexmo .Oomcoaxo mo0H>umm Oo umoo amuue comb cues are co cusses NOH e ovH>ouO o» mosco>ou.cueo cu vosoHHe OH OuHHHus uou sH vooeHe OH usmHe cos: muse one an eueeuuucH omen cue» urea OOO.~OH O OOO.~OH O OOO.NOH mammmmwlu OOO.OnO.H OOO.O~O.H OHKOH OOO.mOH O OOO.mOH O OOO.mOH OO0.0~O OOO.nnO.H OO0.0nO.H OXOH ooe.wcH o ooe.moH o coc.OCH coo_mNO coc.cno.H occ.O usoocH ooc.O OO0.0 unoueucu I coo.wO cOo.wO emchuem mcHueano OOO.OOO OO0.000 uaoocuaxm coo.OOO coo.OOO . buosso>o¢ oco.ccc.H occ.coo.H Queen Guam meH QKOH 34 are recorded as earnings without an attendant cash flow. The addition of AFUDC to the construction work in progress account (CWIP) during this construction period causes the asset value to increase by the amount of the AFUDC. In this case the asset value increases by $8,000 per year for a total of $40,000. This increases the asset value of the asset account (CWIP) to $140,000. This $8,000 per year is recorded as other income, AFUDC. The $8,000 per year can be found by multiplying the CWIP account, $100,000, by the allowance rate (rate of return allowed on CWIP) of 8%. Notice that the total income the utility reports is $8,000 greater than the revenues they are actually allowed to collect from customers. The next step occurs when the plant is completed. The CWIP account is allowed to enter the rate base in year 6, thereby increasing the rate base from $1,000,000 to $1,140,000. At the same time the depreciation expense increases by $28,000 due to the addition of CWIP to the rate base. This additional depreciation expense really has two components, $20,000 is the original cost of the plant depreciated on a straight-line basis over the five year life. $8,000 is the depreciation due to the addition of $40,000 of AFUDC to CWIP over the construction period and the subsequent depreciating of this $40,000 over the five year life of the plant. The $8,000 can now be col- lected from customers as a cost of service expense. The problem is that AFUDC is reported as $8,000 of income in 35 years l9x1—19x6 and the $8,000 is not actually collected from customers until years l9x6-l9xlO. The rate base in years 19x6-19x10 declines by $28,000 per year reflecting this depreciation expense. Also, revenues in years l9x6-19x10 also decline reflecting the lowering of the rate base each year. Years 19x1 through l9x5 show an interest expense of $6,000 per year which is the $50,000 of debt times the cost of 12%. The assumption is that the debt is retired at the end of the fifth year. Again this may be a simplification since a utility could be expected to pay off the debt with the proceeds of the new plant, but for purposes of clarity it was thought necessary to use this format. The crucial point that one may glean from this example is the recording of $8,000 of income in years l9xl through 19x5 when that income is not currently a cash flow and will not be a cash flow until some future time, years 19x6 through 19x10. Figure 4 substitutes actual numbers into the theoretical framework of Figure 2. This may offer the reader additional insight into the process of revenue determination for an electric utility. The effect of an increased rate base and additional depreciation charges are evident from this exhibit. 1.12 Taxes and the Disclosure of AFUDC The Internal Revenue Service does not consider AFUDC income as taxable income. AFUDC is a device used purely for rate-making and financial reporting purposes. 36 To Equity ‘ 4.5M.“ O mmm ' .. 05.54,“, 8I0.0oo.000x .01 . 8 700.000 quuited Profit Mm- “.121” O * mu 5 Ptefetted * W 59”“, Required Ptofit “* we we ‘ ammtua tummy» x " x . . . etunt hmglate l l Dmdendlue OnEquity L Intetest Expeme 1 “0.“.on .0. _ 8 «my» Hfi ‘tmmmma 4.0% 1% 4.3% Iefote lntetett “LIN.” l I I J I 1 L & Tue: :- ° ' o Depreciation (Manet!) ] ‘ LII”!!! O OwnwnC~u (Assumed) _._____‘”'m'°m "1""- euro Tate! Revenue .30. .000 '90.!!!”me .IS - Allowed Return to Capital xyz Power Co. Total Revenue Requirements xvz Power Co. Dennis T. Officer, "Issues in Public Utility Regulation," Arizona Business, January, 1981, p. 19. Source: Figure 4 Illustration of Rate-Making Process 37 Therefore, the recording of allowance income does not in- crease the company's income tax liability. In the year the AFUDC credit is recognized, book income exceeds tax- able income by the amount of the credit, but in later years book income is less than taxable income due to the difference in depreciation expense allowed. The depre- ciation expense due to the allowance is not a tax deduc- tible depreciation, therefore there is a reduction in book income compared to taxable income. AFUDC then causes book income to be greater than taxable income in earlier years, but this difference "turns around" in later years. This is a classic example of the timing difference for inter- period tax allocation. Figure 5 is a simplified example illustrating the timing differences for AFUDC. The total tax paid over the six year life is $18,000. In the first three years book income exceeds taxable income by $2,000, the amount of AFUDC. $1,000 is the debt portion and $1,000 is the equity portion. Only the debt portion creates a deferred tax liability which will be eliminated in years 4-6. In years 4r6 taxable income exceeds book income by $1,000 which is the additional depreciation charges due to the capitalized debt portion of AFUDC. However, the total amount of AFUDC reported as income does not represent a timing difference between book and taxable income. There are two aspects which may affect the differences between book and taxable income. The timing 38 oomev oomem oooeb oooew oooem ooo.~ ooo.m mxma UQDMd mo OmOodem xmB How OOUGOHOMMfio mcflefla mo magamxm OO~.¢ mmmqm OOO.~ OOO.O oooem ooo.~ ooo.m mme m ousmwm mIe Onmmm aw BHmm Omosomu I H>\ooo~ mo .omumoo Hmsofluwood OO~.« mmmwm OOO.> OOO.O ooo.m oooem ooo.m exma oom.m oom.m OO0.0 OOO.OH oooea ooo.m oooea ooo.m oooeoa mxma OOOOOO m + name x oom.m oom.m ooo.m oooeoa oooea ooo.m oooea ooo.m ocoeoa Nme wow u open xma m o H H GOCM3OHH¢ OO0.0~ u mHso oofinom coHuosHumcoo HemmIm oom.o oom.m ooo.m cooeoa ooo.a ooo.m oooea cooem ooo.oa HXmH OMHH vomnoum ummem ”msowumssmmfi oEoosH uoz Omxma mEoocH manmxma wEoosH xoom ucmsomfioo avenge UQDM¢ + usmcomaoo unmo 099mm + umoumucH I BHmm 39 difference, which is explained above, and a permanent dif- ference. A permanent difference results from income re- ported for book purposes which is never reported as tax- able income. For example, income on a municipal bond which is of course federal tax exempt. Recall that AFUDC con- sists of two components, the amount attributed to debt financing, and the amount attributed to equity financing. The portion of AFUDC related to debt financing is a timing difference. The interest expense related to debt funds is tax deductible immediately in the period incurred for tax purposes, but is deferred and expensed as a depreciation expense over the life of the plant for book purposes. The portion attributed to the equity financing which is considered "Other Income" is a different case, a case of permanent differences. Since this component is due to the opportunity cost foregone by investors it is of course not a tax deductible expense for any corporation. There- fore, this portion of AFUDC, which is classified as "Other Income" gives rise to a permanent difference between book and taxable income, for which no inter-period tax allocation is appropriate.25 In summary, only the $1,000 attributed to the debt funds is a timing difference which reverses in later years. The Securities and Exchange Commission does not have any specific pronouncements related to AFUDC on the utility's income statement, except one discussed below. However, most utilities do disclose in footnotes to their financial 40 statements, the nature of the allowance and the rate used to compute the allowance. Recently the SEC has required one or more of the following disclosures in registration statements filed with the commission on a case by case basis: 1) Disclosure of the portion of net income available to common stock which is attri- buted to the common equity portion of the allowance for funds used during con- struction. 2) An exhibit to the registration statement which illustrates the computation of the common equity portion of the allowance. 3) Deduction of the allowance, or a portion thereof, from sources of funds from oper- . ations in the statement of changes in financial position.26 The other federal agency, which does impose industry wide accounting requirements, is the Federal Energy Regu- latory Commission. The requirements are: (1) To specify the account title, allowance for funds used during con- struction, for the income statement, (2) To define the allowance for funds used during construction as "the net cost during the period of construction of borrowed funds used for construction purposes and a reasonable rate on other borrowed funds used for construction purposes and a reasonable rate on other funds when so used,"27 and (3) To 41 specify that the entire amount be deducted from sources of . . . . 28 funds from operations in reports filed Wlth the commiSSIOn. In 1977 both the SEC and FPC issued a new requirement. The total amount of AFUDC must be separated into two com- ponents as stated earlier.29 These are the debt component and the equity component. 1.13 Differences Between Method I and Method II Many commissions use method I, a few use method II 30 It is im- and a fair number use a combination of both. portant that the difference is clearly understood. An illustration accentuating the differences follows, assume the following facts: 1) Rate base excluding CWIP until completion of the plant ‘ 10,000,000 2) Investment in CWIP including 100,000 of capitalized allowance for funds from this year 1,000,000 3) Allowed rate of return on rate base 12% 4) Allowance rate for AFUDC (applied to CWIP) to get current year's AFUDC 10% The basic difference between method I and method II is that CWIP is included in the rate base for method II and CWIP is excluded for method I. For method I companies, the current year's allowance for funds and the firm's 42 investment in CWIP are both ignored in setting current revenue requirements.31 Rates charged to consumers would then be set to provide income of $1,200,000 (the allowed rate of return of 12% times the rate base of $10,000,000) for method I firms. Method II firms are quite different because both the current year's allowance for funds and the investment in CWIP are considered in the rate making process.32 Even though method II does allow CWIP in the rate base, they do not allow the current year's capitalized allowance for funds to be included. This prevents the utility from cap- italizing the allowance and then earning a return on it in the same year. This would be double counting. The procedure is really two steps. First, the method II firm includes CWIP in the rate base. In this example, the initial amount would be $1,320,000. This is the allowed rate of return of 12% times the rate base, including CWIP, of 11,000,000 (12% x (10,000,000 + 1,000,000) = 1,320,000). Second, the current year's allowance for funds is subtracted from this total revenue figure. In this example, the cur- rent year's allowance is $90,000 (10% x 1,000,000 - 100,000). The total revenue figure minus this year's allowance yields the return an utility would be allowed to earn. Here, $1,320,000 - $90,000 equals $1,230,000. This is an important point. Remember for method I firms, the income the utility is allowed to earn is $1,200,000. The full cost of the construction program 43 12% x 1,000,000 = 120,000 is borne by future customers. Method II firms split the cost of construction between current and future customers. In this case $30,000 (1,230,000 - 1,200,000) is paid by current customers and the majority of the cost (the $90,000 deduction in the determination of income allowed) is paid by future cus- tomers. The next section discussed the interest on interest question and the effect this has on the amount of AFUDC. Also, Figure 6 provides examples of this split cost under different alternatives. 1.14 The Concept of Interest on_Interest Interest on interest is an important concept which very few commissions allow. In a 1970 survey of 130 electric utilities on this question, the authors state: ". . .of the (130) utilities that reported capitalizing AFC, only four reported that they, at one time or another, compounded the allowance, that is, left prior AFC capitalized in the ex- 33 penditure base in computing the current portion." Interest on interest pertains to the question of, whether or not previously capitalized allowance for funds should be included in construction work in progress for this year's allowance computation. In the example above, if the commission allows this year's allowance for funds to be 10% of $1,000,000 or $100,000 it is allowing interest on interest. The concept really has to do with compounding interest because the $100,000 of previously capitalized allowance for funds is allowed to earn an additional $HL000 (10% x $100,000), of allowance for funds in this 44 year. Since most commissions don't allow this, the amount that would be the allowance in this year is $90,000 [10% (1,000,000 - 100,000)]. One can also see that this amount which is not allowed, compounds year by year over the life of the construction period. Considering the construction period may be 10 to 12 years, this could be a substantial amount which the utility is not allowed to earn. This question can be examined from another perspective, that is, the rate of return allowed on this allowance amount once it has been added to the construction work in pro- gress. For method II, without interest on interest, the allowance for funds dollars are immediately invested at the rate of return allowed by the commission. Also, since they are allowed in the rate base from year 1 (start of construction) through year N (end of life of the plant) when the plant is fully depreciated, these dollars earn the rate of return allowed by the commission for the same period of time in which income is reported. Again using the previous example, the $100,000 of previously capital- ized allowance earns a return of 12% in this year. Con- tinuing another year and assuming the only change is the allowance entry made in this year, then the next year the amount previously capitalized will be $190,000 (100,000 + 10% (1,000,000 - 100,000)). This amount will then earn 12% in the next year or whatever rate the commission allows.34 The example could be continued far into the future, but the critical point is that under method II, 45 theoretically from the year the allowance is recorded until the plant is fully depreciated, the allowance dollars are invested at the appropriate rate of return. The important point is that under method II the interest question is not relevant. As stated before, for method II firms, interest on interest results in the parti- tioning of the return into the amount received in the current period and the amount to be received in future periods. If interest on interest was allowed in the first year, the $100,000 or previously capitalized allowance would still earn 12% or $12,000. But, $10,000 (10% x 100,000) would then be added to the rate base to be received in future periods (through additional depreciation expenses) and $2,000 would be received in the current period. The interest on interest question affects method I firms inla far different fashion when viewed from the per- spective of the rate of return allowed on the allowance for funds dollars. First, if interest on interest is allowed, the allowance dollars are invested during the construction period at the rate used to capitalize the allowance. Turning again to our example, if interest on interest is allowed, the $100,000 of previously capital- ized allowance dollars earn a return of 10% or $10,000, for the current year. Second, if interest on interest is not allowed, the allowance dollars are invested during the construction period at a zero rate. For example, the $100,000 of previously capitalized allowance produces no 46 Ooosoumuuwo mcwswa uuououcn so unmumusn mo mumaexm w muomam .Oofiun> u::OEO Os» mucuouosu .Oomcmno cusuou anuou can use acmumsoo OsaOEmu cannon acouuso use H oozuoe you .Omaun> Oswaau ozu OHOMOuozu .so«uuom uneasy use coduuom acouuso e owed omow>wu OH u« use Ewan HH panama n u0u scannws u.~ no men Ow susuou aeuou on» onno >uo>o :H .OOOOOoasoo OOOOO can .OOH Oo Ouau Omzosaa .OOH uo ouuu OOOOO I o oamo .OOOOOoOeoo no: ooame can .OOO mo mung eozoadu ..O~ yo mama OOOOO I O OOOo .ooccsomeoo so: Doomd use .mOa mo oueu om3oHAo .oo mo oueu oo=h¢ I 4 omen ooo.oo~ n Dasha oonwamuammo >an:0«>0ud mo ooo.o~w mswoauocq mazo :« acmEuOo>sH omouo>< OO0.000.0 n muse ozqesaoxm Oman 00oz «Oscaudasumt OO0.0HO OO0.0HH OO0.0H~ OO0.00~ OO0.0HO OO~.OOH . Ousuoz deuce OO0.0~ OOOqOO OO0.0 OO0.0 .OOm¢~ room.» .oaamee maze 0» cocoa Ounces: OO0.00~ OO0.00~ OO0.00~ OO0.00~ OOO.~OH OO0.00H Ono» acouuso 0:» new money ca noosaoca cucumm fiance mo acquuom OO0.00~ OO0.00~ OO0.0H~ OO0.00~ OO0.00H OO~.OOH Ouzuom sauce ooowmw coo.” OO0.0 O OOO.H O OOOOO Omuwnuuuauo OHOOoO>Oua :o Ousuoz OO0.0 OO0.0 OO0.0 OO0.0 OO0.0 OO~.O OOOOO unannouamuo >~O50~>oud mcaosaocw mnzo so susuom OO0.00H OO0.00H OO0.00H OO0.000 OO0.00H OO0.00~ 00e>u0O an Ocean :0 assume OH H HH H HH H U OOOU n OOOU t undo 47 return in the current period. Even with interest on interest, the returns on the previously capitalized allowance for method I, are in the form of an additional allowance until the plant is placed in service at which time they will be recouped through additional depreciation charges (cost of service expenses). Contrast this to method II, where the returns on allowance dollars can take two forms: 1) Immediate cash flows if interest on interest is now allowed or 2) Parti- tioned between immediate and future cash of interest on interest is allowed.35 Summarizing for method II firms, the compounding (interest on interest) effects the timing of the utility's return on construction work in progress, but these factors do not affect the amount of the return. On the other hand, for a method I firm the compounding effects the amount but not the timing of the utility's return on construction work in progress.36 A method II firm should prefer no interest on interest while a method I firm should prefer to use interest on interest.. Since the majority of firms don't allow interest on interest it points out another difference between method I and method II firms. Method I firms are again penalized with respect to method II firms. Figure 6 on the previous page illustrates these differences between method I and method II firms.37 48 1.15 Mathematical Treatment of Differences Between Method I and Method 1138 v—vv—w v7- Method I Firms. Let Cl be the amount of allowance in the plant accounts at the time when the plant is placed in service. Let C1 represent just the allowance for 1 year, all other years of the construction period will be identical. The future value is then (1 + r)N2Cl where r = return allowed on the rate base by the commission, and N2 = the number of years for plant life. If this sum is discounted to the year when the plant is placed in service, the present value is .N (1 + r) C (1+ r) 2 But now if C1 is discounted back to the date the allowance income is reported, (time 0 + N1) the reported income will exceed this discounted value of Cl‘ Figure 7 on the next page presents a schematic drawing of this concept. The interest on interest rate is important in the N 1.= Co (1 + i) 1 where i = interest determination of Cl' C on interest rate and N1 = construction period. If i = 0, then C1.= C0. Therefore, the difference between the allow- ance income reported and the present value of the related cash flows depends on three factors: 1) Length of the construction period (N1) 2) Interest on interest rate (i) 3) Opportunity cost of funds (assume r = k = marginal cost of capital) 49 OMHA ucmam we» no com ou sofluosnumcou mo uumum was scum eonumm mafia 0:» OOO3oOO OOOO mafia h whomflm cusumn mo Oven Umsoaad u u mums ummumusfl so uOmHmucH u H o mafia um_mocmzoaam may no ussoem M o0 Hz + 0 mafia um mosmzoaam may no usaoE< u HO OMHH ucmflm u NZ coauom coHuosuumsoo u Hz AH+HO Hz H 0 Ho 1w. o u H AO+HO o z mow>umm ca omomam o ucmam . _ N. I. _ «MHA Osman q ooflumm coauoouumcoo 50 A major difference between method I and method II is that under method I the allowance amount at time 0 does not compound (interest on interest) over the construction period (N1) to give the value C1. It is really the value Co. Then when discounted back over Nl years to time 0, the income reported is greater than the present value of C1' The present value of each dollar of Cl can be found by solving the following equation (1 + i)N1 N (1 + r) l where the parameters are as defined above. Figure 8 shows the calculations for different para- meter values when this equation is solved. An examination of the exhibit reveals a startling fact. There are sub- stantial differences between each dollar of allowance in- come reported and the present value of the related cash flow. For example, 1) In the typical case where no interest on interest is allowed and the opportunity cost of funds is 12%, the ratio of in- crease in value to income reported de- creases from 89% to 32% as the con— struction period increases from 1 year to 10 years. 2) Even if interest on interest is allowed at the rate of allowance, which is still 51 Figure'8 Present Value of the Cash Flows From $1 of Allowance Income Opportunity Cost of Funds 12% Length of Interest on Interest Rate Construction Period 0% 8% 1 .893 .962 2 .797 .925 3 .712 .889 4 .635 .855 S .567 .822 6 .506 .790 7 .452 .760 8 .404 .731 9 .360 .703 10 .322 .676 9% .971 .943 .915 .888 .863 .837 .813 .789 .766 .744 52 less than the Opportunity cost of funds, and if the differential is 3%, the ratio decreases from 97% to 74% over the 10 year period. Therefore, for method I firms, there are substantial dif- ferences between the income reported and the present value of the income actually earned over the life of a construction project. Method II Firms. Under this method construction work in progress is included in the rate base. It remains in the rate base from year 1 until year N where N is time when the plant is fully depreciated. Therefore, the value of the firm is increased by the present value of the annual return on the allowance plus the amount Of the allowance. Let C2 be the allowance, r the return allowed by the com- mission, N1 the construction period, and N2 the life of the plant to which the allowance is added. The future value N1 + N2 of the cash flows is then (1 + r) C and the present 2 value of this stream at the time when the income is re- ported is N +N (1 + r) l 2 N1 + N2 C2 (1 + r) or C2, which is the amount of income reported. Therefore, for method II firms, the present value of the increase in value of the firm due to the allowance is equal to the amount of the income reported. 53 Summarizing, it was pointed out earlier that the "quality" of earnings refers to the timing and amount of the cash flows related to those earnings. The above dis- cussion provides a mathematical framework within which the cash flows of AFUDC earnings are shown to be signifi- cantly less than the cash flows of earnings from Operations for method I firms. This indicates that the "inferior quality" connotation that many investors place on AFUDC earnings does have some theoretical justification. In the case of method II firms this may be an unfair conno- tation, since the cash flows of AFUDC earnings are equiv- alent to the cash flows of earnings from operations. The mathematical treatment of AFUDC and the comparison of method I with method II both suggest that there is a difference between the income reported and the income or cash flow actually earned by the utility. The discrepancy between the income numbers gives rise to the argument about the ”quality" of these earnings. It is not hard to see why a large number of people associated with electric utilities might be concerned with the market reaction to a utility's earnings. Particularly when the percent of earnings represented by allowance for funds used during construction hovers around 40-50%. Since disclosure of the item is minimal and the practice of capitalizing this component varies by state and by year, it is not surprising to find that AFUDC comes under attack as one of the reasons ‘:fin:the decline of market valuation of electric utilities. In the ne provide I have a d actually "- It.“ | uni-€29! far fu: t2. 54 In the next chapter empirical tests are discussed which provide evidence that those earnings attributed to AFUDC have a different market reaction due to their increased risk. The market then values those earnings differently than earnings from operations. In other words, earnings actually earned through revenues from customers are valued differently than earnings due to an accounting entry. There certainly seems to be some justification for a study that will present evidence concerning the immediate market reaction to AFUDC earnings. Hopefully the evidence presented here will be useful in the regulation of electric utilities and aid investors in evaluating those securities. 1.16 Summary In this chapter first, an overview of the regulatory process was presented. Second, the concept of allowance for funds used during construction and how it fits into the regulatory process was presented. Special attention was noted of the fact that the primary purpose of AFUDC is to prevent current customers from paying for future supplies of power. Third, a very brief history of the concept was traced to the early 1900's. Fourth, the deter- mination of the appropriate capitalization rate and the specific accounting for AFUDC was detailed. Fifth, the scarcity of formal disclosure requirements by the SEC and FPC was discussed; Sixth, the interest on interest question was discussed with implications for both method one and 55 method two firms. Finally, a mathematical development was presented for method I and method II and for AFUDC in general. 56 Footnotes 1Lawrence Pomerantz and James E. Suelflow, "Allowance for Funds Used During Construction: Theory and Application," (East Lansing, Michigan: Division of Research, Michigan State University, 1975) p. 75. 2Discussion with Thomas Harrimon, Senior Utility Analyst, Duff & Phelps on July 2, 1980. 3Differences between method 1 and method 2 will be discussed at length in later sections of this chapter. Briefly, method 1 does not include and method 2 does in- clude CWIP in the rate base. 4Richard Walker, "Testimony and Exhibits of Richard Walker for Arthur Andersen & Co.," Florida Public Service Commission Docket No. 72609-PU, June 18, 1973. 5Affidavit by Mr. W. Trueslo Hyde, Jr., filed with the N. Carolina Utilities Commission in Re Duke Power Co. Docket No. E-7, Sub 128. 6Arthur L. Litke, "Allowance for Funds Used During Construction," Public Utilities Fortnightly, (September 28, 1972). pp. 19-22. 7O'Donnell, John L., "Relationships Between Reported Earnings and Stock Prices in the Electric Utility Industry," The Accounting Review, (January, 1965), p. 135-143. Pro- fessor O‘Donnell was one of the first to examine the quality of earnings question and the affect it may have on stock prices. 8Robert M. Bowen, "Valuation of Earnings Components in the Electric Utility Industry," The Accounting Review, Vol. LVl, No. 1, January, 1981, p. 1-21. 9F.P.C. v. Hope Natural Gas Company, 320, U.S. 591 (1944).. 10Natural Gas Pipeline Co. of America v. F.P.C., 315, U.S. 575 (1942). 11Each of these items is defined and discussed in the following pages. Also, Figure 2 is a schematic presentation of these items. 57 12Stewart C. Meyers, "The Application of Finance Theory to Public Utility Rate Cases," The Bell Journal of Economics and Management Science (Spring, 1972), p. 59. 13Dennis T. Officer, "Issues in Public Utility Regu- lation," Arizona Business, January, 1981, p. 21. 14Murray L. Weidenbaum, "Variation in Public Utility Regulation," Public Utilities Fortnightly (October 24, 1974), p. 30. 15Everett L. Morris, "Capitalization of Interest on Construction: Time for Reappraisal?" Public Utilities Fortnightly, (March 4, 1971), p. 26. 16Arthur L. Litke, "Allowance for Funds Used During Construction," pp. 20-23. l7Ibid. 18The account title Allowance for Funds Used During Construction was established by Order No. 436 issued by the Federal Power Commission in August, 1971. 19Paul B. Coughlan, "Interest During Construction: A Little Appreciated Cost," Public Utilities Fortnightly (September 28, 1972), p. 19. 20Pomerantz and Suelflow, p. 93. 21Dennis R. Bolster, "Should Plant Under Construction be Included in Rate Base?" Public Utilities Fortnightly (May 27, 1971), p. 27. 22Lyle M. Dahlenbury, "Allowance for Funds Used During Construction - Calculation and Disclosure Problems," Public Utilities Fortnightly (January 17, 1974), pp. 68-72. 23Pomerantz and Suelflow, p. 93. 24Federal Power Commission, Federal and State Commis- sion Jurisdiction and Regulation of Electric and Gas Utilities, 1973 (Washington, D.C.: U.S. Government Printing Office, 1973), p. 64. 25Bunch, c. Robert, "The Tax Effects of AFUDC: Financial Accounting Aspects," Public Utilities Fortnightly (August 14, 1980). A detailed treatment of this topic is given in this article. 26Dahlenbury, "Calculation and Disclosure," p. 17. 58 27Federal Power Commission, Uniform System of Accounts Prescribed for Public Utilities and Licenses (Washington, D.C.: U.S. Government Printing Office, 1973). 28Federal Power Commission, Accounting Release No. AR-10 (Washington, D.C.: Federal Power Commission, 1971). 29Federal Power Commission, Order No. 561 (Washington, D.C.: Federal Power Commission, 1977) and Securities Ex- change Commission, Accounting Bulletin No. 15, Washington, D.C.: 1977. A 30Bolster, "Should Plant Under Construction," p. 27 made the initial distinction between method 1 and method 2. 31 Ibid, p. 28. 32Ibid, p. 28. 33Pomerantz and Suelflow, p. 73. 34It must be recognized that this rate of return is what the utility should earn, not what they actually earn. The same problem exists with the other assets of the utility, what they are allowed to earn and what they actually earn are very likely to be different. Second, the rate of return allowed can change from year to year. 35This treatment of interest on interest draws heavily on Johnson's work in his dissertation. 36Johnny R. Johnson, "Losses on Investment in Con- struction Work in Progress," Public Utilities Fortnightly (September 14, 1978), p. 22-29. 37Ibid, p. 23. 38Johnny R. Johnson, "The Allowance for Funds Used During Construction: Market Reaction to Current Accounting Practice," unpublished dissertation, Virginia Polytechnic Institute and State University, 1976, pp. 45-50. This mathematical treatment of the differences between method I and method II is first developed here. CHAPTER 2 DEFENSES AND CRITICISMS OF ALLOWANCE FOR FUNDS USED DURING CONSTRUCTION This chapter reviews and discusses the major defenses and criticisms of AFUDC as these have been presented in the financial literature. First, an examination of the principal defenses given for the practice of AFUDC, with the emphasis placed on the validity of these arguments. The counter arguments and criticisms of this practice are then reviewed. It should be noted that these arguments are not necessarily the views of the author. The dis- cussion is merely an attempt at delineating the pros and cons of this controversial topic. Finally, the chapter concludes with a presentation of "other utility income," that is, depreciation and deferred taxes. AFUDC is com- pared to depreciation and deferred taxes in terms of the timing and amount of their respective cash flows. 2.1 Defenses 2.1.1 Matching. The matching concept as defined in accounting theory is frequently given as a defense for AFUDC. 59 60 Matching implies that any costs associated with an asset must be matched with the revenues that asset produces for a correct determination of income in that period. Eldon Hendriksen gives a succinct summary of this idea. When costs can be reasonably associated with revenue, they should be charged to expense in this same period in which the related revenue is recognized. . . .When costs can be reasonably associated with specific revenue of a future period. . .they should be carried forward as deferred charges and matched with the related revenue when it is recognized.1 Defenders of AFUDC argue that the capital costs, both debt and equity costs, associated with the construction of a new plant are no different than the costs of labor and materials. Proper accounting then requires these costs be capitalized, that is, added to the costs of that asset, and expensed through additional depreciation charges when the plant is complete, that is, when it starts producing revenues. They conclude, this results in the proper matching of all costs associated with the construction of this asset and the revenues produced by this asset.2 Defenders of this argument present a two case scenario much like the following. In the first case, the plant is purchased complete from an outside contractor. The alternative scenario is construction of the plant by the utility. It seems illogical to conclude that the cost of the plant should be different depending on how it is acquired. Surely the contractor would in- clude his capital costs during the construction period 61 as part of the price of the plant. Therefore, it is appropriate for the utility to capitalize their capital costs and thereby add these costs to the price (asset value) of the plant. These capital costs consist of both the explicit costs of debt and the implicit costs of equity. Proponents of this argument go to great lengths to justify equity costs as "true" costs of construction. Yet, no justifi- cation is really needed. Financial theory indicates there is no difference between these two costs. Equity owners have just as much right to a fair return as debt holders. The opportunity cost foregone by equity holders is as valid as any cash outlay, therefore these equity costs should be capitalized just like the interest costs on debt. In fact, the Federal Energy Regulatory Commission advocates the separation of these components on the in- come statement. Prior to 1977, the capital costs were not separated into the two components of debt and equity. But, starting in the 1977 annual reports, the Federal Energy Regulatory Commission by Order No. 561 mandated that AFUDC should appear as two entries. One entry re- flects the AFUDC provided by debt funds, while the other entry reflects the AFUDC provided by equity funds. Nevertheless, even though financial theory considers equity costs a legitimate cost, proponents of the matching concept for AFUDC still go to great lengths to justify these costs, and by this justification conclude the matching 62 argument is then true. Defenders argue that the cost due to the equity investment in the plant under construction is more important in the utility industry. "This equity component of AFUDC applies only to regu- lated industries, for its creation is properly allowed by utility commissions and cannot be recorded justifiably by managements of industrial companies."3 This argument is based on the fact that for nonregu- lated companies, stockholders could forego a return on the investment in the plant during this period of con- struction because of the possibility of larger returns in future periods, thereby compensating investors for the lost return in the construction period. Since electric utilities are regulated there is no possibility of abnormal returns, profits above those allowed by the commission, in future periods. Therefore, the opportunity cost fore- gone on equity funds is a true cost just like any cash outlay. It should then be capitalized. One author argues that these equity costs represent a created asset, which is included in the rate base, there- by earning the company a rate of return or cash inflow. By adding these capitalized equity funds to the asset value, the asset value increases, and therefore, increases the depreciation charges, which result in a cash inflow recouped through ”cost of service expenses." He argues it is a claim on cash similar in nature to accounts re- ceivable.4 63 In summary, this defense relies on the concept of matching as presented in accounting theory and on the justification of equity costs as real costs, in fact more real than in the non-regulated sector. The criticisms of this view are now presented. The matching argument does seem to be deficient in a number of areas. First, this author fails to understand how the justification of equity costs make the matching argument true. Financial theory indicates the opportunity costs foregone on equity funds are a very real cost in both the regulated and non-regulated sectors. Arguing that these costs are more real in one sector appears to contradict financial theory. Therefore, justification of the matching argument based on this concept is apparently erroneous. Second, the allowance for funds on equity is not truly similar to accounts receivables. Earnings and there- fore cash flows for accounts receivables are not recognized and reported, until those earnings are actually realized. In contrast, allowance for equity funds results in earnings reported on the income statement long before actual reali- zation. There appears to be a substantial difference in the risk between these two assets due to the time factor. The time from realization of income to accounting for the asset value in the case of accounts receivable is less than one year. The allowance for equity funds may have a time lag of ten years or longer, depending on the length 64 of the construction period. It seems tenuous to make the analogy between accounts receivables and the allowance on equity funds since there is a real difference in the time between reporting earnings and realizing cash flows. The next two deficiencies relate to the allowance for funds in total, not just to the equity component. One problem is the fact that the allowance for funds does not result from the systematic allocation of interest, preferred dividends and return on common equity. This concept of the capitalization rate is discussed further in Chapter Two in the section "Determination of Capitali- zation Rates." The crucial point needed for the dis- cussion here, is that the rate used to compute the allow- ance for funds is not the average or marginal cost of capital. The rate used is determined arbitrarily and only by accident would the rate used equal the average cost of capital. The second problem is that the capitalization of an allowance for funds will not result in the same cost of plant that would occur if an outside contractor had built the plant. The rate of return on any asset should be commensurate with the risk of that asset. Therefore, the asset risk should not be affected by who builds the plant, the cost should be the same for both the contractor and the utility. Practical considerations of the regulatory climate make this application of matching the required return with the risk of the asset virtually impossible. 65 The utilities use an "embedded" cost of capital. The utilities' calculation of the cost of capital is based on an average of all outstanding debt issues. This is more properly identified as an average cost of capital, not a marginal cost of capital. Secondly, the allowance rate, which is used to capitalize these debt and equity costs for the construction project, is not equivalent to the average cost of capital. The allowance rate is set lower than the average cost of capital. Therefore, the construction firm could use the marginal cost of capital, the average cost of capital or some other rate, adjusted for the risk of the projedt, but it is unlikely that the rate the contractor chooses is the same as the allowance rate the utility uses. It appears the statement, "the costs should be the same whether the plant is built by an outside contractor or by the utility itself," is apparently unrealistic. 2.1.2 Separation Argument. This defense is based on the premise that the construction and operating activities should be separated.‘ Earnings per share for any utility should be the same during the construction period as they would have been if there had been no construction program. An illustration shown in Figure 9 is presented on the following pages.5 Defenders of this approach argue the construction program should not affect earnings per share. This is exactly what is achieved by recording the allowance for 66 funds, if the allowance rate equals the firm's cost of capital. An examination of the illustration reveals company B's earnings per share drop relative to company A's and company C's EPS remains the same as company A's given the same construction program. The only difference between company B and company C is that company C includes an allowance for funds when calculating their earnings per share. Company D is the last case presented. In this case the earnings per share are not equal to company A's because the allowance rate is not equal to the marginal cost of capital. One can see a substantial difference in EPS does result from uSing an allowance rate less than the cost of capital for the utility. The separation argument seems to be deficient in at least two respects. The argument is based on the fact that net income should not be allowed to fluctuate due to construction activities. In other words, net income should be shielded from any effects the construction pro- gram might have on net income. Today, construction activities are an integral part of most utilities opera- tions. .To defend a procedure that results in a net income figure, which is independent of a significant economic event occurring during some period, it not only illogical, but also does not agree with generally accepted accounting principles. This is a particularly important point given the nature and activism of some consumers towards nuclear power plants. By no means is a utility assured of completing 67 Figure 9 The Separation Argument for Allowance for Funds Assumptions 1. Three companies A, B, and C are identical except A has no construction program but B and C do. 2. C capitalizes an allowance for funds used during con— struction at the marginal cost of capital but B does not. 3. At the end of l year, A purchases a plant identical to the completed construction owned by B and C. 4. Company D capitalizes the allowance at 8%, a rate less than the marginal cost of capital. 5. All companies began the year with the following capital structure: Book , Amounts weights Bonds - 8% 27,500 .55 Preferred Stock - 8% 5,000 .10 Common Stock - $10 par 17,500 .35 6. The new construction of $10,000 is financed as follows: Bonds - 10% 5,500 Preferred Stock - 10% 1,000 Common Stock - $10 par 3,500 7. Cost of new equity is 17.85%. 8. Tax rate is 50%. *Marginal cost of capital: Source After Tax Cost1tWeight = Component Cost Bonds .05 .55 .0275 Preferred Stock .10 .10 .0100 Common Stock .1785 .35 .0625 .1000 or 10% 68 Figure 9 (cont'd) Revenues Cost of service expenses (excluding taxes) Taxes Operating income Interest Preferred dividends Income available before allowance Allowance for funds Net income available to common stockholders Earnings per share 2 1 A 0,000 0,000 3,900 6,100 2,220 400 3,500 B 20,000 10,000 3,625 6,375 2,750 500 3,125 3,125 1.47 20,000 10,000 3,625 6,375 2,750 500 3,125 1,000 4,125 1.97 D 20,000 10,000 else 6,375 2,750 500 3,125 800 3&925 1.87 The decrease in EPS under Case C was due to the increase in the cost of debt and preferred stock compared to previous years. EPS under Case C would equal Case A. If these capital costs were to remain the same the *It should be noted that for purposes of illustration the marginal cost of capital is used here. that this method is not employed by utilities. The author realizes Utilities use what is more properly called an average or embedded cost of capital. The embedded cost of capital employs the use of previous debt issues in the calculation of the cost of capital. Again it is beyond the sc0pe of this study to decide the correct method of calculation. ‘ 69 the nuclear power plant. Witness the trouble the Con- sumers Power Company of Michigan has had with its Midland nuclear plant. Even if the plant is completed, there is no assurance that it will operate for the intended time period for which it has been designed. Therefore, a false sense of security may be given to investors by showing earnings as if there was no construction program and no attendant difficulties with the future operation of that power plant. Second, the argument does not result in the correct statement of earnings per share unless the allowance rate is equal to the marginal Cost of capital. As discussed above, this is a false assumption. Therefore, the separa- tion argument is not valid on the assumption that it pro- perly shields net income from the construction activities of that period. This procedure does not result in the separation, which is claimed by its followers unless the allowance rate is equivalent to the marginal cost of capital. 2.1.3 Valid Asset Argument. This argument is based on the concepts developed in the accounting literature. "Assets represent service potentials or rights to pro- spective benefits."6 When the allowance for funds is created and added to the asset account, it becomes in- distinguishable from previous amounts in this asset account. Therefore, this amount attributed to AFUDC gives rise to future benefits in the form of revenues the 70 utility will earn, as a function of the rate of return allowed on the expanded rate base. Hence, the allowance for funds is a valid asset. Litke states: Income generated from the capitalization of an allowance for funds used during construction is valid income because it will become part of the assets of the utility on which they will be entitled to earn a return and which will be recovered through depreciation expense in their cost of service over the service life of the property. Accordingly, the allowance capital- ized represents a valid asset and the derivative is equally valid.7 David Kosh states: Any investment (capital costs of construction) made in a utility or for that matter in any other business must wait for its return in the future. The only difference is that as concerns AFUDC there is a short lag before any returns can be anticipated; the lag being the duration of the construction period.8 There is no major criticism of this argument. In fact, it is the only valid defense for AFUDC. But, there are some minor points of clarification. First, the time lag for AFUDC is not a short period of time. It frequently can be ten years or longer. The importance of this point was examined further in Chapter One when the present values of the AFUDC cash flows are developed. Second, the argu- ment makes no mention of the validity of this income stream, which will occur in future periods, in comparison with the validity of an income stream, which is derived in the cur- rent period, that is a result of a rate of return which is allowed on the rate base. This specific tOpic will be discussed at length in Chapter One, where the differences between method I and method II and their respective income 7l streams are presented. Briefly, much of the literature on this particular topic contends there is a significant difference between these two income streams. Therefore, even though the valid asset argument appears to have some theoretical support, it suffers from the fact that no distinction is made between the two income streams derived from the different regulatory treatments of AFUDC. 2.1.4 Additional Defenses. There are two minor de- fenses in addition to the three major arguments previously presented, which can be found in the literature. The first is simply the fact that this practice has been well established for many years and it is in widespread use. One author goes so far as to use the Taiwan Power Company, of the Peoples Republic of China, as an example of its worldwide impact. The author also notes that a few in- dustrial corporations also use this practice.9 In other words everybody uses it so it must be correct. The second defense relies on the fact that stable earnings should result from this practice even during a construction 10 Stable earnings is a goal that many corporations, period. including those in the competitive environment, strive to achieve. In the utility regulatory climate, stable earnings also have a well developed following. The author does not intend to imply that this idea per se is wrong. Rather, the implication is that, manipulation through use of this questionable accounting practice might be viewed with a somewhat more jaundiced eye by members of the 72 investment community. The first defense clearly appears without serious merit. The defense offered by the stable earnings concept also appears to be somewhat questionable, but perhaps defendable given the regulatory climate towards risk. 2.2 Criticisms There are three major criticisms of the allowance for funds, the next section will discuss each one. 2.2.1 Manipulation. A serious concern to users of financial statements is the possibility that some of the numbers may be manipulated by management. This could occur simply by changing the allowance rate or the con- struction work in progress account to which it is applied.11 Proponents contend this is not a valid point because allowance practices are rigorously examined by the regula- 12 Nevertheless, the allowance rate does tory commissions. change quite often. It varies among firms within any one year and it varies from year to year for any one firm. There is not just one model that all commissions use to calculate this allowance rate and the scrutiny of this practice does vary from commission to commission. It appears that manipulation whether intentionally or unintentionally, as a direct result of the regulatory climate, does occur. For example, a $500,000,000 con- struction program and a change in the allowance rate from 8% to 8%% adds $2,500,000 to earnings. This is a consider- able sum of money that might create some skepticism in 73 the investment community. This is especially true, since as already noted, there is very little theoretical explana- tion for either 8% or 8%%. Investors might be expected to view this practice with some concern, especially since the allowance rate changes quite often. 2.2.2 Quality of Earnings. This is the most serious and frequent criticism encountered in the literature. Chapter One is devoted to a discussion of this issue. Therefore, the discussion here will be brief. Statements along the following lines are common in the literature on this particular area. Today's utility income accounts for the most part are full of what have come to be known as "gimmicks". . .One of the most serious of the gimmicks results from including a credit for interest charges to construction in cur- rent earnings.13 Serious question arises as to whether the reported figures really reflect even current, or latest reported, earnings,. . .This results from the fact that earnings as reported in- clude for many, if not most, companies a large bookkeeping item formerly called, "interest during construction.14 The peculiar regulatory accounting concept that interest accruing against a utility during a construction period is an asset rather than an expense, and that the offsetting credit somehow produces income.15 There has also been an empirical study where investor attitudes towards the allowance were examined. The con- clusion was that allowance earnings are inferior to other earnings. Mitchell and Hutchins surveyed "110 of the largest institutions in the U.S.," they conclude "no one among informed investors believes that AFUDC earnings are 74 the same quality as operating earnings."16 Again it should be pointed out here that no distinction between method I and method II was discussed when allowance earnings were judged to be inferior to other utility earnings. There are major differences between these two methods in terms of the quality of the earnings that are derived. The con- cept of quality earnings was discussed previously and Chapter One contains an extended development of this con- cept. 2.2.3 Unstable Earnings. This criticism is tenuous since it is the counterpoint to stable earnings as a defense. Both the criticism and the defense implies that the income statement should be shielded from a significant economic event of the period. Clearly this is illogical given the purpose of accounting principles. But to give some idea of the concern for this topic in the utility industry, chairman George I. Bloom, of the Pennsylvania Public Utility Commission makes this point. The bad thing about the IDC contribution to net is that it is unstable. During the course of construction of a new generating station, the successive quarterly reports of a utility show a large and rapid build-up of the IDC credit on the income statement. Then suddenly, on the day the station is completed and goes into operation, the supply of IDC credits ceases and, barring some other offsetting factor, such as another generating station in the early stages of construction, the re- ported net per share shows a decline. A declining net per share tends to lower the price at which the utility can market new common, and this lower price increases the dilution of earnings for the existing shares. Since utility managements are elected 75 principally with the votes of the existing common shareholders, they are naturally reluctant to sell new common, and tend to do as much financing as possible through sale of senior securities.l7 Looking back at Figure 9, critics argue that when the plant is placed in service in Case C or Case D, the plant is unlikely to produce utility income sufficient to offset the loss of $1,000 or $800, respectively, that will no longer be present. Defenders of AFUDC contend this point is irrelevant. Companies A and C will have the same earnings next year. This is exactly what the allowance is designed to do, separate the operations of the firm from the effects of large construction programs. As pointed out earlier, this is illogical given the pur- pose of generally accepted accounting principles. 2.3 Comparison of AFUDC With Other Income AFUDC can now be compared to other utility income such as depreciation and deferred taxes. The valid asset argument focuses on the determination of future cash flows. It is then necessary to compare the cash flows associated with AFUDC and the cash flows associated with depreciation and deferred taxes to see if they are the same or different. If they are dissimilar then of course another criticism would have justification. Any income figure can be con- verted to cash provided by operations with this simple equation: cash flow provided by operations = net income reported : noncash items recognized in computing net in- come. There are many noncash items recognized in computing 76 utility income. However, if the term cash flow is ex- panded to near cash flow (to include receivables and pay- ables) the principal adjustments are for depreciation and deferred taxes. Then for all practical purposes the equa- tion becomes: cash flow provided by operations = utility income reported + depreciation + deferred taxes. 2.3.1 Depreciation. In a public utility the depre- ciation charges are considered a cost of service expense whereby the future rate base is reduced and therefore the future cash flows will be reduced. But, the cash flows which result in this period are exactly equal to the present value of the future cash flows which are foregone by taking the depreciation in this period. For example, let D be the current depreciation expense. If depreciation is not taken this period the rate base would not be reduced by an amount equal to D. Furthermore, if D remains in the rate base until some future period N, D will produce return in each period of r where r is the rate of return allowed by the commission. The future value at any period N of D and the returns on D is then, D (1 + r)N. Discounting this to the present we simply divide by (l + r)N which leaves us with D. Therefore, the income reported in the period and the income actually earned will be the same. There should be no differences in determining the value of utility income since the cash flows are equal to recorded depreciation. 77 The same formulation can be given to AFUDC up to a certain point. A number of people argue that AFUDC and depreciation are the same in terms of the cash flows actually earned by the firm. In Chapter One a detailed discussion of AFUDC earnings is given, with the relation of AFUDC earnings to cash flow. A brief overview will give some idea of the basic difference between depreciation and AFUDC earnings. The amount of the allowance that is recorded in the plant accounts at the time the plant is placed in service will produce future cash flows. Let Al represent the amount of the allowance in the plant accounts at this time. Then the future value of Al is Al (1 + r)N. Discounting this to the beginning time period when the plant is first placed in service, we divide by (l + r)N which leaves us with Al'- Up to this point depreciation and AFUDC are the same, but here a crucial divergence occurs. The depre- ciation D is recorded in the period and the amount D is recovered in the same period through "cost of services expenses." But in the case of AFUDC remember the AFUDC for this plant is recorded as income long before the plant is actually placed in service. Therefore, Al must still be discounted by (l + r)n where n represents the length of the construction period. Referring to Figures 7 and 8 in Chapter One and the appropriate discussion, one can see that there is a substantial difference between the income reported as AFUDC and the cash flows associated 78 with that income. It would seem that depreciation and AFUDC are similar, but that AFUDC must be discounted by this additional period of time. Thus, AFUDC and depre- ciation do not contribute to the earnings and cash flows of an electric utility in the same manner. 2.3.2 DeferredTaxes.18 This discussion centers on deferred taxes as a component of cash flow for an electric utility. It does not focus on the relationship between AFUDC and deferred taxes. The relationship between AFUDC and deferred taxes is treated in Chapter One. There are two procedures followed by utilities in the treatment of deferred taxes. A number of commissions requires that 19 In this the credit be deducted from the rate base. case the present value of the future benefit foregone is the current cash received. In algebraic terms, let T represent current deferred taxes, r the rate of return allowed on the rate base and P the future tax payment to be made. If T is not recognized now, then the rate base will be higher by this amount until some period n when P is made. The future value of not recognizing T is the return foregone each year (T x r) plus P in year n. Assuming the tax rate is constant, P equals T, and the future value is then T (1 + r)N. Discounting this back to the present (dividing by (1 + r)N we get just T. The second procedure requires that the deferred tax 20 credit be treated as a cost-free source of capital. An illustration follows, assume the following definitions: 79 T = amount of the deferred tax credit R = rate base under the first procedure R + T = rate base under the second procedure r = rate of return allowed under the first pro- cedure R (first) I rate of return under the second procedure Under this second procedure you can see the return foregone is the lower return represented by (§%T)r, (this is less than Rr) times the higher rate base represented by (R+T). The rate base is higher because of the absence of the deduction for the tax payment made in period n. The stream of payments in each future year is given by: (§%5)r x (R+T) = Rr Rr is exactly the same stream of payments which accrues to the firm as in the first procedure above. Therefore, the two procedures offer identical treatments of the future returns to utilities. The determination of cash flows which occurs in any year because the firm recognizes depreciation and deferred taxes is different from AFUDC in the determination of other utility income. The cash flows from depreciation and deferred taxes does not increase or decrease the value of the firm beyond the income reported because the cash flows received are at the expense of foregoing future cash flows which have a present value equal to the current benefit 80 received. In other words, the income reported is equal to the present value of the inflows of cash from both depreciation and deferred taxes. As discussed in previous paragraphs this is not the case for AFUDC. The cash in- flow from AFUDC is less than the income reported. There- fore, in the determination of cash flow accruing to a utility, depreciation and deferred taxes are quite dif- ferent from AFUDC. 2.4 Summary This chapter presents the principal criticisms and defenses of AFUDC. The defenses were shown to be somewhat lacking inthe theoretical support. The only defense with any merit is the valid asset argument, but this was also shown to be questionable since no distinction is made between method I and method II income streams. The criticisms of AFUDC center on the quality, stability and manipulation of these earnings. The investment community's reservations regarding these earnings was also presented. Finally, a short discussion of two components of other utility income, depreciation and deferred taxes, and their similarities and differences to AFUDC was developed. 81 Footnotes 1Eldon Hendriksen, Accounting Theory (Richard D. Irwin, Inc.), Homewood, Illinois, Third Edition, 1977. 2Everett L. Morris, "Construction of Interest on Construction: Time for Reappraisal," Public Utilities Fortnightly (March 4, 1971), p. 23. 3Robert E. Frazer and Richard C. Ranson, "Is Interest During Construction 'Funny Money'?" Public Utilities Fort- nightly (December 21, 1972), p. 21. 4Frazer and Ranson, "Funny Money," p. 20. 5This example is similar to the illustration in Robert Frazer and Richard Ranson, "Is Interest During Con- struction Funny Money?" Public Utilities Fortnightly (December 21, 1972), p. 21. Also Johnny R. Johnson dis- sertation. 6 Eldon Hendriksen, p. 257. 7Arthur L. Litke, "Allowance for Funds Used During Construction," Public Utilities Fortnightly (September 28’ 1972)! p- 20. 8David Kosh, "Interest During Construction Renamed Allowance for Funds Used During Construction," Regulatory Commission Development Short Course, Michigan State Uni- versity,'l974. 9Frazer and Ranson, p. 23. 10Pomerantz and Suelflow, p. 119. 11Morris, "Time for Reappraisal," p. 21. lzIbid. 13‘ Frazer and Ranson, Ibid., p. 21. 14Charles Tatham, "Interest During Construction and Price-Earnings Ratios," p. 33. 15George I. Bloom, Chairman of the Pennsylvania Public Utility Commission, speech before New York Society of Security Analysts or quoted in Public Utilities Fortnightly, September 28, 1972, p. 37. 82 16Charles A. Benore, "Accounting Procedures and Stan- dards Related to Capital Formation in the Electric Power Industry and the Differences in Earnings Quality Among Major Electric Power Companies," paper presented to the NARUC staff Subcommittee at Washington, D.C. on September 10, 1975. 17George I. Bloom, Interest During Construction Evaluated," Public Utilities Fortnightly (September 28, 1972): p. 37. 18The algebraic formulation of deferred taxes and depreciation was developed in Johnny R. Johnson disser- tation. 19National Association of Regulatory Utility Commis- sioners, 1974, Annual Report on Utility and Carrie Regu- lation (Washington: National Association of Regulatory Utility Commissions, 1976), p. 396. 20 Ibid. CHAPTER 3 LITERATURE REVIEW This chapter will review the financial literature pertaining to empirical tests of AFUDC. Previous empirical tests have focused on differences in total market valuation. In this context, differences in market valuation have been empirically tested, using three widely followed financial ratios; price earnings ratio, market-to-book ratio, and intereSt coverage ratio. This chapter will discuss the effect AFUDC might have on each of these ratios and the empirical work in each area. Additional empirical work on AFUDC will also be reviewed. 3.1 Price Earnings Ratio The first area of interest is empirical tests using price-earnings ratios (hereafter P/E). P/E ratios are widely published and followed by investors as some indi- cation of relative earnings quality and future growth prospects for earnings. From the security analyst's point of view, serious questions have been raised about the validity of a utility's earnings when they contain a large component attributed to AFUDC. They view this as 83 84 having a deleterious effect on P/E ratios. Common stock valuation is a complicated and controversial topic. For purposes of this discussion we will assume most financial analysts use some method of capitalization of earnings for common stock valuation. This entails some problems; earnings must first be adequately defined and measured. The security analyst is more concerned with earnings that represent cash flow, therefore he would be more con- cerned with the so called "revenues minus expenses concept." Irving Fisher eloquently outlines this position. He points out that "no concept of income, which includes appreciation or depreciation in capital value as a positive or negative item of income, is acceptable as a basis of valuation under the "capitalized income“ method. The only concept, which is strictly valid for this purpose is the "flow-of-services" concept, of which the "cash receipts- minus-cash-disbursements" concept is a modified version."1 "Under such a definition of income, no account is taken of the increase or decrease in the capital value of the property from which the income is derived. Hence no vicious-circle fallacy is involved in the assumption that the present value of the property is the discounted value of the anticipated flow of services. On the other hand, the increased net-worth concept of income is clearly 2 disqualified." ."In other words, if income is used as the basis of capital value, we cannot include in income 85 an amount which itself represents an enhancement in capital value. This would obviously be double counting."3 The previous discussion implies AFUDC does not meet the revenues minus expenses concept of earnings. There- fore, AFUDC would be expected to have some effect on the P/E ratios of electric utilities. For example, using Standard and Poor's 40 utilities for comparison purposes only and the average AFUDC percentage calculated from the compustat tapes in 1976 gives some idea of the dis- crepancy that might occur. When AFUDC is included in earnings the P/E ratio is 6.44, but if AFUDC is substracted from earnings the P/E ratio becomes 8.25. Which multiple is correct? It depends on who is calculating the ratio and their attitude towards AFUDC. A problem arises in that the P/E ratio could change sub- stantially if the amount of AFUDC changed, even though earnings from operation remain unchanged. The historical pattern of P/E ratios and future estimates of the P/E ratio would be distorted because AFUDC can change drasti- cally from year to year, depending on a number of both internal and external factors. A number of researchers (Severiens, Stich, Tatham) have examined the effect of AFUDC on P/E ratios. Severiens further develops the concept of why P/E ratios might be effected by AFUDC.4 The growth in earnings is an important variable that investors attempt to assess in determining stock prices. Malkiel and Cragg's study noted that 86 expectations of changes in the normal trend of growth in earnings per share have been the principal reason for price-earnings differences during most of the postwar period.5 Perhaps a more important variable in electric utilities is increases in dividends. Due to regulation and the reputation of quality stocks as income stocks with high dividend yields, the dividend factor should be of serious concern to investors. Since AFUDC is a non- cash item, which is added to income, it does not enhance the ability of utilities to pay increased dividends in the immediate future. Another factor is that the growth rates in earnings can also be irratic due to the uncertain nature of AFUDC earnings. These two arguments are presented to provide some theoretical basis for why increasing amounts of AFUDC might have a negative association with price- earnings multiples. Severiens studied 12 electric utilities using linear regressions, where the change in P/E ratio is regressed against a number of factors. He found the most significant variable to be AFUDC, both in individual firm regression and in the combined regression of all firms. He states, ”the sign is negative suggesting that the market discounts its effect on earnings."6 Stated differently, the market valuation of a firm as represented by the P/E ratio declines with increasing amounts of AFUDC. Tatham and Stich both present theoretical arguments similar to the above argument. Tatham makes the point that 87 since the portion of AFUDC as a percentage of income has increased substantially from 1965 to 1975, the P/E ratios based on reported earnings are not the same. "We can hardly use historic averages as a guide to what an appro- priate price-times-reported earnings should be today."7 If the utility analyst chooses to capitalize reported earnings, with their large and fluctuating AFUDC component, he has no historic guide as to what price-earnings multiple might be appropriate. But, the analyst also might be very hesitant to conclude that the proper earnings figure to use would entirely exclude AFUDC. AFUDC is a current recording of future earnings, which should be recouped, at least to a certain extent at a later date. Therefore, when using P/E ratios as guides to stock values the portion of AFUDC should have an important impact on this ratio. The implications, from the work of Severiens, Tatham and Stich, are that as the percentage of AFUDC increased the P/E ratios of electric utilities declined. Johnson in an unpublished dissertation did a more comprehensive study of AFUDC and P/E ratios. Unlike the previous studies, Johnson focused on the differences in P/E ratios between method I and method II firms. As stated previously, these two methods are two different regulatory treatments of AFUDC. Beaver and Morse attri- bute at least some of the differences in P/E ratios to 8 be different accounting methods. A serious problem arises in that it would prove difficult to isolate and 88 quantify the effect AFUDC might have on the P/E ratios. A number of confounding variables could effect the statistical nature of any test done on P/E ratios. Con- founding variables are variables that might affect the test, in this case, differences in P/E ratios, but aren't under control in the design. Numerous factors effect the market valuation of a firm as represented by its P/E ratio other than the amount of AFUDC. This is not to cast doubt on the previous arguments on why P/E ratios might be effected by AFUDC, but to suggest that better statistical techniques might be used to ascertain the real influence of AFUDC.. Nevertheless, Johnson examined this question of AFUDC as an accounting difference and the effect on P/E ratios for a sample of 45 method I and 9 method II com- panies for the period 1965-1974. His hypothesis was concerned with the earnings/price ratio which is the inverse of the P/E ratio. Johnson hypothesized that method II companies would not be effected by AFUDC since CWIP is included in the rate base and the value of the firm due to the allowance is equal to the amount of the allowance income that is reported. In this case the accounting entry accurately reflects the cash flow the firm receives. Method I firms do not include the CWIP in the rate base, consequently any AFUDC recognized as income has no immediate cash flow implications. It is only an accounting entry, the firm does not receive any 89 cash from the transaction. Therefore, the market should react to these two methods in different ways. The Wilcoxon Rank Sum Test, which is a non-parametric test, was used. His conclusion was that the average price earnings ratio for method I firms was lower than the price earnings ratio for method II firms. The statistical significance or differences between the P/E ratios of method I and method II firms was found at the .0749 level.9 Since this is one of the very few empirical tests performed on the differences between method I and method II firms, a detailed discussion of the non-parametric tests and a replication of Johnson's tests on P/E ratio and market- to-book ratios with more current data is provided in an appendix to Chapter Three. 3.2 Market-to-Book Ratio Another ratio of particular interest to the utility industry is the market-to-book ratio. Both the P/E ratio and the market-to-book ratio are surrogates of investment risk to the common equity. These ratios have suffered a marked decline as compared to the industrial averages in the last fifteen years. The downward trend in the market- 10 The table shows to—book ratio is outlined in Table 3. that the ratio for utilities is not much different from the industrials until 1966. After this date the utility ratios suffer a continuous decline up through 1974. Whereas, in 1966 utility stocks were priced comparably 90 Table 3 A Comparison of Market Value to Book Value Ratios Between Industrials and Utilities for the Years 1960-1974 Moody's 125 Moody's 24 ngg Industrials Utilities 1960 i 1.99 1.77 1961 2.21 2.20 1962 2.03 2.11 1963 2.22 2.27 1964 2.48 2.28 1965 2.55 2.35 1966 2.25 2.00 1967 2.33 1.90 1968 2.41 V 1.74 1969 2.24 1.60 1970 ' 1.85 1.27 1971 2.09 1.29 1972 2.26 1.17 1973 2.07 1.00 1974 1.42 .67 Source: Robert S. Stich, "An Additional Standard for Measuring Common Equity Costs," Public Utilities Fortnightly, February 26, 1976, p. 36. ‘ 91 to industrials, in 1974 they sold for less than half on a book value basis. Obviously, this is of serious con? cern to public utility officials and investors since any new common stock that is sold must be sold below book value. This might lead to dilution of the common equity and therefore, the determent of current stockholders. An examination of Table 4, which contains the P/E ratios through 1977, reinforces this observation of a declining trend in the market valuation of electric utilities. The P/E ratios and the market-to-book ratios for utilities were comparable to the industrials until 1966, at which time they started on a downward slide which continued through the 1970's. The market-to-book ratio can be likened to the P/E ratio, in that it is also thought of as some measure of investor confidence in the utility's future earnings capabilities. It is also argued that not just earnings, 11 As but "real" earnings are the key to this ratio. already stated, this ratio has suffered a serious decline since 1965. In one study, 80 electric utilities were divided into deciles by market-to-book ratios in 1965 and then compared in 1974. This ratio declined from 192.0% to 68.8% for the lst decile and from 377.3% to 106.4% for the 10th decile. Clearly the market has reacted to declining real earnings of utilities with in- creased skepticism, to the point where the majority of utilities are selling at prices below book value. This 92 Table 4 A Comparison of P/E Ratios Between Industrials and Utilities for the Years 1960-1977 400 130 Electric X233 Industrials Utilities 1960 18.14 18.6 1961 22.47 23.4 1962 17.05 20.3 1963 18.69 20.3 1964 18.55 21.4 1965 17.87 19.5 1966 14.47 16.7 1967 18.57 14.6 1968 18.38 15.7 1969 16.45 12.1 1970 18.58 12.8 1971 18.72 12.2 1972 19.31 11.6 1973 12.32 8.9 1974 7.89 6.7 1975 11.80 8.4 1976 11.19 8.9 1977 9.01 8.4 Source: Computed by the author from Compustat data and Standard and Poors Statistical References Guide. 93 does not bode well for the electric utilities' future ability to raise capital. Electric utilities will need to raise massive amounts of capital for large construction programs in the future. As one author states, "it will be difficult enough for the healthiest of firms; trying to compete for this limited resource requires that utilities regain the interest of the investment community, and the only way they can do this is by improving their earnings."13 There have been a number of reasons given for the decline in this ratio. The principal factors are infla- tion, regulatory lag, and as a number of authors suggest, the increased use of AFUDC (lower quality earnings). The focus of this argument is the concept of real earnings and future earnings prospects. AFUDC as already explained can effect the earnings of the utility. It is not sur- prising to see AFUDC come under scrutiny as the cause of the decline in the market-to-book ratio. The market-to-book ratio was examined by a number of researchers, (Burkhardt and Viren, Hyman, Marx, Fitzpatrick and Statzel, Johnson, among others). Con- flicting statistical conclusions are the product of this research. This could be due to differences in the statistical design and multicollinearity among the vari- ables tested in that specific design. The following is a brief review of each of these tests. 94 Hyman first examined 24 of the largest electric utilities in the period 1967-72 using multiple regres- sion.14 A regression with total return as the dependent variable and a number of variables including return on equity, change in return on equity, payout ratio, a quality index represented by the percentage amount of AFUDC, growth in equity and market-to-book ratio, among others as the independent variables were used to examine what might cause differences in total return. He con- cludes, "Quality of earnings does seem to have some. effect on the market action of the stocks. . . .Perhaps, then, there will have to be some de-emphasis on earnings per share, and more attention paid to how these earnings are generated if they are to benefit the stockholder."15 In a similar study, the dependent variable was the market- to-book ratio and the independent variables included the payout ratio, percentage of AFUDC, coverage ratio and return on equity. Hyman16 found the return on equity to be the most statistically significant of the variables tested, with the coverage ratio also being significant. He concluded that the effect of AFUDC was already in- cluded in the coverage ratio. The implications of these studies are tentative at best due to the small sample size, limited time period and multicollinearity among the variables. Multicollinearity refers to the fact that one or more independent variables in the regression 95 equation are correlated with each other. Any statistical significance attached to any of these variables may then be misstated. 17 studied 95 electric utilities Burkhardt and Viren but only used 1976 as the test period. They examined a large number of independent variables and the effect on the market-to-book value. Among the independent variables were earnings growth in EPS, bond rating, regulatory ranking, cash dividend payment and AFUDC. They found that AFUDC was not significant, but growth rate in EPS, the rating of the company's bonds and the coverage of the firm's cash dividend payment was significant. AFUDC would affect all three of the variables here and due to multicollinearity any conclusions about AFUDC are suspect. 18 used data from 1969-1975 Fitzpatrick and Stitzel and the 90 utilities on the Value Line data base. The independent variables included AFUDC, risk measures (Beta), profitability ratios (for example return on equity) and cash flow measures. They found AFUDC statistically significant only for years 1972-75. "These results suggest that as AFUDC as a percent of net income in- creases, more and more concerned attention is focused on the AFUDC issue by the investment community. This concern is evidently being reflected in a measurable 19 These results reduction in market-to-book valuations." suffer from the same statistical problems, as mentioned above. 96 Finally, Johnson examined the differences in market- to-book ratio for method I and method II firms. He ranked companies according to their market price per share compared to book value per share. They by using the Wilcoxon Rank Sum test he was able to test whether the ranking of firms according to market-to-book ratio was actually different for method I and method II firms. Johnson found the method I firms' market-to-book ratio was statistically different from method II firms at the significance level of .1230.20 It appears then, that there is no conclusive evidence to support the hypothesis that AFUDC is a different quality earnings and therefore this should be reflected in the market-to-book ratio. This is due as much to the problems of statistical design as to the results of the studies mentioned here. Johnson's study appears to be the best designed study since the others suffer from problems al- ready mentioned. Nevertheless, Johnson's study is also deficient in a number of areas that the research by this author will try to eliminate; (l) the test of differences in the market-to-book ratio was significant at only the .1230 level. This might be economically insignificant at such a high level of significance. The implication here is that even though there is statistical significance the dif- ferences in economic values upon which this study is based may be so minor as to render any economic policy decisions 97 useless. In fact, looking at the original data of market- to-book ratios, one finds it takes very little change in the ratio to substantially change the ranking. For example, a change in the ratio from 1.31 to 1.41 changed the ranking from 13 to 21. Since there was only 9 firms in the method II grouping, but 45 firms in the method I grouping, any slight change in the ratio due to some other factor besides AFUDC could bias the results. (2) The time period for the study ended in 1974. Even though the upward trend in AFUDC started in 1966, AFUDC accounted for only 13.3% of net income on average as late as 1970 (Table 1). The large jumps in the period 1972-1974 might have effected the market differently and therefore the statistical nature of the tests. (3) Dividing the companies into method I and method II would add a substantial risk of confounding variables due to the regulatory climate. If the commissions approved method II, where CWIP is included in the rate base, perhaps they might be more lenient in other matters such as rate increases. The question then arises whether the market reacted differently to the method of accounting for AFUDC or to differences in regulatory climate. Surely these two aspects must be strongly cor- related. This is then a crucial problem which will be eliminated in the research undertaken here, by equating the two groups of firms according to their relative market risk as reflected in their Betas, which should account for any perceived differences in regulatory climate. Any 98 of the above problems may have biased Johnson's results, therefore, the results while not inconclusive, may be suspect. 3.3 Interest_§overage Ratio The last financial ratio which has been hypothesized as being affected by AFUDC is the interest coverage ratio. Here AFUDC poses an obvious problem since AFUDC is in theory an offset to interest expense. Prior to 1970 AFUDC was shown as a deduction from interest expense, then for a time it appeared under the section "Other Income" but now it is divided between both these sections of the income statement. The component attributed to debt funds is shown as an offset to interest and the component attributed to equity funds is shown as other income. "To the security analyst this is an interesting aspect of the matter, for he finds that, by shuffling the items on the income account a portion of the interest deduction is turned into income and used to demonstrate its own margin of protection. It is conceivable that a bond buyer might view this a bit darkly."21 The fixed charge coverage ratio represents the relationship between earnings available to pay interest charges and the total amount of fixed interest charges. AFUDC can effect this ratio depending on whether the interest charges are reduced (denominator) or the earnings are reduced (numerator). There seems to be no consistent 99 or uniform method for calculating the coverage ratios for utilities. Even though there is one appropriate method for SEC reporting requirements, the conclusions analysts draw from these numbers may be quire different. "Some analysts completely exclude the AFUDC credit; others include it as a part of the earnings available for interest; a third group deducts it from the total 22 This would create substantial dif- interest change." ferences in the ratio calculated. In the case of the Consumers Power Company, for example, total interest charges for 1977 amounted to $131,388,000; total AFUDC was 54,211,000 and earnings before interest and taxes was $263,733,000. The ratio could be 2.00 times if AFUDC was included in earnings or 1.59 times if not in- cluded and finally 3.41 times if AFUDC was deducted from total interest charges. These are substantial differences which may affect the utilities credit rating and bond ratings. The fixed charge coverage ratio is not the only analytical tool used by the rating agencies in determining the quality of bonds for investment purposes but it is probably the 23 Therefore, the treatment of AFUDC can most important. effect the coverage ratio and the bond ratings for the utility. Donald W. Johns in research done for the Public Service Commission of Michigan examined the relationship 100 between AFUDC and bond ratings. He examined a sample of 55 utilities taken from the Compustat data base. His test hypothesis was that pretax interest coverage ratios adjusted for AFUDC will predict utility bonds ratings with less error than unadjusted pretax interest coverage ratios. His conclusion was that pretax coverage ratios with AFUDC removed was not statistically different from pretax coverage ratios including AFUDC when used to predict bond ratings. He further states, "no statis- tical evidence that investors or rating agencies discount AFUDC when determining the cost of debt on bond ratings, respectively, was found to exist in the information 24 This research is the strongest endorsement examined." given here for the conclusion that AFUDC has no effect on market values. 3.4 Other Empirical Tests of AFUDC The final review to be included in this chapter is another test of differences between method I firms and method II firms performed by Johnson in his dissertation.25 This test is the most statistically valid test of those reviewed in this chapter. It does provide some valid evidence for differences in total market valuation between these two groups of firms. This design is a two- step process. First, the percentage change in earnings is regressed against the percentage change in stock prices for each of the ten years in this study, 1965- 1974. These ten regressions were of the form: 101 P1 = oi + BiEi + ei where: Pi = the change in stock price for year i. Ei = the change in earnings in year 1. 8i = the residual or error in year 1. Johnson based his design on the premise that the residuals contain all the information not explained by the independent variable, for the observed variation in the dependent variable. Therefore, Johnson's hypothesis was that an allowance factor representing the percentage of AFUDC should be included in a regression model of Pi and Ei' He could test this hypothesis by going to the second step. The second step is a regression of the form: Si = ai + biAFUDCi + di where: 8i = the residuals from the previous regressions in each year i. AFUDCi = the change in the allowance factor for each year i. The slopes of this regression, represented by bi' then served as the basis for the statistical tests. Johnson hypothesized, that since method I AFUDC earnings are not the same quality as method II AFUDC earnings, a regression of the change in AFUDC percentage from year to year should have some relationship to residuals for method I firms. But there should exist no such relation- ship for method II firms. The slopes for both groups 102 of firms are combined and ranked from smallest to largest. Then the sum of the ranks for one of the groups is com- pared to the expected sum of the ranks for that group using the Wilcoxon Rank Sum test, a non-parametric test. Johnson concluded; there is a statistically significant difference between method I firms and method II firms at a significance level of .033. This aspect of Johnson's study does seem to provide some valid evidence for differences in market valuation between these two groups of firms. A crucial difference does exist between Johnson's work and the statistical tests that are performed in this study. Johnson con- centrated on the total market valuation for individual firms without regard for differences in market risk, which may have biased any regression, where the regression is a change in stock prices on changes in earnings. This research attempts to eliminate any differences in market risk and thereby examine any changes in market valuation which may be due to the disclosure of AFUDC earnings. In summary, whereas previous empirical work has looked at the total market valuation of firms at some point in time, this research concentrates on the changes in market valuation which may be due to the disclosure of AFUDC earnings. The previous statistical designs have suffered from confounding variables, that is, vari- ables that could effect the total market valuation of firms, but where these variables have not been accounted 103 for in the statistical design. This research makes every attempt to present an unbiased experimental design, where- by all factors effecting the firm are held constant except for the variable under study. This variable is the dis- closure of AFUDC earnings. The specific statistical de- sign which is used to accomplish these goals is developed in the next chapter. 104 Footnotes lIrving Fisher, "The Nature of Capital and Income," Augustus M. Kelly, New York, New York, 1965. 2J. C. Bonbright, The Valuation of PrOperty, New York: McGraw-Hill, 1937, Chapter 26. 3Charles Tatham, "Interest During Construction and Price-Earnings Ratios," Public Utilities Fortnightly (September 27, 1973), pp. 32-36. 4Jacobus T. Severiens, "The Price-Earnings' Behavior of Electric Utilities: Some Determinants," Public Utilities Fortnightly, December 7, 1978, pp. 27-32. 5B. G. Malkiel and J. G. Cragg, "Expectations and the Structure of Share Prices," American Economic Review, September, 1970. 6 Severiens, p. 31. 7Charles Tatham, "Interest During Construction and Price-Earnings Ratios," Public Utilities Fortnightly, September 27, 1973, pp. 32-36. 8William Beaver and Dale Morse, "What Determines Price-Earnings Ratios," Financial Analysts Journal (July-August, 1978), pp. 65-75. 9Johnny R. Johnson, "The Allowance for Funds Used During Construction: Market Reaction to Current Accounting Practice," unpublished dissertation, Virginia Polytechnic Institute and State University, 1976. 10Robert Stich, "An Additional Standard for Measuring Common Equity Costs," Public Utilities Fortnlghtl , February 26, 1976, pp. 34#39. 11Thomas G. Marx, "Market-to-Book Return on Equity Correlation," Public Utilities Fortnightly (December 4, 1975): PP. 28-31. 12 Ibid 0 13Robert Stich, p. 39. 14Leonard S.‘Hyman, "Market-to-Book Ratio: Statis- tical Confirmation or Aberration?" Public Utilities Fortnightly (December 19, 1974), p. 28. 105 lsLeonard s. Hyman, "Utility Stocks in 1967-72: A Rale of Woe," Public Utilities Fortnightly, February 28, 1974, pp. 23-28. 16Leonard S. Hyman, "Market-to-Book Ratio: Statis- tical Confirmation or Aberration?" Public Utilities Fortnightly (December 19, 1974). 17Daniel A. Burkhardt and Michael A. Viren, "Investor Criteria for Valuing Utility Common Stocks," Public Utilities Fortnightly (July 20, 1978), pp. 27-35. 18Dennis B. Fitzpatrick and Thomas E. Stitzel, "Capitalizing on Allowance for Funds Used During Con- struction: The Impact on Earnings Quality," Public Utilities Fortnightly (January 19, 1978), pp. 18-22. 19 Ibid., Pp. 21-22. 20Johnson, p. 89. 21George E. Phelps, "That Bothersome Construction Interest Allowance," Public Utilities Fortnightly (January 31, 1974), pp. 21-27. 22Pomerantz and Suelflow, Allowance for Funds Used During Construction, p. 138. 231bid., p. 137. 24Donald W. Johns, "Testimony Before Michigan Public Service Commission, Investigating the Effects of AFUDC on Bond Ratings and Interest Coverage." 25Johnson, dissertation, pp. 77-82._ CHAPTER 4 METHODOLOGY:' THEORETICAL DEVELOPMENT AND EMPIRICAL SUPPORT This chapter is divided into six sections. A brief summary of the theoretical concepts used here, that is; information content, market efficiency, the capital asset pricing model, and the diStribution of security returns comprises the first four sections. The fifth section will integrate these concepts, and discuss their applica- tion to the present study. The methodology that is used here, that is; the equivalence of relative risks and the testing of vector means, has been well developed in the literature. Consequently this study will draw heavily on a number of articles by Gonedes and Dopuch, and Gonedes, for notational guidance.1 The last section details the specific hypotheses to be tested. 4.1 Information Content As already discussed the purpose of this study is to determine the information content of AFUDC. In other 106 107 words, does AFUDC provide information that the capital markets use to assess the distribution of future security returns. An event x can be shown to have information content if the conditional distribution f(R/x) is not equivalent to the unconditional distribution f(r) for any R, where R is some random variable. Of course there must be some reason for believing that x and R are in some way related. If f(R/x) = f(R) then x is ignored in agent's assessment of R. Stated differently, x has no information content which is used to evaluate changes in R.2 Accounting variables are random variables that may reflect information about the distributions of future values of firm's securities. The equilibrium prices of a firm's ownership shares are dependent upon assessed dis- tribution functions of the dollar returns per share. These distribution functions are dependent on the available in- formation about the firms operating and financing deci- sions. This dependency provides the basis for the many available studies on the accounting items that may reflect new information pertinent to establishing a firm's equili- brium values.3 See, for example, Ball and Brown, Eskew and Wright, and Gonedes.4 In the context of this study, an information content can be inferred to an accounting event x by observing the behavior of some random variable R during a period of time when it is thought the accounting event might have some 108 impact on the random variable. The event x being con- sidered here is AFUDC disclosure. The random variable, R is the stock returns of electric utilities, or the firm's equilibrium value. 4.2 Market Efficiency To be able to assess the impact of accounting events and their disclosures some period must be identified with- in which any information effects could be expected to be apparent. Much of the evidence supporting market effi- ciency suggests that stock prices adjust instantaneously to new information once it becomes publicly available. Therefore, one potential time period is the time period immediately following the disclosure of the accounting event, which in this case is AFUDC earnings. Previous work by a number of researchers has also implied that the information content of some economic events that are reflected in accounting numbers may ap- pear in stock prices a number of months before actual public disclosure.5 This could certainly be true of AFUDC. Disclosure of the specific practices concerning AFUDC to the general public is scarce. In fact, the classification of firms into either method I or method II does not exist in the company's annual reports or 10-K reports. The amount of AFUDC is disclosed quarterly and annually. But informa- tion as to the practices used to calculate AFUDC is limited to two or three sentences in a footnote to the financial Sta 0th cer of Se; The: Ca: do an. 50 di r8 109 statements. Security analysts could be expected to obtain other company documents such as filings with the SEC con- cerning the issuance of new securities, but an examination of these documents reveals basically the same information as in the annual reports. Nevertheless, security analysts who follow utilities could be expected to have a better understanding of what AFUDC represents than the general public. But as the previous chapters have indicated, there is substantial disagreement even among these pro- fessionals. The security analyst might also be expected to have a forecast of AFUDC in relation to operating earnings sometime before the annual report is released. But it is doubtful whether all these forecasts are correct. If the forecasts are not correct then there is some justification for examining the annual reports and the disclosure of AFUDC for information content. In fact, the most widely followed investment advisory services such as Value Line and Standard and Poors do not mention AFUDC when discussing earnings or earnings fore- casts for utilities. Therefore, since forecasts of AFUDC do exist, and because there are a number of security analysts who follow utilities, there should be some market anticipation to the disclosure of AFUDC earnings. But since forecasts are not always correct and since there is disagreement among analysts, there may also be some market reaction. Whether the anticipation is more important any reactio of this the per time pe reports An is now capital tween 5 time, 5 Ship be have st Of marl Either market infOrme infOrme CannOt 80mg a: effeCt: and Cii: 4'3 c. I: I One world . the Dr, 110 reaction to AFUDC earnings will be examined in the context of this research. The time period for this study is then the period immediately surrounding AFUDC disclosure. The time period was chosen so that any reaction to quarterly reports can also be tested. An important point about tests of market efficiency is now discussed. A number of studies have relied upon capital market efficiency to measure the relationship be- tween stock prices and accounting events. At the same time, studies, which are designed to measure the relation; ship between accounting information and market prices, have stated that their evidence supports the hypothesis of market efficiency.6 This is circular reasoning. Either the research design makes some assumption about market efficiency and tests for the effects of accounting information or it makes some assumptions about accounting information effects and tests for efficiency. It clearly cannot be both. Therefore, the present study will make some assumption of market efficiency and test for the effects of accounting information, which is the magnitude and disclosure of AFUDC earnings. 4.3 Capital Asset Pricing In order to infer information content to some event x, one must hold all other things constant. In a real world setting this is impossible to achieve, but by using the properties of the two-parameter asses pricing model one can overcome this problem of holding all other things lll constant, insofar as tests on vectors of mean security returns are concerned. By focusing attention on the values of relative risks, as defined by the two-parameter asset pricing model, one can distinguish among securities of firms with different production-investment and financ- ing decisions. The theoretical connection between rela- tive risk and firms' production-investment and financing decisions is developed in Fama and Miller and Hamada.7 Already stated is the fact that the distribution function of security returns is dependent on the firm's operating and financing decisions, which are reflected in accounting numbers. Therefore, if one assumes that all other competing factors of a firm's financing and produc- tion-investment decisions have been controlled through use of the two-parameter model and the equivalency of the relative risks, then a statement can be made about the differences between any two distributions and the event under study. That is, if two firm's, which are identical in every respect except for the event under investigation, have differences in the distribution function of security returns, then that difference can be attributed to the information content of the event under investigation. The capital asset pricing model combined with the efficiency of capital markets provides a well defined framework for assessing the information content of AFUDC. The model used here is the one proposed by Black and tested by Black, Jensen and Scholes and by Fama and MacBeth. 112 Jensen also provides a detailed summary of the various forms of the CAPM, the tests on each form, and the model's assumptions.8 These assumptions are_also used here. The two-parameter model implies that the equilibrium expected returns are given by: .E(Rt) = E error LOW %. “2 AFUDC ....... F - - - - _ _ - - _ - - _ - - - _ High % AFUDC “3 Low 3. EPS forecast /~ error h- - Low % ' AFUDC u4 132 These four portfolios are independent and will result in four sets of mean monthly return observations which can be tested in either of two ways. The multivariate test, Hotelling T2, can be used to test for the equivalency of the four means simultaneously. A significant T2 statistic implies at least one of the means is not equivalent to the other means. If this test is significant then individual contrasts can be used to test more specific questions using the simple t-test. In this case, it might be hypothesized that the means of the high forecast error portfolios are not equivalent to the means of the low forecast error port- folios. An additional question can also be examined, that is, is there some difference between the high and low AFUDC portfolios within either the high or low forecast error groups. This design allows us to use both the univariate and multivariate version of the t-test to examine mean re- turn differences in portfolios depending on the specific hypothesis that is to be tested. b) Rank the firms from the highest to lowest per- centage forecast error. Then split these firms into two separate groups by choosing every other firm for each group. This should provide as close a matching as pos- sible for this specific attribute. There are now twenty- one firms in each group. The groups should be as similar as possible, given that they are all method I firms and they have been matched on the earnings forecast error. 133 The next stage in the actual portfolio construction ' is to rank each firm in each of the two groups by the per- centage of AFUDC. The balance of this discussion is iden- tical to the previous design and to discuss it here would be redundant. The end result is that four portfolios are formed. .In this case though, any differences due to the forecast error has been controlled by matching the groups based on the EPS forecast error so that each group of twenty-one firms is nearly identical. This is not a multivariate test. It is two separate univariate tests, in which the simple t-test is used to test for differences in mean monthly returns for two port- folios. In one case the portfolios are high percentage AFUDC versus low percentage AFUDC and in the second case, it is low percentage AFUDC versus high percentage AFUDC. All firms are method I firms. A diagram may facilitate iunderstanding. 21 firms ‘Hethod I firms only _F_ . 10 f. 10 f 21 firms F 1rms irms '7 “‘ 1 . with hi with high . ’1 percent- percent- ‘ ‘ ge _ age AFUDC * split 10 rim—s" 0 a; th low ch low ercent- ercent- 89 AFUDC ge AFUDC . . .... ‘ __I L..— 1'21 _ - __ l'21 134 It is true that this design really tests the same hypothesis. That is; is there some difference between high percentage AFUDC portfolio and low percentage AFUDC portfolios. The fact that four distinct portfolios are formed does not change this hypothesis. It just allows one to test this hypothesis in two separate tests. 4.7.2 Changes in the Percent AFUDC This next section concerns changes in the percentage of income represented by AFUDC. The first stage is iden- tical to the previous section. The method I firms were ranked on EPS forecast error. Then by either of the two methods outlined previously, the total group of method I firms is split into two distinct groups. In one case the groups were formed to be as identical as possible based on the forecast error. In the other case, the groups were formed so that one group contained high fore- cast error firms and the other group contained low fore- cast error firms. The second stage is different. Here the firms were stratified into portfolios based on the change in percent- age of income represented by AFUDC, not just the percent- age AFUDC. For example, if the percentage of AFUDC in- creased from 10% to 20%, this is classified as a 100% change in AFUDC percentage. The premise in this case is that the percentage of income which is represented by AFUDC can be thought of as a quality index. It can be thought of as a quality index 135 because for method I firms the present value of the AFUDC earnings is less than the income reported. Therefore, the larger the percentage of income represented by AFUDC earn- ings, the lower the quality index of that income. In other words, the ratio of cash or near cash (near cash re- presents other utility income such as depreciation and de- ferred taxes, which is discussed in Chapter Two) to the total reported income is decreasing. Let the allowance factor represent the percentage of total income represented by AFUDC in any year. Then when one considers the year to year percentage change in earn- ings with the year to year percentage change in the allow- ance factor, four distinct possibilities arise:38 A) If both the allowance factor increases and in- come increases compared to the previous year, the actual increase_in income is less than the reported increase in income. This is true because the reported income has de- creased in quality. It includes a bigger component of AFUDC earnings. B) If the reported income increases and the allow- ance factor decreases, the actual increase in income is greater than the reported increase in income. In this case, reported income becomes higher quality because the percentage of earnings represented by AFUDC declines. C) If reported income decreases and the allowance factor increases, the actual decrease in income is greater than the reported decrease in income. This is true because 136 not only does reported income decrease, but it is also of lower quality since AFUDC represents a larger percentage of those earnings. D) If the reported income decreases and the allow- ance factor decreases, the decrease in actual (real) income is smaller than the decrease in reported income. This is true because even though reported income decreased, a smaller percentage was represented by AFUDC and therefore it is a higher quality. In algebraic form, a percentage change in income can be represented by I T - 1. T-l I Also recognize that to find the real change in earnings the allowance factor must be subtracted from both the nu- merator and the denominator. 'When the allowance factor in- creases, proportionately more is subtracted from the nu- merator. Therefore, the percentage change in income is less than reported. When the allowance factor decreases, proportionately more is subtracted from the denominator. Therefore, the percentage change in income will then be greater than reported. A numerical example will help illustrate this concept. Let IT = 1.00, IT-l = allowance factor‘T-l = 10%. The change in reported income .50, allowance factor T = 20% and the is 1.00 -.—5-(—)-"1=100% 137 The change in actual income is 1.00 - .20(l.00) .50 - .10(.50) which is equal to .8 _ 715 1 - 77.8% In the second case, let the allowance factor T = 10% and the allowance factor T-l = 20%. The change in reported income is still 100%. But the change in actual income is 1.00 - .10(l.00) .50 - .20(.50) - 1 = —3% - 1 = 125% The same example can be extended to the case when re- ported income decreases. _Let IT = .50, IT-l = 1.00, AFT = 20%, and AFT-l = 10%. The change in reported in- come is .50 _ _ The change in actual income is .50 - .20(.50) _ _ .40 _ = _ 1.00 - .10(1.00) l ‘ 290 I 55'5%° Under the second scenario when reported income de— creases and the allowance factor decreases, the actual de- crease should be less. Let IT = .50, IT-l = 1.00, AFT = 10%, AFT-1 = 20%. The change in reported income is .50 _ _ The change in actual income is .50 - .10(.50) _ _ .45 _ = _ 1.00 - .20(1.00) 1 ‘ .80 1 43'75% '138 These four cases can be combined into two for our purposes. Whether income increases or decreases, if the allowance factor increases, the actual increase or decrease in income is less than reported. In the second scenario, whether income increases or decreases, if the allowance factor decreases, the actual increase or decrease in income is greater than reported. This allows one to stratify both forecast error groups into two portfolios based on whether the allowance factor increases or decreases. Once the firms are grouped accord- ing to percentage forecast error in EPS, they are ranked from highest positive change in the allowance factor to lowest negative change and then split into two portfolios for both the high and low portfolio. These two portfolios are then matched, whereby the positive change portfolio is compared against the negative change portfolio and vice versa. The hypothesis is that the positive change port- folio should have a lower market return relative to the negative change portfolio if AFUDC provides some informa- tion which is used by the market to assess security re- turns.- Of course, as in the previous hypothesis, the betas for each portfolio are weighted to that they are equivalent. A diagram of these designs follows: 139 Case A - Firms ranked on Forecast error High ~-- - '--‘ 1 change “1 AFUDC -High .__. .___ Forecast Low Error 1 change “2 L0" AFUDC Forecast Error Case A above can be tested in an identical fashion to Case A in Section 4.6.1. Case B 1... Firms Firms “ ' a . ___— irms with Firms wi h -' .-1 F1 igh per— high per- 1 centage centage J change AFUDC L. _. split Firms‘with low per- centage change AFUDC L— F'21 21 ‘— _I. I_' — Case Case net] exax met hav rat tot eaI Th Va. H0 3r de 140 Case B above is also tested in an identical fashion to Case B in Section 4.6.1. Summarizing, the above designs deal exclusively with method I firms. A number of designs are formulated to examine the question of information content of AFUDC for method I firms. There are two stages with each stage having two alternatives. This means there are four sepa- rate designs which are used to test AFUDC. Stage 1 is the use of EPS forecast error to split the total number of method I firms into two groups. A) Split into high and low groups B) Split into nearly identical groups Stage 2 is the use of AFUDC characteristics to split each of the above groups into two portfolios. A) Portfolios based on the percent AFUDC B) _ Portfolios based on the percent change in AFUDC These four designs are then tested using either the uni- variate t-test, the multivariate version of the test, Hotelling T2, or a combination of both. Also the tests are used for mean differences (ud) or for the means (u) depending on the design. 4.7.3 Method I versus Method II Firms The next basic grouping of hypotheses concerns a com- parison of method I firms with method II firms. The dif- ferences in income streams for method I and method II are presented in Chapter I. The next step is to test whether there is any difference in the market reaction to the 141 announcement of earnings. In this case total earnings consist of two different methods of accounting for AFUDC. A major problem with any research into the differential market valuation of method I and method II firms is the confounding variables attributed to regulatory climate. A second problem is that any differences in market returns attributable to the announcement of AFUDC earnings may in fact be due to differences in the percentage forecast error or to differences in earnings changes from year to year. It is felt that any differences in regulatory climate should already be impounded in the relative risk measure, beta. Also any changes in this regulatory climate should be captured by the beta since a progressive moving average beta estimation period is employed in this study. There- fore, the equivalence of betas between the two portfolios of firms, method I and method II, implies the equivalence of all factors that are already public information which might have some effect on the market returns. This allows us to examine any event that might become known after capital market equilibrium has been established. For example, the disclosure of AFUDC earnings and any changes in this policy which actually occurs. A number of points must again be emphasized to pro- vide some justification for the tests performed here. First, empirical work on the differential market valuation of method I and method II firms is sparse. None of the previous empirical work has examined the instantaneous 142 market reaction to the disclosure of AFUDC earnings and the differences between method I and method II. Second, disclosure of which firms are method I and which are method II is not found in the typical accounting reports of the company, nor in widely used investment advisory services, such as Value Line and Standard and Poors. In fact, the classification of firms must be made by individual inves- tors from information provided in reports from the National Association of Regulatory Utility Commissioners. It is conceivable that some firms could be classified as either method I or method II depending on how these reports are used. Third, it is not at all clear that people familiar with AFUDC recognize that there is major differences in the income streams which accrue to these firms.39 As noted in previous chapters much of the empirical research on AFUchdoes not make any distinction between these two methods. Therefore, any empirical evidence on this par- ticular comparison may be of special interest to'a number of groups. 4.7.4 Procedure for Testing Method I vs. Method II To eliminate the second problem, that is; the dif- ference in market return may be influenced by differences in the forecast error of EPS, a matching procedure for individual firms is implemented. The percentage forecast error of earnings per share is calculated for method II firms. Method I firms are then matched with these firms on the basis of this percentage forecast error. To ensure 880 It CI: be 511 th- th me as ac I] bc 143 enough firms for the formation of portfolios a range of i 3% is used when matching method I with method II firms. It is felt that this range serves as a reasonable matching criteria for this particular attribute. Two specific designs are used to test for differences between method I and method II firms. Each-of these de- signs is the second stage of the two-stage process, whereby the first stage involves the matching of these firms on the percentage forecast error of earnings per share in the manner discussed above. Let the method II portfolio act as the control portfolio and let the method I portfolio act as the treatment portfolio. Since only thirteen method II firms are available for testing, only one portfolio for both the treatment and control is testable at any time, but this does not preclude the use of the same method II portfolio as a control portfolio in other tests. 4.7.5 Method I versus Method II: Percentage AFUDC This section details the designs used to test for differences between method I and method II when method I firms are selected based on the percentage AFUDC. Method I firms are chosen so that the firms with the highest per- centage of total income represented by AFUDC are formed into one treatment portfolio and the lowest percentage firms are formed into a separate treatment portfolio. We now have two independent treatment portfolios, but only one control portfolio. This allows one to test for dife ferences in method I and method II firms using the 144 univariate t-test in two separate, but not independent, procedures. The first is a comparison of the highest per- centage method I firms with method II firms. The second is a comparison of the lowest percentage method I firms with the method II firms. This allows one to test for any differential market reaction between method I and method II firms, inherent in the quality of those earnings as repre- sented by either high or-low percentages of AFUDC. A dia- gram follows: u-' -_T .__. l_. ‘ Firms with 1 highest compare with Z AFUDC t-test ‘73’ ,L___ __ All Firms . Mefimd II matched ‘ split rms wiflh #thod II -' "' ' on EPS . forecast Firms with ._ .__J error lowest compare with Z AFUDC t-test . Mathod I firms Method 11 firms Summarizing, method I firms are formed into portfolios in a two-setp procedure. First, these firms are matched 145 to method II firms according to the percentage forecast error of EPS. This leaves approximately thirty firms from the total of fOrty-five which are then split into two groups. Second, only the thirteen firms that have the highest percentage of AFUDC are formed into one portfolio. The thirteen firms that have the lowest percentage of AFUDC are formed into another portfolio. Again, the multivariate test is not appropriate here because the two tests are not independent, One can only perform the univariate t-test on each of the two treatment-control matched pairs. The hypo- thesis is that there should be a negative market reaction for method I firms relative to method II firms for the higher percentage treatment group, but no such relation- ship should exist for the lower percentage treatment. This is so because in the later case, it is a comparison of firms with essentially equivalent earnings, rather than a comparison of firms, where one portfolio consists of large amounts of earnings thought to be of inferior quality, as in the former case. 4.7.6 Method I vs. Method II: Percentage Change in AFUDC The next design is similar to the design developed for method I firms, whereby the change in the AFUDC per- centage is used to stratify firms into the treatment port- folios. As in the previous designs for method I and method II firms, method I firms are matched to method II firms on the basis of the percentage forecast error in EPS with- in a range of + 3%. The next stage is identical to the 146 previous design except for one modification. The method I firms are divided into two separate treatment groups based on the change in the AFUDC percentage, not just the percentage of AFUDC. The two extremes of this attribute (the change in AFUDC percentage) are used to form port- folios. The firms with the highest positive percentage change in the AFUDC percentage compose one treatment group. While the firms with the lowest negative percentage change in the AFUDC percentage compose the other treatment group. Since these firms are matched on the percentage fore- cast error of EPS, an increase in the AFUDC percentage from year to year implies the actual EPS and therefore the fore- cast error is not the same for method I firms as it is for method II firms. A positive percentage change in AFUDC for method I firms should have a negative market reaction re- 40 In the case of negative per- lative to method II firms. centage change in AFUDC, the relationship should be the opposite. In any case a difference in the actual market return between method I and method II firms, where these firms have been formed into portfolios, which are essen- tially identical except for the method of accounting, would imply some information content to AFUDC earnings. A diagram follows: .147. test differences Firms' ___AEUD in mean returns matched ' using t—test with ‘-- ‘-- Method II on fore— Firms with cast error lowest » - negative t-test , change in 1 AFUDC Z ...—- —_J. 1L— _‘ 1 Method I firms Method 11 firms The test procedure is identical to the previous de- sign developed for method I and methOd II. Two separate ‘t-tests are performed on the difference in mean monthly returns. 'One t-test is a test of high positive percentage changes in AFUDC for method I with method II firms. The other t-test is a test of low negative percentage changes in AFUDC for method I firms compared with method II firms. 4.7.7 Test Design: gguarterlyReports The test period used in this study encompasses not only the disclosure of AFUDC in the annual report, but also the quarterly announcements of earnings. The quarterly earnings reports, especially the first quarter reports, may contain information on AFUDC that may be used to assess the impact of AFUDC on firm valuation before the annual report_is available. For this reason similar tests 148 of mean returns are performed on the months surrounding the disclosure of first quarter reports. In essence, the sub-period, T = 0 to T = +6 is also a test of the disclosure of first quarter results. These results imply no basis for rejection of the null hypothesis in any of the designs used here. Nevertheless, the author feels a six month time period may bias these results by including other factors which could effect the market return for these firms. Therefore, a slightly different procedure is used in this situation, which enables one to focus more closely on the disclosure of the first quarter earnings and the amount of AFUDC. The first quarter ends on March 31, for all firms in this study since all firms have a December 31, fiscal year- end. A problem occurs in that the time when the first quarter reports are publicly available varies from firm to firm and year to year. The first quarter results are unaudited statements but they are reviewed by the company's auditors. This should ensure their accuracy and the fact that they are unaudited implies these reports are publicly available rather quickly. An examination of the date of the auditor's statement may give an approximate date when these reports are publicly available. There should be no reason to delay releasing these reports after the auditor has reviewed the statements. The SEC also puts some pres- sure on companies to release financial statements as quick- ly as possible. A random sampling of a few of these qua: as i qua: clo per in ave ced non Eac tic EdC the a 1 or be tu: Cr, mo; av, to in mg; the Prq 149 quarterly reports reveals two to five weeks after March 31 as a reasonable approximation for the release of the first quarter financial statements. Therefore, to test the dis- closure of AFUDC in the first quarter reports, a time period consisting of April and May, months T = l and T = 2 in each year is used as a reasonable approximation to the average time of disclosure for these companies. The pro- cedure used here differs from the preceding in that the monthly returns are not averaged over the eight years. Each month of each year serves as an independent observa- tion. This allows a total of sixteen observations for each treatment and control portfolio. In the use of this procedure, two observations about the power of the test became apparent. First, the use of a larger sample size, in this case sixteen versus thirteen or less as in the previous procedure, means the test will be more powerful. In general as sample size increases power also increases.41 Second, the use of monthly re- turns not averaged over some year to year period will de- crease the power Of the test. This is true because these monthly returns are not as homogeneous as if they had been averaged over years to eliminate any variation due to year to year differences. Since these two effects are opposite in direction it is not clear whether this procedure is more or less powerful than the previous procedure. Never- theless, it should provide some validation of the results presented so far. 08' ea it at 150 The specific test designs used for the quarterly earnings are identical to the designs used on the annual earnings except for the aforementioned differences. A further discussion of these designs would be redundant and add very little to the understanding of this research. Therefore, the designs on quarterly earnings are not dis- cussed again. 4.8 Data Selection and Validation The initial universe of firms used in this study con- sisted of all electric service utilities contained on the Compustat tapes. This includes two groups of electric utilities. First, there are those firms which derive all their income from electric power generation. Second, there are those firms whose primary source of income is electric power generation, but these firms also derive some minOr portion of their income from other services such as sales of natural gas. This initial sample con- sisted of one hundred forty-seven companies. A number of companies were then eliminated for various reasons. The first criteria used to eliminate firms was the dichoto- mizatiOn of firms into either method I or method II. This classification was made using the 1974 and 1977 annual re- ports of the National Association of Regulatory Utility Commissioners. Every attempt was made to keep these group- ings of firms into method I and method II as homogeneous as possible. Firms were then eliminated for the following reasons: 1) firms that had more than one state with 151 jurisdiction over their regulatory environment and where those states differed in the application of method I or method II; 2) firms that had some limits on the use of either method I or method II such as, only a certain amount of AFUDC was allowed, or only a certain amount of CWIP was allowed in the rate base. This filter left approximately fifty-four method I firms and eighteen method II firms. The next stage was to eliminate firms that did not have the proper data avail- able. Beta is used to match the relative risks of firms, therefore, at least sixty months of monthly security re- turns had to be available on the CRSP tapes before the start of the test period. The test period started on September, 1969, thus monthly returns must have been avail- able from August, 1964 to September, 1978. Also data must be available on the Compustat tapes for the calculation of earnings per share and calculation of the percentage of income represented by AFUDC. Therefore, data items on the Compustat tapes must be available form 1969 to 1978. This left a sample set consisting of firty three method I firms and thirteen method II firms. This sample set of firms is given in Tables 5 and 6, In Johnson's study, firms were also classified into method I and method II, but since a different set of criteria was used for data avail- ability, the sample set was not exactly the same as this study. Nevertheless, a majority of the firms were the same. The previous author called a number of firms to 152 Table 5 METHOD I COMPANIES Boston Edison Carolina Power Light Central Maine Power _Cincinnati Gas Cleveland Electric POwer Columbus 8 Southern Ohio Electric Co. Commonwealth Edison Duke Power Eastern Utilities Florida Power Light Florida Power Corp. Hawaiian Electric. Idaho Power. Indianapolis Power Nevada Power New England Electric Northeast Utilities Ohio Edison Pennsylvania Power Portland Electric Public Service Indiana Public Service New Hampshire Puget Sound Power Southern Cal Edison Tampa Electric ‘Union Electric united Illuminating Utah Power Arizona Public Service Central Illinois Light Central Illinois Public Service Iowa Electric Light a Power Co. Iowa Illinois Gas 8 Electric Co. Iowa Power a Light Co. ' Iowa Public Service Missouri Public Service Montana Dakota Utilities .Montana Power Northern Indiana Public Service Co. Northern States Power Co. Philadelphia Electric San Diego Gas Sierra Power Southern Indiana Gas Tucson Gas 8 Electric Co. Washington Water Power Wisconsin Electric Power Wisconsin Public Service 153 Table 6 METHOD II COMPANIES. Atlantic City Electric Power Baltimore Gas - Central Southwest Consumers Power Delmarva Power Detroit Edison” Gulf States Utilities Kentucky Utilities Co. Public Service Colorado. Savannah Electric . South Carolina Electric & Gas Co. ~Southwestern Public Service ' ‘Virginia Electric'& Power Co. 154 validate his classification and verify that the firms are indeed method I or method II. In this study, the 1974 and 1977 National Association Regulatory Commissioners annual report were used, and these reports were consistent with each other and the previous validation, therefore it was not thought necessary to validate again this classification scheme. This classification of firms into method I and method II was thought to represent as homogeneous a group- ing of firms as possible. This study makes exclusive use of both the CRSP and Compustat data bases. These tapes are used extensively for empirical research by'a large number of researchers. Thus, validation of those actual figures taken from these tapes was not necessary. Nevertheless, a number of random figures were checked against Value Line, especially with regards to EPS calculations which were crucial to this research. All figures were found to be consistent with the data on the Compustat tapes. Summary This chapter first briefly outlines the four theoreti- cal areas which serve as the basis for this research. These are: information content, market efficiency, capital asset pricing model and the distribution of security returns. These are then integrated into the specific application of research methodology that is used here. A general statement of the hypothesis and the appropriate statistical tests was also presented in that section. The next section detailed 155 the specific hypotheses to be tested. Problems with con- founding variables in previous research were discussed and the proposed solutions to those problems that this research methodology hopes to eliminate was also presented. The last section presented the data selection techniques used to obtain the sample for this research. 156 Footnotes 1N. J. Gonedes and D. Dopuch, "Capital Market Equi- librium, Information-Production, and Selected Accounting Techniques: Theoretical Framework and Review of Empirical Work. " Studies on Financial Accounting Objectives: 1974. Supplement to Journal ofIAccounting Research 12: 48- 169. N. J. Gonedes, Capital Market Equilibrium and Annual accounting Numbers: Empirical Evidence." Journal of Accounting Research (Spring 1974): 26- 62. N. J. Gonedes, "Risk Information, and the Effects of Special Accounting Items on Capital Market Equilibrium." Journal of AccountingWResearch (Autumn 1975b):220-56. N. J. Gonedes, Corporate Signaling, External Accounting, and Capital Market Equilibrium: Evidence on Dividends, Income, and Extraordinary Items." Journal of Accounting Research (Spring l978):26-79. 2Walter Thomas Harrison, unpublished dissertation, "The Information Content of Accounting Changes," Michigan State University, 1976, pp. 13-14. 3 Gonedes and Dopuch, p. 50-54. 4Raymond J. Ball and Phillip Brown, "An Empirical Evaluation of Accounting Income Numbers." Journal of Accounting_Research (Autumn 1968), p. 139- 177. R. F. Eskew and W. F. Wright, "An Empirical Analysis of Differential Capital Market Reactions to Extraordinary Accounting Items." Research paper no. 297, Graduate School of Business, Stanford University, 1976. N. J. Gonedes, Journal of Accounting Research (Spring 1974), 26-62. 5See, for example: Raymond J. Ball and Phillip Brown, "An Empirical Evaluation of Accounting Income Numbers," Journal of Accounting Research, 6 (Autumn 1968), 139-177. Eugene Fama, L. Fisher, M. Jensen, and R. Ball, "The Adjustment of Stock Prices to New Information," Inter- national Economic Review 10 (February l969):l-21. 6Shyam Sunder, "Relationship Between Accounting Changes and Stock Prices: Problems of Measurement and Some Empirical Evidence," Empirical Research in Accounting: Selected Studies 1973, Supplement to Journal of Accounting Research 11 (1973):l-45. 7Eugene F. Fama and Merton H. Miller, The Theory of Finance (Hinsdale, Illinois, Dryden Press, 1972), Chapter 7. 157 8Fisher Black, "Capital Market Equilibrium with Restricted Borrowing," Journal of Business 45 (July 1972), pp. 444-455. Fisher Black, Michael Jensen, and Myron Scholes, "The Capital Asset Pricing Model: Some Empirical Tests," Studies in the Theory of Capital Markets, ed. Michael Jensen (New York: Praeger, 1972). Eugene F. Fama and James MacBeth, "Risk, Return, and Equilibrium: Empirical Tests," Journal of Political Economy 81 (May/June 1973). PP- 607- 636. Michael Jensen, "Capital Markets: Theory and Evidence," Bell Journal of Economics and Management Science 3 (Autumn 1972), pp. 357-398. 9 Ibid. loEugene F. Fama, "The Behavior of Stock Prices," Journal of Business 28 (January 1965), pp. 34-105. 11 Ibid. 12Eugene F. Fama and Merton H. Miller, The Theory of Finance (Hinsdale, Illinois, Dryden Press, 1972), p. 264. 13N. J. Gonedes, "Risk Information, and the Effects of Special Accounting Items on Capital Market Equilibrium," Journal of Accountinngesearch (Autumn 1975), pp. 220-256. 14 Ibid., p. 221-230. 15Gonedes and Dopuch, pp. 48-169. 16Ibid., p. 50-55. 17Louis H. Rappaport, SEC Accounting Practice and Procedure, 3rd edition (New York: Ronald Press Co., 1972), p. 145. 18Walter Thomas Harrison, unpublished dissertation, "The Information Content of Accounting Changes," Michigan State University, 1976, p. 24. 19 Gonedes, 1978, p. 26-79. 20Harrison, 1976. 21R. Darrel Bock and Ernest A. Haggard, "The Use of Multivariate Analysis of Variance in Behavioral Research," in Handbook of Measurement and Assessment in Behavioral Sciences, ed. Dean K. Whitla (Reading: Mass.: Addison- Wesley, 1968), p. 102. 158 22Neil H. Timm, Multivariate Analysis with Applications in Education and Psychology, Wadsworth Publishing Company, Belmont, California, 1975, p. 226-229. Also Jeremy Finn, A General ModelfgrgMultivariate Analysis, Holt, Rinehart & Winston, p. 150-155 and 315- 319. 23William L. Hays, Statistics for the Social Sciences, Holt, Rinehart and Winston, New York, New York, 1973, p. 393.‘ 24 Timm, p. 226-229. ZSIbid. ZGIbid. 27Fama and Miller, Chapter 7. 28M. E. Blume, "Portfolio Theory: A Step Toward Its Practical Application," Journal of Business (April 1970): 152-73. 29Ibid. 3°Jensen, 1972, p. 357-398. 31Fama and MacBeth, "Risk, Return and Equilibrium: Empirical Tests," Journal of Political Economy (May/June 1973):607-636. 32 See empirical evidence cited in Chapter 3. 33See Johnson's dissertation, Chapter 5. 34Conversation with Z. Lew Melynck, utility industry consultant, Chairman, Department of Finance, University of Cincinnati, January 14, 1980. 35Lawrence D. Brown and Michael S. Rozeff, "The Superiority of Analyst Forecasts as Measure of Expectations; Evidence from Earnings," Journal of Finance, March 1978, pp. 1-16. 361bid. 37Ibid. 38Johnson firSt formulated this procedure for use in a different statistical design. 159 39The author is aware of people employed in the finance and accounting functions of method II firms who do not recognize these differences. 40See examples, p. 147-150. 4lwnliam L. Hays, p. 360-362. te Ch an mu 31 CC CHAPTER 5 EMPIRICAL RESULTS: DISCUSSION OF STATISTICAL TESTS This chapter reports the results of the empirical tests performed on the set of hypotheses outlined in Chapter 4. The statistical designs are both univariate and multivariate. Therefore, both the univariate and multivariate forms of the t-test were used when it was appropriate. The first two sets of hypotheses which are concerned with method I firms only, were tested with both the univariate and multivariate tests. In the discussion of the results from these tests, Case A refers to a multi- variate test design where two matched pairs are tested simultaneously. Case B refers to a simple univariate test design. As discussed in Chapter 4 the formation of the portfolios is slightly different in each case leading to the use of these different tests. The univariate tests could have been eliminated since the results are consis- tent with the multivariate tests in all time periods. But it was thought useful to include them here to verify the multivariate results in a slightly different design. 160 161 The set of null and alternative hypotheses that are tested in this research are first presented in summary form. 1) 2) 3) 4) 5) Ho: Ha: Ho: Ha: Ho: Ha: Ho: Ha: Ho: There is no difference in expected monthly rates of return between port- folios of method I firms, one consist- ing of high amounts of AFUDC as repre- sented by the percentage of net income, and the other consisting of low amounts of AFUDC. There is a difference in expected month- ly market rates of return. There is no difference in expected monthly market rates of return between portfolios of method I firms, one con- sisting of high percentage changes in the amount of AFUDC as represented by the percentage of net income, and the other consisting of low percentage changes. There is a difference in expected month- ly market rates of return. There is no difference in expected monthly market rates of return between a portfolio consisting of method I firms with high amounts of AFUDC as represented by the percentage of net income and a portfolio consisting of method II firms. There is a difference in the expected monthly market rates of return. There is no difference in expected month- ly market rates of return between a portfolio consisting of method I firms with low amounts of AFUDC as represented by the percentage of net income and a portfolio of method II firms. There is a difference in the expected' monthly market rates of return. There is no difference in expected month- ly market rates of return between a portfolio consisting of method I firms with a positive percentage change in com; 0161‘ son des Th5 wh Th 162 the amount of AFUDC as represented by the percentage of net income and a portfolio of method II firms. Ha: There is a difference in the expected monthly market rates of return. 6) Ho: There is no difference in expected month- ly market rates of return between a portfolio consisting of method I firms with a negative percentage change in the amount of AFUDC as represented by the percentage of net income and a portfolio of method II firms. Ha: There is a difference in the expected monthly market rates of return. At this point there are two hypotheses that refer to comparisons between method I firms only and four hypotheses that are concerned with method I versus method II compari- sons. Recall from Chapter 4 that there are two different designs for each hypotheses that concerns method I firms. This allows one to test two hypotheses in different ways, which should provide some verification of the results. The four hypotheses that concern method I versus method II comparisons use the same method II portfolio as a con- trol portfolio. Nevertheless, the hypotheses are separate and independent and the simple t-test will be used to test each one. This might be stated better as a general hypo- thesis, that is: Ho: There is no difference in expected monthly rates of return between a portfolio of method I firms and a portfolio of method II firms. This general hypothesis can be examined more thoroughly through the use of the four hypotheses given above. 163 5.1 Tests of-Percentage AFUDC: High EPS Forecast Error vs. Low EPS Forecast Error There are six hypotheses given above. Five specific tests corresponding to different sub-periods are performed using each hypothesis. Each sub-period test is discussed separately. The results for the first hypothesis are given in Table 7. Table 7 and all future tables which Show the multivariate results also include the separate univariate statistics. The first test is the total time period of T = -6 to T = +6. In this case the first uni- variate test is a comparison of a portfolio of high EPS forecast error high percentage AFUDC (for ease of under- standing and future reference, signified high-high) and a portfolio of high.EPS forecast error low percentage AFUDC (high-low). The t-statistic is -.58. The second uni- variate comparison is a portfolio of low EPS forecast erorr high percentage AFUDC (low-high) and a portfolio of low EPS forecast error low percentage AFUDC (low-low). The t-statistic is .09. Finally, the multivariate T2 statistic which is a simultaneous test of High-High vs. High-Low and Low-High vs. Low-Low is .6807. All of these values given so far imply that the null hypothesis of no difference in monthly returns can be accepted at the .10 level. The two sub-periods, T = -3 to +3 and T = -2 to +2, focus more closely on the time period immediately around the disclosure month, T = 0. The univariate t-statistics are -.60 and .30 for the high-high vs. high-low and low-high 164 Table 7 Percentage AFUDC: Method I Firms High EPS Forecast Error Versus Method I Firms Low EPS Forecast Error Case A Total Period 13 Months (Time = -6 to +6) Mean 0 t T2 High Forecast High % .0044 .018 -.58 .6807 High Forecast Low % .0089 .017 Low Forecast High % .0086 .017 .09 Low Forecast Low % .0082 .018 7 Months (Time = -3 to +3) High Forecast High % .0044 .024 -.60 6.18** High Forecast Low % .0108 .019 Low Forecast High % .0085 .026 .30 Low Forecast Low % .0043 .021 5 Months (Time = -2 to +2) High Forecast High % -.0009 .027 -.577 6.86** High Forecast Low % .0082 .022 Low Forecast High % .0020 .029 .263 Low Forecast Low % -.0021 .021 Anticipation Period 6 Months (Time = -6 to -1) High Forecast High % .0141 .016 -.34 .36 High Forecast Low % .0180 .018 Low Forecast High % .0179 .021 .04 Low Forecast Low % .0176 .017 Reaction Period 7 Months (Time = 0 to +6) High Forecast High % -.0039 .015 .52 1.0468 High Forecast Low % .0010 .012 ' Low Forecast High % .0007 .016 .08 Low Forecast Low % .0001 .015 Critical Values for T Statistic, a = .10 t, 12 = 1.356 t, 6 = 1.440 t, 5 = 1.476 t, 4 = 1.533 **Significant at the .10 Level 165 vs. low-low portfolios, respectively, for T = -3 to T = +3. Both.values are well within the bounds for acceptance of the null hypothesis. The multivariate T2 statistic shows quite different results. The value is 6.18, which is sig- nificant at a = .10. At first glance this might seem im- portant, but significance may be due to the attribute that is not of interest here, that is; high EPS forecast error versus low EPS forecast error. Therefore, a second test was performed by rearranging the portfolios so that the AFUDC component was held constant and the EPS forecast erros was tested. This comparison was then high-high vs. low-high and high-low vs. low-low. The test statistic in this case was virtually identical and of course signifi- cant at a = .10. This implies that the significance was more probably due to the EPS forecast error rather than the percentage of AFUDC. This is also consistent with the results and discussions given by a number of other re- searchers concerned with.EPS forecast error, which is out- lined in Chapter 4. The time period, T = -2 to +2, gives similar results. The univariate t-statistics are -.577 and .263 which of course are not significant. The T2 statistic is 6.86 which is significant at a = .12. But again rearranging the portfolios as in the previous case gives the same re- sults. The significance is apparently due to differences in the EPS forecast error rather than differences in the percentage of AFUDC. 166 The anticipation sub-period, T = -6 to T = -1 yields results similar to the total period. The t-statistics in the univariate cases are -.34 and .04, respectively. The T2- statistic in the multivariate case is .36. These values are well within the limits needed for accepting the null hypothesis. The reaction sub-period, T = 0 to T = +6, has results consistent with the above. An interesting aspect is that mean monthly return for the high-high portfolio is negative. This should be expected since a high EPS forecast error and high amounts of AFUDC should imply a negative view by the market. In any case the t-statistics are .52 and .08, respectively. The T2- statistic is 1.046. All the values imply the null hypothesis can be safely accepted at the .10 level. 5.2 TeSts of Percentage Changes in AFUDC: High.EPS Forecast Error vs. Low EPS Forecast Error‘ This next hypothesis is concerned with changes in the percentage of income represented by AFUDC. The for- mation of these portfolios is described in detail in the previous chapter. Briefly, a portfolio of method I firms with increases (high.percentage change) in the percentage of net income represented by AFUDC is compared with a port- folio of method I firms with decreases (low percentage change) in the percentage of net income represented by AFUDC. The results are summarized in Table 8. This scenario is analogous to the previous discussion, in that, 167 Table 8 Percentage Change AFUDC: Method I Firms High EPS Forecast Error Versus Method I Firms Low EPS Forecast Error Case A Total Period 13 Months (Time = -6 to +6) Mean 0 t T2 High Forecast High % Change .0066 .017 -.02 .127 High Forecast Low % Change .0068 .019 Low Forecast High % Change .0075 .018 -.17 Low Forecast Low % Change .0088 .021 7 Months (Time = -3 to +3) High Forecast High % Change .0050 .021 -.10 .049 High Forecast Low % Change .0062 .023 Low Forecast High % Change .0038 .023 -.16 Low Forecast Low % Change .0060 .027 5 Months (Time = -2 to +2) High Forecast High % Change .0035 .026 .19 .927 High Forecast Low % Change .0004 .025 Low Forecast High % Change -.0007 .025 -.07 Low Forecast Low % Change .0004 .031 Anticipation Period 6 Months (Time = -6 to -1) High Forecast High % Change .0182 .015 .09 .659 High Forecast Low % Change .0173 .020 Low Forecast High % Change .0176 .019 -.33 Low Forecast Low % Change .0214 .021 Reaction Period 7 Months (Time = 0 to +6) High Forecast High % Change -.0034 .013 -.15 .241 High Forecast Low % Change -.0023 .013 Low Forecast High % Change -.0012 .013 .10 Low Forecast Low % Change .0020 .016 168 a 2 x 1 vector of mean returns is examined. Each element of this vector can be tested using the univariate t-test, but a simultaneous test of both means requires the use of Hotelling's T2 test. The 2 x 1 vector represents the pair- ings of portfolios as follows: high EPS forecast error high percentage changes in AFUDC (High-High) vs. high EPS forecast error low percentage change (High-Low) and the second element is low EPS forecast error high percentage change in AFUDC (Low-High) vs. low EPS forecast error low percentage change in AFUDC (Low-Low). In the vast majority of cases a high percentage change in AFUDC actually meant an increase in the percentage from year to year and low percentage change actually meant a decrease in the percent- age from year to year. As in the previous case the first time period to be discussed is the total period, T = -6 to +6. The t-statis- tics for the univariate cases are -.02 and -.17, respec- tively. The T2- statistic is .127. All of these values indicate that one fails to reject the null hypothesis at the .10 significance level. The two sub-periods, T = -3 to +3 and T = -2 to +2 give similar results. For T = -3 to +3, the t-statistics are -.10 and -.16, respectively and the T2- statistic is .049. For T = -2 to +2, the t-statistics are .19 and -.07, respectively and the T2- statistic is .927. Again, all these values imply an acceptance of the null hypothesis at the .10 level. 169 The anticipation period, T = -6 to -1 provides re- sults consistent with the above. The t-statistics are .09 and -.33, respectively, and the T2- statistic is .659. The reaction sub-period is also similar. The t-statistics are -.15 and .10 and the T2- statistic is .241. All these values are well within the area of acceptance of the null hypothesis. As in the previous design the average monthly return is negative in the reaction period, in this case for three of four portfolios. This can be compared against the anticipation period where the monthly returns are positive. Even though no statistical significance can be attached to this observation it may provide some basis for further research. This reaction period also includes the disclosure of the first quarter earnings and any ne- gative reaction in this time period may be influenced by this initial disclosure of AFUDC, even though the dis- closure is limited. In later sections of this chapter, further tests are performed that focus on just the first quarter. 5.3 Percentage AFUDC: Method I Firms Matched on EPS Forecast Error, Case B This is a slightly different design from the design discussed in section 5.1. This case tries to eliminate any differences in mean monthly returns that may be due to differences in the EPS forecast error. As it was pointed out in Chapter 4, two distinct portfolios of method I firms are formed so as to match as closely as 170 possible the EPS forecast error. The second step is to compare high percentage AFUDC portfolios versus low per- centage portfolios. There were actually two portfolios for both high and low percentage AFUDC firms. This allows two separate univariate tests, which are essentially iden- tical. Therefore, to prevent redundancy only one result is reported in Table 9. The results are entirely con- sistent with the preceding results given in Section 5.1. All the t-statistics are below the critical value and the null hypothesis can be accepted at the .10 level. 5.4 Percentage Change in AFUDC: Method I Firms Matched on EPS Forecast Error, Case B This design is identical to the design given in Sec- tion 5.3 except the portfolios are formed on the AFUDC variable in a different way. Here the change in the AFUDC percentage rather than the absolute percentage is used to form portfolios. The results are given in Table 10. The results are again consistent with results given in Sec- tion 5.2 and 5.3. All the t-statistics are insignificant, implying acceptance of the null hypothesis at the .10 level.' Summarizing the tests to this point, there appears to be no basis for rejecting the null hypothesis, where the null hypothesis as briefly stated is as follows: 'there is no information content in AFUDC earnings as dis- <:1osed in the annual report among portfolios consisting cof method I forms. Any information concerning AFUDC 171 Table 9 Percentage AFUDC: . Method I Firms Matched on EPS Forecast Error Case B Total Period 13 Months (Time = -6 to +6) Mean 0 t High Percentage . .0080 .019 .17 Low Percentage .0067 .020 7 Months (Time = -3 to +3) High Percentage .0070 .023 .18 Low Percentage .0046 .025 5 Months (Time = -2 to +2) High Percentage .0000 .028 .03 Low Percentage .0005 .025 Anticipation Period (Time = -6 to -1) High Percentage .0174 .019 -.21 Low Percentage .0152 .020 Reaction Period (Time = 0 to +6) High Percentage -.0026 .014 .24 Low Percentage -.0006 .015 172 Table 10 Percentage Change AFUDC: Method I Firms Matched on EPS Forecast Error Case B Total Period 13 Months (Time = -6 to +6) Mean 0 t High % Change _.0059 .021 .09 Low % Change .0066 .020 7 Months (Time = -3 to +3) High % Change .0019 .024 .30 Low % Change .0057 .025 5 Months (Time = -2 to +2) High % Change -.0020 .026 .22 Low % Change -.0009 .027 Anticipation Period 6 Months (Time = -6 to -1) High % Change .0159 .019 .16 Low % Change .0176 .019 Reaction Period 7 Months (Time = 0 to +6) High % Change -.0012 .011 .16 Low % Change -.0025 .017 173' earnings, that is, whether a firm has higher or lower amounts of AFUDC or whether a firm decreases or increases the amount of net income represented by AFUDC is apparent- ly of little value in assessing the future market returns for these firms. At least for differences between these method I firms for the time period, 1970-1977, the dis- closure of AFUDC in the annual report is already impounded in the relative risk measure, beta. The beta of the firms in this sample apparently incorporates any information about AFUDC that may be of use in investors' judgment of future security returns. In this sense the market is ef- ficient. AFUDC earnings are already reflected in stock values for method I firms or the process of adjustment to these earnings is too fast for investors to act on this information in a profitable manner. 5.5. Method I Low-Percentage-AFUDC Versus Method II The following discussion is concerned with the dif- ferential market reaction to the two methods used for AFUDC. In the following designs, only one portfolio of method II firms can be formed, therefore this portfolio is used as a control portfolio in all future designs. Thus, the appropriate statistical test is the univariate t-test. The first hypothesis concerns a comparison of ,a method II firm with.a matched groups of method I firms, *where these firms contain low percentages of AFUDC. The results are contained in Table 11. A test of the mean 174 difference for the total time period, T = -6 to T = +6, gives a t-statistic of -.25. The results for the antici- pation sub-period, T = -6 to T = -l, and the reaction sub- period, T = 0 to T = +6 are consistent with the total period. The t-statistics are -.02 and -.43, respectively. All of these values are well within the limits needed for acceptance of the null hypothesis at the .10 level. As in the previous tests of method I firms, further tests are performed with the time period focusing more closely on the disclosure month. These results are similar to the above. The seven month period, T = -3 to T = +3, and the five month period, T = -2 to T = +2, result in t-statistics of -.21 and .02, respectively. Again, the null hypothesis can be accepted at a a significance level of .10. None of the time periods tested for these groups of firms, offer any evidence for rejection of the null hypothesis. In fact, a closer examination of Table 17 reveals the actual market return for method II firms is actually less than the actual market return for method I firms in four of five test periods. These differences are extremely small. Nevertheless, the direction may have some signi- ficance since the previous chapters imply method II firms' earnings are of higher quality and therefore the actual market return should be greater than the returns for method I firms. II I Low II I LOW II I Low II I LOW II I LOW 175 Tab 1e 11 Low Percent Method I Versus Method II Total Period 13 Months (T = -6 to T = +6) Mean 0 .0067 .022 AFUDC .0086 .017 Anticipation Period (T = -6 to T = -l) .0181 .024 AFUDC .0183 .017 Reaction Period (T = 0 to T = +6) -.0031 .015 AFUDC .0003 .014 7 Months (T = -3 to T = +3) .0032 .027 AFUDC .0059 .021 5 Months (T = -2 to T = +2) -.0012 .031 AFUDC -.0016 .019 -.25 -.02 -.43 -021 .02 176 5.6 Method I High.Percentage AFUDC Versus Method II Firms The next hypothesis compares method I firms contain- ing high percentages of AFUDC with a portfolio of method II firms. It should be noted that the firms used to form either the high percentage portfolio or the low percent- age portfolio are mutually exclusive. This, a completely different group of firms serves as the basis of compari- son for the portfolio of method II firms. Therefore, these are separate tests even though the control port- folio is the same in all cases. The test results are contained in Table 12. The thirteen month time period, T = -6 to T = +6, reveals results similar and consistent with the previous hypo- thesis. In this case the t-statistic is -.08. The avers age monthly returns for the anticipation sub-period, T = -6 to T = -l, are equivalent, thus the t-statistic is .00. The reaction sub-period, T = 0 to T = +6, gives a t-statistic of -.15. As-in the previous cases, the aver- age monthly returns for the reaction period are negative for both portfolios. This reaction period contains both the disclosure of first quarter earnings and annual earn- ings. The negative monthly returns while not significant mey‘imply a small market reaction to this disclosure even though it is unclear which.disclosure may be important. IFurther tests of the first quarter results are provided in a later section of this chapter. II I High II I High II I High II I High II I High 177‘ Table 12 High Percent Method I Versus Method II Total Period 13 Months (T = -6 to T = +6) Mean 0 t-Value .0066 .022 -.08 % AFUDC .0073 .020 Anticipation Period (T = -6 to T = -l) .0183 .024 .00 % AFUDC .0183 .021 Reaction Period (T = 0 to T = +6) -.0034 .015 -.15 % AFUDC -.0021 .016 7 Months (T = -3 to T = +3) .0030 .027 .02 % AFUDC .0027 .023 5 Months (T = -2 to T = +2) -.0017 .031 -.07 % AFUDC .0004 .027 178 The tests of the two sub-periods of T = —3 to T = +3 and T = -2 to T = +2 reveals t-statistics of .02 and -.07, respectively. All of the t-statistics for this hypothesis imply an acceptance of the null hypothesis at a a signifi- cance level of .10. In fact, as in the previous case, the actual monthly returns for the method II portfolio is less than the method I portfolio in three of the five test periods. For this hypothesis and the previous hypo- thesis, there appears to be little evidence for rejection of the null hypothesis in favor of the alternative hypo- thesis. That is, there is no difference in actual month- ly returns for a portfolio of method I firms containing either low percentages of AFUDC or high percentages of AFUDC in comparison with a portfolio of method II firms. The hypothesis of no information content for AFUDC earnings can be accepted. 5.7 Method I Positive Percentage Change AFUDC Versus Method I I The next two hypotheses compare method I firms that have changes in the percentage of AFUDC with method II firms.- The author feels this procedure is perhaps the best design for testing the market reaction to the dis- closure of AFUDC. This is due to the fact, that if AFUDC is important in the investors' assessment of the proper :market valuation'of these firms' common stock, any change in AFUDC should also be reflected in changes in common stock values. The first hypothesis concerns a comparison 179 of method I firms, which have an increase in the percent- age of net income represented by AFUDC, with method II firms. The results for this hypothesis are summarized in Table 13. The first test is of the total time period, T = -6 to T = +6. The t-statistic in this case is -.l7. The an- ticipation sub-period, T = -6 to T = -l and the reaction sub-period, T = 0 to T +6 give similar results. The t- statistics are .14 and -.53, respectively. These values indicate that the null hypothesis can be accepted at a a -2 to T = +2, significance level of .10. The tests of T and T = -3 to T = +3 offer identical conclusions. The t- statistics are -.20 and -.03, respectively. The results indicate no significant market reaction to a change in the amount of net income represented by AFUDC. This is sur- prising in view of the fact that many firms had the per- centage of net income represented by AFUDC change substan- tially during this test period. In fact, for a number of cases the percentages doubled or tripled from year to year. Even though there is no statistically significant (iifferences between the two portfolios, the biggest dif- ference does occur during the reaction period when any market reaction to a change in accounting numbers may be expected to occur. The results here also confirm previous results that the‘average monthly return for method II firms is lower than the return for method I firms in three of the five time periods. This may have some significance. 180 Table 13 Positive Percent Change: Total Period 13 Months (T II I Positive % Change Anticipation Period (T II II Positive % Change Reaction Period (T II I Positive % Change 7 Months II I Positive % Change 5 Months (T II I Positive % Change (T, Method I Versus Method II = -6 to T = +6) Mean 0 t .0065 .022 -.17 .0078 .015 = -6 to T = -l) .0149 .024 .14 .0134 .017 = 0 to T = +6) -.0034 .015 -.52 .0002 .011 = -3 to T = +3) .0029 .027 .03 .0026 .018 = -2 to T = +2) -0001-8 0031 -020 .0016 .021 181 If method II earnings are actually considered to be of superior quality when compared to method I earnings as previous chapters imply, then the results here do pose some question as to the underlying reason. One would not expect the method II portfolio to have an average monthly return less than the method I portfolio. This is espe- cially true since the method I portfolio consists of a random selection of method I firms which would be different in each year. This difference cannot be attributed to a static group of method I firms. The firms change from year to year based on the criteria previously mentioned. Thus, though these differences are not statistically significant and fall well within the range for acceptance of the null hypothesis, the fact that the direction of the difference is the opposite of what theory may imply is interesting and perhaps a fruitful area for further research. 5.8 Method I Negative Percentage Change Versus Method II The next hypothesis concerns a portfolio of method I firms, where those firms have a decrease in the percent- age of net income represented by AFUDC, compared with a portfolio of method II firms. Again, if AFUDC does have some information content, a portfolio containing firms with changes in the amount of AFUDC compared to a port- folio composed of firms, where the amount of AFUDC is not relevant and therefore any changes in AFUDC should not be relevant, appears to offer best design for testing AFUDC. 182 The results are given in Table 14. The results of these tests on this hypothesis are consistent with the dis- cussion of the previous hypothesis. The null hypothesis can be accepted at the .10 significance levels for all test periods. The total time period, T = -6 to T = +6, has a t-statistic of -.25. The anticipation sub-period, T = -6 to T = -1, again has the lowest t-statistic, -.03 and the reaction sub-period, T = 0 to T = -6, has the highest t-statistic, -.42. The two sub-periods which focus more closely on the disclosure date, T = -3 to T = +3, and T = -2 to T = +2 reveal similar results. The t-statistics are -.34 and -.21, respectively. There is no basis for rejection of the null hypothesis. There is no difference in average monthly returns between method I firms and method II firms when those firms have equivalent relative risks. There may be some problem with using parametric tests on small samples. Siegel states non-parametric tests may be the appropriate statistical tests when the sample sizes 1 Thus, the Welsh test is performed for testing are N i 6. on the five-month sub-period. The results are identical to the parametric test. 5.9 Tests of Quarterly Earnings Announcements The results are summarized in Table 15. These re- sults are consistent and supportive of the null hypothesis, that is; there is no information content contained in the disclosure of AFUDC earnings. These results hold for both Negative Percent Change: Total Period 13 Months (T II I Negative % Change 183 Table 14 Mean .0066 .0086 Anticipation Period (T = -6 to T = 0' .023 .019 +6) -6 to T = -1) II .0180 .026 I Negative % Change .0184 .018 Reaction Period (T = 0 to T = +6) II -.0032 .015 I Negative % Change .0002 .016 7 Months (T.= -3 to T = +3) II .0029 .027 I Negative % Change .0075 .023 5 Months (T = -2 to T = +2) II -.0017 .031 I Negative % Change .0020 .026 Method I Versus Method II -.25 -003 -.42 -.34 -.21 184 Table 15 lst Quarter Results Method II Method I Negative % Change Difference Method II . Method I Positive % Change Difference Method II Method I High % Change Difference Method II Method I Low % Change Difference Method I High % Change Method I Low % Change Difference Method I Low % Change Method I High % Change Difference Mean -.022 -.0143 -.00786 -.0227 -.0122 -.01027 -.0228 -.0179 -.00497 -.0232 .0121 .0059 -.0219 .0217 .0033 -.0145 -.0217 .0046 .048 .046 .022 .048 .043 .020 .048 .050 .026 .047 .049 .016 .044 .036 .025 .049 .048 .027 t-Value -.34 -.52 .19 -.36 .13 .17 185 the comparison of method I firms with a similar group of method I firms differing only in the information variable, AFUDC, and for the comparison of method I firms with method II firms. Table 15 contains the mean and standard devia- tion for the sixteen monthly observations for both the treatment and control portfolio, as well as, the mean and standard deviation for the differences in the monthly re- turns. Also included is the t-statistic which is used to test whether the mean difference is statistically signifi- cantly different from zero. A detailed discussion of these results would be redun- dant and add very little to any conclusions that might be inferred from the results on the annual reports. An exam- ination of Table 21 reveals all the results on the first quarter disclosure are consistent with the results of the annual disclosure of AFUDC. In each case the null hypo- thesis can be accepted at the .10 level. One additional observation should be discussed at this point. One may notice that the mean monthly returns, for the months of April and May, over this eight-year period 'were consistently negative in all cases. In fact in most cases the return averaged around negative 2%. As a point of reference the mean monthly return for the market index taken from the CRSP tapes were .0065 for these same months. Since, the monthly market return for the market index is «quite small, it could be due to chance that the monthly re- turn for utility stocks is negative in this eight year pt 186 period. On the other hand, the controversy that surrounds AFUDC and the disclosure of AFUDC in the first quarter may have some effect on the monthly return for all utility stocks in comparison with.stocks in general. This is an empirical question that must wait for future research to answer. It certainly bears further scrutiny in light of the differences of opinion among followers of the elec- tric utility industry concerning the use of AFUDC. 5.10 Summary The results of this chapter offer no basis for re- jection of the null hypotheses. As financial theory would indicate the semi-strong form of market efficiency implies that all publicly available information should be impounded in stock prices. Therefore, the equivalence of two port- folios' relative risks as measured by beta should imply the expected market returns are also equivalent. Any informa- tion content contained in the disclosure of AFUDC is al- ready impounded in this risk assessment. These results hold for comparisons of portfolios of :method I firms. The results alSo are consistent for com- parisons of method I and method II firms. Differences that in the past have been attributed to AFUDC earnings :may in fact have been due to some other attribute, fOr «example, differences in EPS forecast error. The conclusion Ihere is that the market is efficient with respect to dis- «closure of AFUDC earnings. This does not imply that AFUDC 187 does not affect the equity risk. It merely suggests the disclosure of AFUDC earnings is properly impounded in the systematic risk of a security as reflected in that se- curity's beta. CHAPTER 6 CONCLUSIONS AND SUMMARY The idea of lower quality of earnings is shown to involve both the risk of future cash flows and the dis- crepancy between the income reported and the present value of the income stream actually received. A mathematical development first proposed by Johnson is used here to show why AFUDC earnings should actually be considered of lower quality when compared to other utility earnings. This development is also used to describe the differences between the present value of the cash flows reported by method I firms and those cash flows reported by method II firms. Briefly stated, for method I firms in the normal situation of no compounding interest, there is a discrep- ancy between the income reported in the annual reports and the present value of the income actually earned. In the competitive environment, the reporting of earnings should signal the market that the present value of the firm has increased by an identical amount. This is not true of AFUDC earnings in the electric utility industry. This is due to the fact that the income is reported in 188 a *-~— ___.—.~._. __ the valu muck in ' I f sub whi one in: Of re 189 the current period, but it does not actually increase the value of the firm through additional cash flows until a much later date when the plant is completed and included in the rate base. Therefore, due to this time lag, method I firms report income in the current period that can be substantially greater in present value terms than the cash which is actually received by the firm in later periods. Method II firms are different from method I firms in one aspect. The construction work-in-progress account is included in the rate base, which then earns the same rate of return as the firm's other assets. Method II firms then report an amount, equivalent to the increase in the present value of the firm, as income. For method II firms, no discrepancy exists between the income reported and the present value of the cash flows the firm actually receives. This leads to two observations. First, an increase in the percentage of net income re- presented by AFUDC for method I may be viewed unfavorably by the market. Therefore, any increase in the absolute amount of AFUDC, if the percentage is held constant, may be viewed favorably. This is true because an increase in (AFUDC implies higher cash flows at some later date compared to the situation where there is no AFUDC. Given the effi- ciency of security markets, there certainly may be some anticipation of future AFUDC amounts. However, finding the exact percentage of earnings represented by AFUDC may ibe quite difficult to forecast. This may be due to two 190. reasons. The disclosure of AFUDC and the affects it may have on both the income statement and balance sheet are extremely limited and certainly unclear. In fact, as discussed in earlier chapters, two different researchers looking at the same set of numbers may draw different conclusions as to the affects AFUDC may have on the value of the firm. Second, forecasting AFUDC percentages requires both a forecast of earnings from operations and a forecast of AFUDC earnings. Again, either or both forecasts may be wrong. The fact that two separate forecasts have to be made only compounds the chances of error in the final anticipated percentage. An examination of the percentages represented by AFUDC suggests substantial instability. The percentages change often and by large magnitudes. These two factors, which are difficulty in forecasting and difficulty of interpretation, indicate that unanticipated changes in AFUDC may be a factor in security valuation. The statis- tical design employed here is essentially a test of differ- ences in mean monthly rates of return of portfolios. These portfolios were comprised of firms where each firm had similar attributes of the AFUDC variable. The firms within each portfolio were weighted so that the betas of the two portfolios in the matched pair were equivalent. The extensive empiriCal and theoretical literature concerning the efficient market hypothesis and the capital asset pricing model suggest portfolios with equivalent betas 191 will have equivalent rates of return. The theory implies that only unanticipated changes or levels of AFUDC should cause some reaction from the security markets. The above observations hold for method I firms but they may not be true for method II firms for the following reason. AFUDC earnings are the same as other utility earnings for method II firms. Therefore, any increase in the per- centage of net income or an increase in the absolute amount of AFUDC may be viewed favorably by the market. The question of anticipating the percentage of earnings re- presented by AFUDC is a moot point since both sets of earnings are indentica1.' The question of anticipation is one of forecasting total earnings and is not one of forecasting AFUDC percentages. The above observations lead to the statistical designs developed in this research. The basic premise is that the market does not adequately anticipate changes in AFUDC and perhaps does not fully understand the difference between method I and method II firms. Preliminary attempts to forecast AFUDC have been abandoned due to the difficulties of formulating adequate forecasting models. Thus, the test results from the portfolios formed from the forecast model are not re- ported here due to the problems mentioned above. The results of these tests, even though few in number, show no statistical differences in market returns due to AFUDC. The empirical tests consist of two basic strategies. The first concerns comparisons of monthly rates of 192 return for only method I firms; the other examines method I firms in comparison to method II firms. Each strategy has two separate designs. The designs test for any infor- mation content in the disclosure of AFUDC numbers found in the company's financial statements. One design is con- cerned with tests of portfolios differing in the absolute percentages of income represented by AFUDC. In other words, high percentage firms are compared against low percentage firms. The second design examines firms with differences in the increases or decreases in the AFUDC percentages. The tests are concerned with any information content contained in AFUDC earnings and changes in AFUDC earnings which may be unanticipated. If this information is useful to investors, changes in AFUDC levels may imply changes in the market valuation of electric utilities. The change in market valuation may be anticipatory or reactionary depending on the efficiency of investors' perceptions of this information. For this reason the test period is separated into two sub-periods, consisting of before and after the disclosure date. ' The semi-strong form of the efficient market hypothesis indicates that any new information will be impounded rapidly in stock prices. In this research, new information may be considered any unanticipated changes in AFUDC levels. If there is no information content in the disclosure of AFUDC earnings, and if all other publicly available information 193 is already impounded in the market assessment of the systematic risk of these electric utilities, then the expected returns from the two portfolios of each matched pair should be equivalent. Any significant difference between the market returns of portfolios, identical except for the specific attri- butes concerning AFUDC, should be the result of informa- tion contained in the disclosure of AFUDC which the market has not adequately anticipated. The market returns are calculated around this disclosure date, which is the date of public availability of the annual or quarterly reports. The empirical results presented in the first sections of Chapter Five, which are concerned with method I firms only, indicate that there is no information con- tent in the disclosure of AFUDC. Apparently any informa- tion contained in AFUDC disclosure which might be used to assess market returns for securities is already impounded in the measure of systematic risk, beta, as financial theory indicates. This does not imply that AFUDC does not affect security risk. It merely suggests that if AFUDC affects security risk, it is already impounded in the beta. The results suggest there may be some slight reaction to this information as disclosed in the quarterly reports. However, the reaction is too small to develop any trading strategy which might use the disclosure of this accounting information to earn abnormal profits. 194 Tests of market rate of return differences between method I and method II firms yield similar conclusions. The market is apparently efficient in the assessment of the quality and risk of the earnings for these two groups of companies though no such distinction exists in the financial statements. The disclosure of these earnings apparently contains no information that may be used to judge market performance of method I firms relative to method II firms. There is some evidence suggesting there are signifi- cant differences in rates of return between method I and method II firms. Upon closer examination, the differences may be attributed to another factor and are not a result of AFUDC disclosure. When tests are performed to separate the affect of total earnings changes from changes in AFUDC, the results indicate differences in rates of return may be due to the total earnings changes and not due to AFUDC. Failure to take into account this factor may have biased previous tests of AFUDC regarding method I and method II firms. The research conducted here may have some implications for both accounting and rate-making. Through SEC and FERC pressure, the accounting profession has required more in- formation concerning the disclosure of AFUDC. The infor- mation provided*has changed over the time period examined here, apparently to the benefit of investors. Assume that the dominant, price-setting investors are the sophisticated 195 investors who understand the nuances of utility accounting which includes AFUDC. Then, even though the disclosure of AFUDC is still minimal and there is no delineation of the differences between method I and method II firms, these investors are apparently not fooled by AFUDC earnings. Any information pertaining to AFUDC is impounded in the security risk measure, Beta. On the other hand, unsophi- sticated investors need only concern themselves with published Beta as a guide to future security returns. It is doubtful whether unsophisticated investors have both the time and technical skills necessary for an accurate understanding of the implications for market rates of return that AFUDC earnings disclosure might have. Again, this does not indicate that AFUDC does not affect security risk. It merely suggests enough investors understand any affects AFUDC might have on security risk; they price the securities properly; and they respond to any changes in AFUDC quite rapidly. Further disclosure of the practices concerning AFUDC may help unsophisticated investors better assess the security risk of electric utilities. However, the research done here implies the market is semi-strong efficient with respect to this information. Additional disclosure may be both unnecessary and an added cost 'that the utility's customers will eventually have to bear . The rate-making implications are harder to assess. lividence implies changes in rates of return for method I 196 firms is not significantly different from method II firms. A hypothesis is now presented which may account for these findings: the market treats all firms as if they use the method I approach and method I earnings have been shown to be of lesser quality. Thus, method II firms are penalized and sincefthey comprise only a small minority of the firms in this industry, they become lost in the confusion and controversy surrounding AFUDC. The solution may be twofold. One, since the cash implications to the firm are greatly improved by using the method II approach, the regulatory agencies should allow all firms to use method II. Secondly, there should be a concerted effort to educate investors about the method II approach being utilized and to convince the investors that method II AFUDC earnings are exactly equivalent to all other utility earnings in their cash flow implications. This approach will, at the very least, stop all the arguments about AFUDC earnings quality which as acted as a detriment to the market valuation of these companies over the last fifteen years. In summary, the primary question this study addresses is whether there is any information contained in the disclosure of AFUDC, which may be unanticipated, and therefore of use to investors in prbperly adjusting to changes in common stock values. In conclusion, there is :no information contained in the disclosure of AFUDC as stated in the financial statements. Apparently, any 197 information concerning AFUDC has already been impounded in security risk measures. The disclosure in the finan- cial statements furnishes no additional information. No information content may be the result of any of the following reasons. First, security analysts may be privy to information concerning AFUDC before the disclos- ure in the financial statements. They use this informa- tion to perfectly forecast any AFUDC and the effects AFUDC may have on market valuation. Second, AFUDC may be totally ignored by the market. The market considers AFUDC earnings worthless and therefore ignores AFUDC in the assessment of market valuation. Third, AFUDC earnings are considered the same quality as other utility earnings. Therefore, the disclosure of AFUDC and any changes in AFUDC are considered to be the same as other utility earn- ings. The results given in Chapter Five support the con- clusion that changes in earnings are significant and used by investors to reassess market value. Since this effect ‘was eliminated in examining AFUDC earnings, then differ- ences in AFUDC earnings should have no market reaction or change _in market valuation. The evidence and theoretical justification presented :in previous chapters indicates, there is some quality mn .oum xoom\uv=z "Em .>mn .wum xoom\uu§ HH H hhma mhma mbmH whma Mhma tha Hhma onma Ham» mansumm manucoz mo coau6fl>mo pumocmpm can owumm m\m .xoomlouuumxumz on OHQMB 215 Table 21 Calculations for P/E Ratios and Market-to-Book Ratios Let T1 = iranks N1 = number of firms in group 1 N2 = number of firms in group 2 N (N +1) _ 1 1 _ U — N1 - N2 + 2 T 02 = N1 - N2(N1 +N2 + 1) u 12 N ' N _ 1 2 E(U) — 2 For P/E ratios T1 = 356 U.= 11.45 + 11%12 - 356 = 205 E(U) = 11545 = 247.5 2 _ 11°45(11+46) _ Cu - 12 — 2351.2 Cu = 48.5 2 {Ely-l:g = 205‘257'5 = .876 probability .19 on 48.5 For Market-to-Book Ratios T1 = 325 U = 11.45 + llglz - 325 = 236 E(U) = 247.5‘ ou = 48.5 z = 235’317°5 = .237 probability = .40 48.5 216 Footnotes 1William L. Hays, Statistics for the Social Sciences, 2nd Ed., New York, Holt Rinehart and Winston, Inc., pp. 760-765. 2Sidney Siegel, "Non-Parametric Statistics, New York, McGrawbHill, 1956, p. 31. 3Ibid., p. 126. 217 APPENDIX B: EARNINGS CHANGES AND EFFECTS ON STOCK RETURNS A final test design is utilized here to provide some evidence on earnings changes in the electric utility in- dustry. This test provides some evidence that the changes in earnings from year to year may influence the market value. This test also provides evidence that differences in market values attributed to differences in AFUDC may in fact be due to some other variable. This design is a comparison of method I firms which have large percentage changes in earnings from year to year with method II firms. The procedure is as follows: 1) ‘Calculate the percentage change in total earnings from T to T+1 for all firms. Let ET be total earnings in year T and E be total earnings in T+1 year T+1. The percentage change is calculated E T 2) Split each of the two groups (method I and method II) into two portfolios. One portfolio contains firms with high positive percentage changes in earnings and the other portfolio contains firms with low negative percentage changes in earnings. Since there is only a limited number of method II 218 firms, portfolios were formed with only about seven firms in each portfolio. The number of firms varies slightly each year depending on the number of firms which have increases in earnings and how many have decreases. The method I port- folios are constrained to include only ten firms with the largest percentage changes, either posi- tive or negative. 3) Equate the relative risks for each set of treat- ment and control (method II) portfolios as in previous procedure. This effectively rebalances and weights the portfolios each year. 4) The statistical tests and the process that gener- ates the monthly returns is now similar to the previously discussed designs. The univariate and .multivariate forms of the t-test will be used to test for differences in mean monthly returns. Table 22 gives the results for this hypothesis. In this case, there exists a 2x1 vector of mean returns which can be tested using Hotelling's T2. The first element of the vedtor is a comparison of firms with increases in earn- ings from year to year. The second element is a comparison of firms with decreases in earnings from year to year. There are then two matched pairs which can be tested simul- taneously using Hotelling's T2. The time period, T = -6 to T = +6, which is an examination of the total time period has an Tgvalue of 74.92. The anticipation sub-period, 219 Table 22 Percentage Change in Annual Earnings: A Comparison Between Method I and Method II Firms Total Period (T = -6 to T = +6) Mean Positive % in Earnings I .0052 Positive % in Earnings II .0060 Negative % in Earnings I -.0677 Negative % in Earnings II .0073 Anticipation Period (T = -6 to Positive % in Earnings I .0146 Positive % in Earnings II .0187 Negative % in Earnings I -.0667 Negative % in Earnings II -.0152 Reaction Period (T = 0 to T Positive % in Earnings I -.0029 Positive % in Earnings II -.0049 Negative % in Earnings I -.0685 Negative % in Earnings II .0006 Critical Values of the t-Distribution, a = t,12 t,6 t,5 2. 3. 3. 681 143 365 13 Months T2- 0 t-Value Value .020 -.10 74.92 .022 .016 -10.34 .021 T = -l) .024 -.32 30.61 .021 .019 -6.49 .024 = +6) .012 .820 114.36 .019 .014 -8.56 .016 .01 220 T -6 to T = -l, and the reaction sub-period, T = 0 to T = -6, have similar results. The Tgvalues are 30.61 and 114.36, respectively. These values imply that the null hypothesis, of no differences between method I and method II, can be rejected at the .01 level. Once the null hypothesis has been rejected, further univariate tests can be performed to examine more closely why these differences do exist. The tests are two indepen- dent T-tests are each matched pair. The first test examines whether there is any difference between method I and method II firms, when those firms have increases in earnings. The t-test is performed on each of the time periods, T = -6 to T = +6, T = -6 to T = -l, T = 0 to T = +6. The t-statis— tics are -.10, -.32, and .82, respectively. These values imply the null hypothesis of no differences can be accepted at the .10 significance level. This may not be surprising given the nature of utility regulation. Increases in earn- ings for a utility are constrained on the upper limit by the rate of return an utility is allowed to earn. Any in- crease in earnings may be apparent long before the annual reports-are issued. In fact, a close examination of the commission hearings, where rate increases are approved, may allow investors to formulate and therefore anticipate any increase in revenues and earnings. If there are no vast differences amongofirms based on the percentage increase in earnings, then no difference in market returns may be ex- pected. In the context of this study, the author proposes 221 that earnings which contain a large component of inferior quality earnings (method I AFUDC earnings) may imply that the increase in earnings is actually different than re- ported. Therefore, one might expect a portfolio of method I firms to have market returns less than a portfolio of method II firms. At least for increases in earnings, this is apparently false. The market values are not significant- ly different. The second univariate test examines if there is any difference between portfolios of firms which have decreases in earnings. The results in this case are highly signifi- cant. The time period, T'= -6 to- T = +6, has a T-statis- tic of —10.34. The anticipation sub-period, T = —6 to T = -1, and the reaction sub-period, T =_0 to T = -6, have T-statistics of -6.49 and -8.56, respectively. All these values imply the null hypothesis of no differences between method I and method II firms, can be rejected at the .01 level. At first glance a researcher may be tempted to ex- plain this significance as resulting from differences in AFUDC earnings between method I and method II firms. But since all other tests have shown non—significant results, perhaps some other explanation is necessary. This design made no attempt to match firms on the per- lcentage decrease in earnings. Also, no attempt was made ‘to ensure that the average percentage change for each port- folio was equivalent. Since utilities are regulated then (any increase in earnings may not only be anticipated but 222 it might also be expected to fall within a narrow range for most utilities. The same may not be true of decreases in earnings. In fact, decreases in earnings may be both unanticipated and widely disparate. This disparity may be expected to be wider for a larger group of firms. The method II group has a very limited number of firms, con- sequently the range of percentage changes in earnings may not be as large as for method I firms. Since only method I firms with the largest decrease in earnings were chosen for that portfolio, one may expect to have more firms with larger decreases in earnings in this group than in the method II group. This may account for the significant results obtained in this case. Perhaps there may be other reasons. The point being, a design that does not properly account for other information may imply significant results when there are none. These results are not presented here as evidence opposing the large number of tests discussed earlier. Rather, they are presented here as evidence showing that significant results implying differences between the mar— ket valuation of method I and method II firms may be due to some confounding variable that has not been controlled in the design. The experimental design discussed earlier is an attempt at eliminating any bias which may influence the results. Once this has been done, one finds that there is no significant information content to the dis- closure of AFUDC earnings. Apparently, the market is 223 efficient, that is, the beta correctly anticipates and reacts to AFUDC earnings some months before the actual disclosure. 224 Footnotes lSidney, Siegel, "Non-parametric Statistics," McGraw- Hill, New York, N.Y., 1956. 2The non-parametric test was again performed for the sub-period, T = -2 to T = +2. The results are consistent with the parametric tests. In fact there may be some dis- agreement whether the data does meet the assumptions needed for a parametric test. The principal criticism may be that the mean monthly returns are not normally distributed. For this reason the author has verified with the appropriate non-parametric tests the results given here. The results are consistent with parametric results. Any results for any of the test periods which are not consistent will be discussed in this chapter. 3Siegel, p. 10. BIBLIOGRAPHY BIBLIOGRAPHY Ball, Ray. "Risk, Return and Disequilibrium - An Appli- cation to Changes in Accounting Techniques," Journal of Finance, (May 1972), p. 343-353. Ball, Ray and Brown, Phillip. "An Empirical Evaluation of Accounting Income Numbers," Journal of Accounting Research, (Autumn 1968), p. 159-178. Bassett, Christopher. "The High Cost of Nuclear Power Plants," Public Utilities Fortnightly, (April 27, 1978), p. 21-26. Beaver, William. "The Information Content of Annual Earnings Announcements," Journal of Accounting Research, (Supplement 1970), p. 62-99. Beaver, William and Morse, Dale. "What Determines Price- Earnings Ratios,?" Financial Analysts Journal, (July- August, 1978), p. 65-75. Black, Fisher. "Capital Market Equilibrium with Restricted Borrowing," Journal of Business 45 (July 1972), pp. 444-455. Black, Fisher; Jensen, Michael; and Scholes, Myron. "The Capital Asset Pricing Model: Some Empirical Tests," Studies in the Theory of Capital Markets, ed. Michael Jensen (New York: Praeger, 1972). Bock, R. Darrel and Haggard, Ernest A. "The Use of Multi- variate Analysis of Variance in Behavioral Research," in Handbook of‘Measurement and Assessment in Behavioral Sciences, ed. Dean K. Whitla (Reading, Mass.: ‘Addison- Wesley, 1968, p. 102. Bolster, Dennis R. "Should Plant Under Construction be Included in Rate Base,?" Public Utilities Fortnightly, (May 27, 1971). P. 25-29. 225 226 ‘ Brown, Lawrence D. and Rozeff, Michael S. "The Superiority of Analyst Forecasts as Measure of Expectations; Evidence from Earnings, Journal of Finance, March 1978. pp. 1-16. Brown, Phillip and Kennelly, J. W. "The Information Con- tent of Quarterly Earnings: An Extension and Some Further Evidence," Journal of Business, (July 1972), p. 403-415. Brown, Stewart. "Earnings Changes, Stock Prices, and Market Efficiency," The Journal of Finance, (March, 1978): P. 17-28. Brown, William R. "Rate Cases and Utility Financing," Public Utilities Fortnightly, (February 15, 1973), p. 39-42. Burkhardt, Daniel A. and Viren, Michael A. "Investor Criteria for Valuing Utility Common Stocks," Public Utilities Fortnightly, (July 20, 1978), p. 27-35. Coughlan, Paul B. "Allowance for Funds in Construction Accounting Stepchild and Regulatory Football," Public Utilities Fortnightly, (November 4, 1976), p. 26-33. Dahlenburg, Lyle M. "AFUDC-Calculation and Disclosure Problems," Public Utilities Fortnightly, (January 17, 1974), p. 17-23. Eskew, R. F. and Wright, W. F. "An Empirical Analysis of Differential Capital Market Reactions to Extra- ordinary Accounting Items," Journal of Finance, (May, 1976), p. 651-673. Fama, Eugene et al. "The Adjustment of Stock Prices to New Information," International Economic Review, (February, 1969), p. 1-21. Fama, Eugene F. "The Behavior of Stock Prices," Journal of Business 28 (January 1965), pp. 34-105. Fama, Eugene F. and MacBeth, James. "Risk, Return, and Equilibrium: Empirical Tests," Journal of Political Economy 81 (May/June 1973), pp. 607-636. Fama, Eugene F. and Miller, Merton H. The Theorygf Finance (Hinsdale, Illinois, Dryden Press, 1972), Chapter 7. 227 Financial Accounting Standards Board. Discussion Memo- randum, Conceptual Framework for Financ1al Accounting and Reporting: Elements of Financial Statements and Their Measurement, December 2, 1976. Finn, Jeremy D. General Model for Multivariate Analysis. New York: Holt, Rhinehart and Winston, Inc., 1974. Firth, Michael. "The Impact of Earnings Announcements on the Share Price Behavior of Similar Type Firms," The Economic Journal, (June, 1976), p. 296-306. Fitzpatrick, Dennis B. and Stitzel, Thomas E. "Capital- izing and Allowance for Funds Used During Construction: The Impact on Earnings Quality," Public Utilities Fortnightly, (January 19, 1978), p. 18-22. Foster and Rodney. Public Utility Accounting, Prentice- Hall, Inc., 1951. Frazer, Robert E. and Ranson, Richard C. "Is Interest During Construction 'Funny Money,'?" Public Utilities Fortnightly, (December 21, 1972), p. 20-27. Glassman, Gerald J. "Fair Return to Equity," Public Utilities Fortnightly, (May 25, 1978), p. 11-15. Gonedes, N. J. "Capital Market Equilibrium and Annual Accounting Numbers: Empirical Evidence," Journal of Accounting Research, (Spring 1974):26-62. Gonedes, N. J. "Corporate Signaling, External Accounting, and Capital Market Equilibrium: Evidence on Dividends, Income, and Extraordinary Items," Journal of Accounting Research (Spring l978):26-79. Gonedes, N. J. "Risk, Information, and the Effects of Special Accounting Items on Capital Market Equi- librium," Journal of Accounting Research (Autumn 1975b):220-56. Gonedes, N. J. and Dopuch, D. "Capital Market Equilibrium, Information-Production, and Selected Accounting Tech- niques: Theoretical Framework and Review of Empirical Work," Studies on Financial Accounting Objectives: 1974. Supplement to Journal of Accounting Research 12:48-169. Gordon, Myron J. *The Cost of Capital to Public Utility. East Lansing: Division of Research, Graduate School of Business Administration, Michigan State University, 1974. 228 Green, Paul E. Analyzlng Multiyariate Data. Hinsdale, Illinois: Dryden Press, 1978. Hagerman, Robert L. "Finance Theory in Rate Hearings," Journal of Financial Management, (Spring, 1976), p. 18-21. Hamada, Robert S. "The Effect of the Firm's Capital Structure on the Systematic Risk of Common Stocks," Journal of Finance, Vol. 27, May 1972. Harrison, Walter Thomas. Unpublished dissertation, "The Information Content of Accounting Changes," Michigan State University, 1976, pp. 13-14. Haugen, R. A.; Stroyny, A. L.; and Wichern, D. W. "Rate Regulation, Capital Structure, and the Sharing of Interest Rate Risk in the Electric Utility Industry," Journal of Finance, (June, 1978), p. 707*721. Hays, William L. Statistics for the Social Sciences, 2nd edition, New York: Holt, Rinehart and Winston, Inc., 1973. Howard, Hayden R. Risk andtgegulated Firms, East Lansing: Division of Research, Graduate School of Business Administration, Michigan State University, 1973. Hyde, W. Truslow. "It Could Not Happen Here-But It Did," Public Utilities Fortnightly, (June 6, 1974), p. 47-50. Hyde Jr., W. Truslow. "A Compromise on Construction Work in Progress Would Benefit Consumers and Investors," Public Utilities Fortnightly, (August 18, 1977), p. 15-18. Hyde Jr., W. Truslow. "Overcoming Regulatory Lag," Public Utilities Fortnightly, (February 27, 1975), p. 34-40. Hyde Jr., W. Truslow. "Rate of Return - An Art or a science," Public Utilities Fortnightly, (December 18, 1975), p. 27¥31. Hyman, Leonard S. "Market to Book Ratio: Statistical Confirmation or Aberration,?" Public Utilities Fort- nightly, (December 19, 1974), p. 27-32. Hyman, Leonard S. "Rate Cases in 1970-74: A Quantitative Examinationf" Public Utilities Fortnightly, (November 6, 1975) I p. 22-260 229 Hyman, Leonard S. "Utility Stocks in 1967-72: A Tale of Noe," Public Utilities Fortnightly, (February 28, 1974), p. 23-28. Jaquette, John J. "What is the Electric Utility Stock Market Saying," Public Utilities Fortnightly, (April 8, 1976), p. 17-19. . Jensen, Michael. "Capital Markets: Theory and Evidence," Bell Journal of Econgmics and Management Science 3 (Autumn 1972), pp. 357-398. Johnson, Johnny R. "The Allowance for Funds Used During Construction: Market Reaction to Current Accounting Practice," unpublished dissertation, Virginia Poly- technic Institute and State University, 1976. Johnson, Johnny R. "The Income Quality of Allowance for Funds Used During Construction," Public Utilities Fortnightly, (June 9, 1977), p. 32-36. Johnson, Johnny R. "Losses on Investments in Construction Work in Progress," Public Utilities Fortnightly, (September 14, 1978), p. 20-24. Joskow, Paul L. "Determination of the Allowed Rate of Return in a Formal Regulatory Hearing," The Bell Journal of Economics and Management Science, (Spring, 1972), p. 632-644. Judy, Paul R. "Current Market Valuation and Utility Securities," Public Utilities Fortnightly, (November 22, 1973). p. 24-28. Kaplan, Robert A. and Roll, Richard. "Investor Evaluation of Accounting Information - Some Empirical Evidence," Journal of Business Volume 45 (April, 1972), p. 225- 257. Kaplan, Robert S. "The Information Content of Financial Accounting Numbers: A Survey of Empirical Evidence," in A. Abdel-Khalik, and T. Keller, eds., The Impact of Accounting Research oniPractice andipisclosure, Durham: (Duke University Press, North Carolina, 1978, p. 135-173. King, B. F. "Market and Industry Factors in Stock Price Behavior," Journal of Business, (Supplement 1966), p. 139-167. LaBlanc, Robert and Luftig, Mark D. "What are the New Market Prospects for Raising Utility Capital," Public Utilities Fortnightly, (February 26, 1976), p. 26-30. 230 Leatson, J. Walter. "Capitalism's Greatest Test - The Electric Utilities," Public Utilities Fortnightly, Leighton, Eric. "Accounting and the Determination of Revenue Requirements," Public Utilities Fortnightly, (September 11, 1975), p. 17-21. Leland, Hayne E. "Regulation of Natural Monopolies and the Fair Rate of Return," The Bell Journal of Economlcs and Management Science, (Spring, 1974), p. 3-15. Levy, Robert E. "Fair Return on Equity for Public Utilities," Business Economics, (Spring, 1978), p. 46-57. Linhart, Peter B.; Lebowitz, Joel L.; and Sinder, Frank W. "The Choice Between Capitalizing and Expensing Under Rate Regulation," The Bell Jourgal of Economlgs and Management Science, (Autumn, 1974), p. 406-419. Litke, Arthur L. "Allowance for Funds Used During Con- struction," Public Utilities Fortnightly, (September 28, 1972), p. 19-22. Manus, Peter C. and Phillips, Charles F. "Earnings Erosion During Inflation," Public Utilities Fortnightly, Marx, Thomas G. "Market-to-Book Return on Equity Corre- lation," Public Utilities Fortnightly, (December 4, 1975), p. 28-31. Mattutat, Edward C. "A Pragmatic Approach to Construction Work in Progress," Public Utilities Fortnightly, (March 3, 1977), p. 31-37: McDiarmid, Fergus J. "Interest Coverage and Bon Ratings," Public Utilities Fortnightly, (May 11, 1972), p. 19-23. McDiarmid, Fergus J. "Return on Equity: The Key to Financial Strength and Cost of Capital," Public Utilities Fortnightly, (February 26, 1976), p. 21-27. McDiarmid, Fergus J. "Why Public Utility Stocks are in the Doghouse," Public Utilities Fortnlghtly, (September 28, 1972), p. 23-27. Melicher, Ronald W. "Risk and Return in the Electric Utility and Natural Gas Industries," Michigan State University Business Topics, (Spring, 1975), p. 48-54. 231 Melicher, Ronald W. and Rush, David F. "Systematic Risk, Financial Data, and Bond Rating Relationships in a Regulated Environment," Journal of Finance, (May, 1974): p. 537-544. Meyers, Steward C. "The Application of Finance Theory to Public Utility Rate Cases," Bell Journal of Economics and Management Science, (Spring, 1972), ’—‘-7 p. 58-97. Morris, Everett L. "Capitalization of Interest on Con- struction: Time for Reappraisal," Public Utilities Fortnightly, (March 4, 1971), p. 19-29. Nash Jr., Alexander A. "Rate Relief and the Construction Funds Allowance - A Big Dollar Question," Public Utilities Fortnightly, (July 6, 1978), p. 15-18. Neter, John and Nasserman, William. Applied Llnear Statistical Models, Homewood, IlIInOIs: Richard D. Irwin, Inc., 1974. Phelps, George E. "A Different Approach to the Fair Rate of Return," Public Utilities Fortnightly, (August 12, 1976), p. 26-31. Phelps, George E. "That Bothersome Construction Interest Allowance," Public Utilities Fortnightly, (January 31, 1974), p. 21-27. Pomerantz, Lawrence S. and Suelflow, James E. Allowance for Funds Used During Construction: Theory and Application, East Lansing, Michigan, Division of Research, Michigan State University, 1975. Public Utilities Fortnightly. "Progress of Regulation, Trends and Topics: Construction Work in Progress - Inclusion," Public Utilities Fortnightly, (March 13, 1975), p. 50-52. Public Utilities Fortnightly. "Progress of Regulation, Trends and Topics: Construction Work in Progress - Exclusion," Pghlic Utilities Fortnightly, (March 27, 1975), p. 46-47. Robichek, Alexander A. "Regulation and Modern Finance Theory," The Journal of Finance, (June, 1978), p. 693-705. a Roseman, Herman G. "Utility Financing Problems and National Energy Policy," Public Utilities Fortnightl , (September 12, 1974), p. 19-30. 232 Rubin, Leon C. "Get the Electric Utility Companies Out of the Construction Business," Public Utilities Fortnightly. (June 5, 1975): P. 35-36. Seligson, Carl H. "Electric Utilities Need More Equity," Public Utilities Fortnightly, (March 16, 1978), p. 17-21. Severiens, Jarolus T. "The Price-Earnings Behavior of Electric Utilities: Some Determinants," Public Utilities Fortnightly, (December 7, 1978), p. 27-32. Sharpe, William. "A Simplified Model for Portfolio Analysis," Management Science, (January, 1963), p. 277-293. Shaw, Donald H. "Erosion of Utility Common Stock Divi- dends," Public Utilities Fortnightl , (May 23, 1974) I p. 26-300 Siegel, Sidney. "Non-Parametric Statistics," McGraw- Hill, New York, New York, 1956. Smolen, Gerald E. and Melnyk, Z. Lew. "Rate Filings and Utility Stock Prices," Puplic Utilities Fortnightly, (September 15, 1977), p. 25-28. Stancill Jr., James M. "Cost of Service Indexing of the Rate of Return," Public Utilities Fortnightly, (August 18, 1977), p. 11-14. Stich, Robert S. "An Additional Standard for Measuring Common Equity Costs," Public Utilities Fortnlghtly, (February 26, 1976), p. 34-39. Suelflow, James E. Public Utility Accounting: Theoty and Application, East Lansing, Michigan: Division of Research, Michigan State University, 1973. Sunder, Shyam. "Relationship Between Accounting Changes and Stock Prices: Problems of Measurement and Some Empirical Evidence," Empirical Research in Accounting: Selected Studies 1973, Supplement to Journal of Accounting Research 11 (1973):l-45. "Survey on Construction Work in Progress," Edison Electric Institute, 1976. Tatham, Charles.‘ "Other Rating Factors: Interest During Construction and Price-Earnings Ratios," Public Utilities Fortnightly, (September 27, 1973), p. 32-35. 233 Timm, Neil H. Multivariate Analysis with Applications in Education and Psychology, Wadsworth Publishing Company, Belmont, California, 1975, p. 226-229. Trout, Robert R. "A Rationale for Preferring Construction Work in Progress in the Rate Base," Public Utilities Fortnightly, (May 10, 1979), p. 22-26. West, David. "Adjusting Rates to Cost of Capital," Public Utilities Fortnightly, (September 15, 1977), p. 19-23. - West, David A. and Eubank Jr., Arthur A. "An Automatic Cost of Capital Adjustment Model for Regulating Public Utilities," Journal of Financial Management, (Spring, 1976), p. 23-31. Williams, Edward E. and Seneca, J. J. "Utility Dividend Policy, Growth, and Retention Rates (1958-1967)," Public Utilities Fortnightly, (March 15, 1973), p. 23-27. NICHIGAN srms UNIV. LIBRARIES WMINI”HIIHIIWII"MW"WI“INIWWIHIWI 31293104499607