J. Account. Public Policy

J. Account. Public Policy 28 (2009) 251 261 Contents lists available at ScienceDirect J. Account. Public Policy journal homepage: www.elsevier.com/locate/jaccpubpol A note on relevance of mark-to-market s of energy contracts under EITF Issue No. 98-10 Li Li Eng a, *, Shahrokh Saudagaran b, Sora Yoon c a Missouri University of Science and Technology, Department of Business and Information Technology, 102 Fulton Hall, Rolla, MO 65409, USA b University of Washington Tacoma, Milgard School of Business, 1900 Commerce Street, Tacoma, WA 98402, USA c Ajou University, Suwon, Kyonggi, Korea article Keywords: Mark-to-market Fair s Book s Energy contracts info abstract This paper examines whether marked-to-market s of energy trading assets and liabilities of companies that enter into energy contracts are related to market of equity. The Emerging Issues Task Force of the Financial Accounting Standards Board ruled in November 2002 to ban the use of mark-to-market accounting for energy contracts out of concern that fair s can be easily inflated. We find that the excess of fair over original of energy trading assets and energy trading liabilities is not relevant for valuation. It may be inferred that fair s which are subject to management estimates and not verifiable are poor signals of worth and performance (Watts, R., 2003. Conservatism in accounting Part I: Explanations and implications. Accounting Horizons 17, 207 221). Ó 2009 Elsevier Inc. All rights reserved. 1. Introduction The debate over fair s or historical costs has become increasingly intense. In June 2005, the Securities Exchange Commission (SEC) endorsed fair accounting (The Economist, July 30th 2005, p. 14, 65). The Financial Accounting Standards Board (FASB) has also issued guidelines for applying fair accounting (example, FAS 157 on Fair Value Measurement (FASB, 2007a), and FAS 159 on The Fair Value Option for Financial Assets and Financial Liabilities (FASB, 2007b)). Proponents of fair s argue that they are more relevant than historical costs. However, critics of fair s are concerned about the reliability of fair measures as they are more heavily based on estimates (The * Corresponding author. Tel.: +1 573 341 4594; fax: +1 573 341 4812. E-mail addresses: engl@mst.edu (L.L. Eng), shahrokh@u.washington.edu (S. Saudagaran), yoonsora@ajou.ac.kr (S. Yoon). 0278-4254/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.jaccpubpol.2009.04.004

252 L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 Economist, July 30th 2005, pp. 65 66). Empirical evidence indicates that management estimates are in error or misstated (Noland et al., 1998; Beaver and McNichols, 2001; Lehavy, 2002), subject to manipulation (Bergstresser et al., 2006), and do not improve the quality of financial information (Lev et al., 2009). This paper examines whether marked-to-market s of energy trading assets and liabilities of companies that enter into energy contracts are relevant for establishing market of equity. Energy trading companies enter into contracts for the purchase and sale of energy commodities. In addition, these energy contracts have been entered into for the purpose of speculating on market movements and for trading purposes. In October 1998, the Emerging Issues Task Force (EITF) of the Financial Accounting Standards Board (FASB) released EITF Issue No. 98-10 (FASB, 2000). The Task Force reached a consensus that energy trading contracts should be marked-to-market (that is, measured at fair as of the balance sheet date) with the gains and losses included in earnings and separately disclosed in the financial statements. EITF Issue No. 00-17 reiterated that the estimate of fair should be based on the best information available in the circumstances. Where quoted market prices are not available for these contracts, companies must use other techniques to estimate fair. The discussion note of EITF Issue No. 00-17 states, When available, current market transactions provide the basis for estimating subsequent changes in fair. Valuation models, including option pricing models, should be used only when market transactions are not available to evidence fair s. When valuation models are used, the Task Force noted that the best information available would consider, but is not limited to, recent spot prices and forward prices, and for option pricing models, the volatility implied by recent transactions, when available, or the historical volatility of the commodities and/or services underlying the contract. 1 A series of events led to EITF Issue No. 98-10 being rescinded in November 2002. EITF Issue No. 02-3 (FASB, 2002) ruled to ban the use of mark-to-market accounting on contracts to deliver and store energy entered into after October 25, 2002. Under mark-to-market accounting, companies can book the estimated profits of a long-term energy contract, which could take years to settle, immediately. Analysts have condemned the practice, saying the of those contracts could easily be inflated with generous assumptions about prices and other market conditions. Enron Corp., Williams and others have been accused of using unrealistic assumptions to boost the mark-to-market of their trading transactions...the Securities and Exchange Commission pushed for the change following the collapse of Enron, once the world s biggest energy trader. The action is an effort to restore credibility to an industry plagued by a series of trading and accounting scandals...the FASB is trying to end an extremely embarrassing period for itself and the accounting profession. Ray (2002), Tulsa World, November 6, 2002. E1, E2. The implication of the EITF ruling to ban mark-to-market accounting is that mark-to-market accounting is subject to manipulation, 2 and therefore, may not be relevant. That is, fair of such assets and liabilities may not be associated with market of the firm. In this paper, we examine this empirical issue of whether fair s of energy assets and liabilities are associated with market valuation. This paper contributes to research on understanding alternative accounting numbers, and an understanding of accounting of energy trading firms. We contribute to extant studies on market valuation of disclosures such as SFAS No. 33 on current cost accounting (for example, Beaver and Landsman, 1983; Beaver and Ryan, 1985, 1987; Bublitz et al., 1985; Haw and Lustgarten, 1988; Hopwood and Schaefer, 1989; Lobo and Song, 1989; Murdoch, 1986), SFAS No. 107 on fair of financial instruments (for example, Barth et al., 1996; Eccher et al., 1996; Nelson, 1996), market valuation of recognized fair s of derivative financial instruments relative to disclosed s (Ahmed et al., 2006), and risk relevance of 1 Industry practices indicate that energy traders contracts at quoted prices in active (liquid) markets. Where markets are less active (illiquid), valuation techniques such as Black Scholes and Monte Carlo simulations are used. Key inputs in the valuation process include determining volatility of commodity prices and estimating future cash flows, which are subject to a certain degree of uncertainty. 2 Example, Williams Cos. Inc. lost $814.5 million, or $1.59 a share in the first quarter [of 2003], chiefly due to the accounting change which bars companies from booking the estimated profits of certain long-term energy contracts, and requires energy companies to book those profits as they are accrued. Ray (2003), Tulsa World, May 14, 2003. E1, E8.

L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 253 full-fair- income of commercial banks relative to net income and comprehensive income (Hodder et al., 2006). EITF Issue No. 02-3 to rescind EITF Issue No. 98-10 raises the empirical question of whether fair s of energy contracts are -relevant in establishing share prices. To examine whether mark-to-market energy assets and liabilities are -relevant, we regress market of equity on net income, original s of net energy assets and book of net non-trading assets, and the excess of fair over the original of energy assets and liabilities. We find that market valuation is positively related to original s of energy trading assets and book s of nontrading assets. We find that the excess of fair over original of energy trading assets and liabilities is not -relevant. Similar to Nelson (1996) and Eccher et al. (1996), we conclude that fair s of marked-to-market assets and liabilities are generally not relevant. Thus, it appears that the market does not place a higher (lower) on the firms when fair of energy trading assets (liabilities) exceeds its original. The EITF was concerned that fair s of energy contracts can be easily inflated (manipulated), and mark-to-market accounting may lead to the market being misled. The evidence indicates that unverifiable valuations are actually not relevant for valuation (Watts, 2003, p. 219). We conjecture that the market may have shared concerns of the EITF regarding fair s of energy assets and liabilities of the sample firms in our study. Thus, these fair s are not incorporated into market valuation of the sample firms. The paper is organized as follows. Section 2 reviews prior literature on valuation of current cost accounting and fair disclosures of bank assets and liabilities. Section 3 describes the hypotheses and methodology employed to test -relevance of fair accounting. Section 4 describes sample selection and the data. The results are contained in Section 5. Finally, Section 6 concludes with a summary of the paper. 2. Prior research SFAS No. 33 requires disclosures of current cost operating income. Beaver and Ryan (1985) examine the ability of various earnings measures derived from SFAS 33 data to explain stock returns. They find that no one SFAS 33 variable adds significant explanatory power consistently to the earnings-returns model. Contrasting results are obtained in Bublitz et al. (1985), Murdoch (1986) and Lobo and Song (1989). Bublitz et al. (1985) find significant explanatory power for SFAS 33 data above that provided by historical earnings. Murdoch (1986) shows that purchasing power returns on equity do show incremental information content but constant dollar, current cost, and net holding returns do not exhibit information content incremental to historical cost returns in explaining security price changes. The findings in Lobo and Song (1989) also indicate that SFAS 33 income measures have incremental information over historical cost income and its cash and accrual component for a subset of the industries examined. Bernard and Ruland (1987) show evidence of incremental information content of current cost income where the correlation between unexpected historical cost income and unexpected current cost income is lowest. The tests in Hopwood and Schaefer (1989) suggest the existence of differential responses across firms depending on the firms ability to respond to cost increases. More recent studies examine the fair disclosures of banks assets and liabilities. Barth et al. (1996) examine the -relevance of banks fair disclosures under SFAS No. 107. They find that fair estimates of bank assets and liabilities disclosed under SFAS No. 107 provide significant explanatory power for bank share prices beyond that provided by related book s. Nelson (1996) also examines the association between the market of banks common equity and fair estimated disclosures under SFAS No. 107. Contrary to Barth et al. (1996), Nelson finds that only the reported fair s of investment securities have incremental explanatory power relative to book. There is no evidence of incremental explanatory power for the fair disclosures of loans, deposits, long-term debt or net off-balance sheet financial instruments. Eccher et al. (1996) also obtain evidence that fair disclosures for financial instruments other than securities are -relevant only in limited settings.

254 L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 Ahmed et al. (2006) examine investor valuation of derivative financial instruments depending on whether the fair of these instruments is recognized or disclosed. They find that recognized derivatives are -relevant but disclosed derivatives are not. Hodder et al. (2006) investigate the risk relevance of the volatility of net income, comprehensive income and full-fair- income of a sample of US commercial banks. They find that full-fair- income has more volatility relative to comprehensive income and net income, and this incremental volatility is reflected in capital-market pricing of that risk. This paper extends prior research on information content of current cost income and fair disclosures of financial instruments. Specifically, we examine the -relevance of fair disclosures of energy trading assets and liabilities under EITF Issue No. 98-10. 3. Hypotheses and methodology This paper examines whether fair s of energy trading assets and liabilities are -relevant. We run a model with market of equity regressed on net book of non-trading assets and liabilities, net original of energy trading assets and liabilities, excess of fair over original of energy trading assets and liabilities, and net income. 3 This model is derived from the Ohlson (1995) model. 4 MV it ¼ a 0 þ a 1 NI it þ a 2 BV it þ e it ð1þ where MV is market of equity, NI is net income and BV is book of equity. From (1), MV it ¼ a 0 þ a 1 NI it þ a 2 ðbva it BVL it Þþe it MV it ¼ a 0 þ a 1 NI it þ b 1 ðbvnta it BVNTL it Þþb 2 ðbvta it BVTL it Þþe it MV it ¼ a 0 þ a 1 NI it þ b 1 ðbvnta it BVNTL it Þþb 3 ðovta it OVTL it Þþb 4 ðeta it ETL it Þþe it MV it ¼ a 0 þ a 1 NI it þ b 1 NetNTA it þ b 3 NetTA it þ b 5 ETA it þ b 6 ETL it þ e it where BVA is book of total assets, BVL is book of total liabilities, BVNTA is book of non-trading assets, BVNTL is book of non-trading liabilities, BVTA is book of trading assets, BVTL is book of trading liabilities, OVTA is original of trading assets at time of inception, OVTL is original of trading liabilities at time of inception, NetNTA is book of non-trading assets less book of non-trading liabilities, NetTA is original of trading assets less original of trading liabilities, ETA is excess of fair over original of energy trading assets, and ETL is excess of fair over original of energy trading liabilities. The coefficient, b 3, measures the relevance of the original s of net trading assets. b 5 and b 6 measure the relevance of change in fair s over original of energy trading assets (ETA) and liabilities (ETL), respectively. Our predictions of the coefficients are limited to their signs with the null hypothesis that the coefficients on ETA and ETL equal zero. That is, if fair s are being used by the market, then excess of fair over original of energy trading assets (liabilities) is expected to be positively (negatively) associated with market of equity. Hence, our hypotheses are: ð2þ H1: The excess of fair over original of energy trading assets is positively associated with market of equity. 3 Original is the fair of the energy contracts at the time of inception of the contracts. 4 Empirical research applying this model includes Harris and Kemsley (1999) and Hanlon et al. (2003). Barth et al. (1996), Nelson (1996) and Eccher et al. (1996) use alternate models to examine the -relevance of banks fair disclosures under SFAS No. 107. They model the excess of market of equity over book of equity as a linear function of the cumulative excess of fair over book of each asset and liability. In the case of SFAS No. 107 disclosures, loans, deposits, long-term debt, etc. are not marked-to-market on the balance sheet; in the case of EITF Issue No. 98 10, energy trading assets and liabilities are marked-tomarket on the balance sheet. We consider the Ohlson (1995) model more appropriate to examine the issue of relevance of fair s of energy trading assets and liabilities that are recognized on the balance sheet than the models in Barth et al. (1996), Nelson (1996) and Eccher et al. (1996).

L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 255 H2: The excess of fair over original of energy trading liabilities is negatively associated with market of equity. The focus of this paper is on the -relevance of fair s of energy assets and liabilities. Appendix A contains an excerpt from Cleco Power LLC, year 2002 10-K Annual Financials that illustrates the footnote disclosure of the original amounts and fair s of energy market positions. As at December 31, 2002, Cleco Power LLC had energy assets with a original of $20.793 million and fair of $24.457 million, and energy liabilities with a original of $32.652 million and fair of $37.239 million. Cleco Power LLC discloses that the fair s of energy market positions may be based on market prices which are verifiable, and management estimates which are not verifiable. 5 Fair s of energy trading assets and liabilities may be largely unverifiable. Asset measures that involve unverifiable estimates are open to considerable manipulation (Holthausen and Watts, 2001). Unverifiable fair measures are not efficient signals of worth and performance (Watts, 2003). The empirical issue is whether fair s are informative to the market, or whether they are inefficient signals. If mark-to-market accounting results in accounting measures that are closer to the underlying fundamental, these fair s will be -relevant. Alternatively, if management manipulates the estimates of fair s, these fair s will not be -relevant. 4. Sample and data 4.1. Sample selection We obtain a list of the top ten energy trading firms through a keyword search of energy trading firms in Google. We then obtain companies with the same 4-digit SIC code in Compustat. The SIC codes are 4911 (Electric Services), 4922 (Natural Gas Transmission), 4923 (Natural Gas Transmission and Distribution), 4931 (Electric and Other Services Combined), 4991 (Cogeneration-SM Power Producer) and 5172 (Petroleum and Petroleum Products Wholesales, except Bulk Stations). We then obtain the annual reports of the companies for the period 1995 2001 from the companies websites or SEC-Edgar. 6 The sample consists of 156 firm-years for which we are able to obtain annual reports. The number of observations obtained for each year is as follows: 2001 44 2000 26 1999 24 1998 21 1997 15 1996 14 1995 12 From the annual reports, we obtain data for shares outstanding, book of equity, fair of energy trading assets, original of energy trading assets, fair of energy trading liabilities, original of energy trading liabilities, book of total assets and book of total liabilities. 5 In reviewing the footnotes to the 2001 annual reports of 64 sample firms, we note that none of the firms based fair solely on market price, 11 percent of the firms based fair s on management estimates, and 70 percent on both market price and management estimates; we were not able to ascertain the valuation basis for 19 percent of the firms from their disclosures. 6 Discussions with Thomas Sell and J. Kevin Vann of Williams Company indicate that fair accounting of energy trading contracts may have begun as early as 1991. Since mark-to-market accounting for energy contracts was promoted through EITF Issue No. 98 10, we decided on a sample period from 1995 through 2001 for two reasons. First, we have three years before and three years after 1998 when the practice was officially sanctioned. Second, we get fewer annual reports as we move back through the years.

256 L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 Table 1 Descriptive statistics of variables. Variable N Mean Median Std Dev Minimum Maximum BV ($ million) 93 3210.62 2499.32 2635.00 133.29 12923.00 MV ($ million) 93 7506.24 4330.26 9779.00 344.76 53410.89 MV BV ($ million) 93 4295.62 1070.71 7957.00 2712.44 43840.89 NetNTA ($ million) 93 6937.48 5275.57 5661.00 706.66 24921.00 NetTA ($ million) 93 3726.85 2776.25 3596.00 20110.00 2204.00 ETA ($ million) 93 147.42 5.40 1001.00 6432.00 1122.00 ETL ($ million) 93 132.01 20.00 919.10 5635.00 933.60 NI ($ million) 93 314.236 211.000 419.873 1922.000 1898.000 BV = book of common equity; MV = market of common equity; MV BV = market less book of common equity; NetNTA = book of non-trading assets less book of non-trading liabilities; NetTA = original of trading assets less original of trading liabilities; ETA = fair less original of energy trading assets; ETL = fair less original of energy trading liabilities; and NI = net income. 5. Descriptive statistics Table 1 presents the descriptive statistics of the sample variables. The average book of equity of the sample firms is $3210.62 million (median = $2499.32 million). The mean market of equity is $7506.24 million (median = $4330.26 million). Thus, the average market less book of equity is $4295.62 million (median = $1070.71 million). The average book of net non-trading assets is $6937.48 million (median = $5275.57 million). On average, the firms in the sample report original of net trading assets equal $3726.85 (median = $2776.25 million). That is, energy trading liabilities are greater than energy trading assets for our sample during the sample period. We find that the average fair of energy trading assets is less than its original with a mean difference of $147.42 million (median = $5.4 million). The fair of energy trading liabilities is also less than its original ; the mean difference is $132.01 million (median = $20.0 million). Sample firms report average net income of $314.236 million (median = $211.0 million). Table 2 Pearson Correlation Analysis of Variables. (Number of observations = 93). MV NetNTA NetTA ETA ETL NI MV 1.000 0.720 0.576 0.052 0.059 0.539 (0.0) (<0.0001) (<0.0001) (0.618) (0.575) (<0.0001) NetNTA 1.000 0.934 0.066 0.155 0.549 (0.0) (<0.0001) (0.531) (0.139) (<0.0001) NetTA 1.000 0.071 0.067 0.380 (0.0) (0.499) (0.524) (0.000) ETA 1.000 0.890 0.233 (0.0) (<0.0001) (0.024) ETL 1.000 0.050 (0.0) (0.638) NI 1.000 (0.0) MV = market of common equity; NetNTA = book of non-trading assets less book of non-trading liabilities; NetTA = original of trading assets less original of trading liabilities; ETA = fair less original of energy trading assets; ETL = fair less original of energy trading liabilities; and NI = net income.

L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 257 6. Results 6.1. Correlation analysis Table 2 presents the correlation analysis results. The correlation analysis shows that market is positively correlated with book of net non-trading assets (q = 0.720) but negatively correlated with original of net trading assets (q = 0.576). Note that original of net trading assets is negative in Table 1. Market is not highly correlated with excess of fair over original of energy trading assets (q = 0.052) and excess of fair over original of energy trading liabilities (q = 0.059). The univariate relations do not support the alternative hypotheses that fair of energy trading assets (liabilities) is positively (negatively) associated with market of equity. Market has a positive correlation with net income (q = 0.539). Thus, market is correlated with book and net income but not fair. Other variables that have significant correlations are net non-trading assets with net trading assets and net income, net trading assets and net income, excess of fair over original of energy trading assets and net income. 6.2. Regression analysis Table 3 presents the results of regression model in Eq. (2). We regress market on net income, net book of non trading assets and liabilities, net original of energy trading assets and liabilities, and excess of fair over original of energy trading assets and liabilities. Column 2 presents the results for the pooled sample for the period 1995 2002. The coefficient on net income (a 1 ) is not significant. Current period fair gains and losses on energy trading contracts are included in net income, and have not been separately measured in our model. To the extent that fair gains and losses are of lower relevance, including these amounts in net income would bias the results to finding lower significance for net income. The variables significant in explaining market valuation are book s of net non-trading assets (b 1 = 2.484, t-statistic = 4.81) and net energy trading assets (b 3= 2.158, t-statistic = 3.24). b 5 and b 6, the coefficients on excess of fair over original of energy trading assets and energy trading liabilities, are not significantly different than zero. The evidence indicates that market valuation does not reflect the excess of fair s reported of energy trading assets and liabilities over the original s. The relation between b 1, b 3 and b 5 (b 6 ) may reflect the proportion of prepaid contracts in a company s energy trading operations. b 1 is the market s valuation coefficient on historical cost. Assume that the market s all historical cost book s at b 1. The market s fair gains at b 5 (assumed and appears to be equal to b 6 ). b 3 would then roughly be equal to cb 1 +(1 c)b 5, where c is the proportion of prepaid contracts. Our slope estimate for NetTA is consistent with this intuition. 7 Column 3 presents the regression results for the sample with outliers removed from the sample (defined as observations with an absolute for the studentized residual greater than three or Cook s distance measure greater than one). The results remain qualitatively similar. Column 4 presents the regression results with the variables scaled by number of shares. Market per share (price) is positively associated with net income. The coefficients on ETA per share and ETL per share are not significantly different than zero. That is, share price does not reflect the change in fair of energy trading assets and liabilities. The adjusted R 2 is lower for the regression model when the variables are scaled by number of shares. The adjusted R 2 reported here is after we clustered standard errors at the firm level. 8 To summarize, the market s of our sample energy trading firms do not reflect the excess of fair over original of energy trading assets and liabilities. The findings that fair s of energy trading assets and liabilities are not d by the market are similar to the findings in some prior 7 We thank a reviewer for suggesting this interpretation. 8 We use Proc Surveyreg in SAS for this regression. We also compare the correlation matrices of the level variables versus the per-share variables. Some correlations are lower or not significant for the per-share variables relative to the levels variables, example: book per share of non-trading assets and trading assets are not significantly correlated with share price. This may be a reason for the adjusted R 2 of the per-share model being lower than that of the levels model.

258 L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 Table 3 Regression of Market Value Value on Net Income, Book Value of Assets and Liabilities, and Excess of Fair Value Over Original of Energy Trading Assets and Liabilities (t-statistics are in parentheses). MV it ¼ a 0 þ a 1 NI i t þ b 1 NetNTA it þ b 3 NetTA it þ b 5 ETA it þ b 6 ETL it þ e it ð2þ Pooled sample (1995 2002) Pooled sample with outliers removed (1995 2002) Intercept 2005.637 * 1012.576 25.493 *** ( 2.46) ( 1.49) (6.84) NI 1.576 1.364 2.550 *** (0.92) (0.85) (3.52) NetNTA 2.484 *** 1.874 *** 0.111 (4.81) (4.15) (0.50) NetTA 2.158 *** 1.486 *** 0.167 (3.24) (2.41) (0.54) ETA 0.604 0.603 0.368 (0.33) ( 0.45) ( 1.14) ETL 0.627 1.200 0.153 (0.37) (0.90) (0.64) Adj R 2 0.5861 0.6514 0.1291 Number of observations 93 91 91 Pooled sample with outliers removed (1995 2002) (scaled by number of shares) MV = market of equity; NI = net income; NetNTA = book of non-trading assets less book of non-trading liabilities; NetTA = original of trading assets less original of trading liabilities; ETA = excess of fair over original of energy trading assets; and ETL = excess of fair over original of energy trading liabilities. Note: the t-statistics are based on heteroscedasticity-consistent estimates of the standard errors (White 1980). research on the explanatory power of SFAS No. 33 disclosures of current cost operating income (for example, Beaver and Ryan, 1985) and -relevance of banks fair disclosures under SFAS No. 107 (for example, Nelson, 1996 and Eccher et al., 1996). The market may be aware that some fair s of energy trading assets and liabilities reflect quoted market prices of these assets and liabilities, and others are best estimates obtained from valuation models, which are not incorporated into the of share equity. Thus, fair s that are unverifiable are not informative of the of the firm (Watts, 2003). However, the results in our paper are subject to several caveats. First, it is possible that omitted variables in our model specification may have resulted in non-significant coefficients for our key variables. Liu and Thomas (2000) observe that omitted variables result in coefficient estimates that are biased and regressions that have low explanatory power. Barth et al. (1996) note that an explanation for Nelson s (1996) findings of non-significance for the SFAS 107 fair estimates is due to important variables omitted from Nelson s (1996) model specification. Second, another limitation of this paper is the small sample size resulting in a lack of power in our tests. Third, many of the companies mention in their annual reports that they use energy contracts for hedging purposes. For example, Aquila mentions, We trade energy commodity contracts daily. Our trading activities attempt to match our portfolio of physical and financial contracts to current or anticipated market conditions. Within the trading portfolio, we take certain positions to hedge physical sale or purchase contracts and we take certain positions to take advantage of market trends and conditions. We record most energy contracts both physical and financial at fair. Changes in are reflected in the consolidated statement of income. We use all forms of financial instruments, including futures, forwards, swaps and options. Each type of financial instrument involves different risks. We believe financial instruments help us manage our contractual commitments, reduce our exposure to changes in cash market prices and take advantage of selected arbitrage opportunities. We refer to these transactions as price risk management activities. These hedging activities to offset fluctuations may explain why fair s of energy trading assets and liabilities are not priced by the market.

L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 259 6.3. Sensitivity analyses 6.3.1. Regression of market on excess of fair over original of net energy trading assets We run Eq. (2) with excess of fair over original of energy trading assets and liabilities combined as a single variable, instead of energy trading assets and energy trading liabilities as two separate variables, that is: MV it ¼ a 0 þ a 1 NI it þ b 1 NetNTA it þ b 3 NetTA it þ b 5 ðeta it ETL it Þþe it We obtain similar results. The excess of fair over original of net energy trading assets and liabilities (ETA ETL) remains not significantly related to market. The adjusted R 2 is 0.6501 for this model compared with 0.6514 for the model with energy trading assets and energy trading liabilities as two separate variables as in Table 3, Column 3. 6.3.2. Regression of market on assets and liabilities In our model, we regress market on book of net non trading assets, original of net energy trading assets, excess of fair over original of energy trading assets and liabilities, and net income. This classification of book of equity into net trading assets and net non-trading assets restricts the corresponding assets and liabilities to have the same valuation coefficient in the regression. We relax this restriction, and regress market on book of non trading assets and non-trading liabilities, original of energy trading assets and energy trading liabilities, excess of fair over original of energy trading assets and liabilities, and net income. 9 The results (not tabulated) indicate that the variables that are significant in explaining market valuation are book s of non-trading assets (coefficient = 2.351) and non-trading liabilities (coefficient = 2.683), original s of energy trading assets (coefficient = 3.157) and energy trading liabilities (coefficient = 1.988). The excess of fair over original of energy trading assets and liabilities, and net income are not significantly related to market valuation. 7. Conclusion This paper examines the -relevance of mark-to-market s of assets and liabilities of energy trading companies. EITF Issue No. 98-10 allowed for energy trading contracts to be marked-tomarket (that is, measured at fair as of the balance sheet date) with the gains and losses included in earnings and separately disclosed in the financial statements. This consensus was reiterated in EITF Issue No. 00-17, which stated that the estimate of fair should be based on the best information available in the circumstances. However, EITF Issue No. 02-3 ruled to ban the use of mark-to-market accounting for energy contracts out of concern that fair s can be easily inflated. The EITF was concerned that energy trading assets and profits may be inflated by s of energy contracts that are marked-to-market. To examine whether mark-to-market energy assets and liabilities are -relevant, we regress market of equity on net income, book of net non-trading assets, original of net energy assets and the excess of fair over the original of energy assets and liabilities. Our results indicate that market valuation is positively related to book of net non-trading assets and original of net energy trading assets. We also find that the excess of fair over original of energy trading assets and liabilities is not -relevant. In conclusion, our findings indicate that the market s of our sample energy trading firms do not reflect the fair s of energy trading assets and liabilities. Our findings support the findings in some prior research on the explanatory power of SFAS No. 33 disclosures of current cost operating income (for example, Beaver and Ryan, 1985) and -relevance of banks fair disclosures under SFAS No. 107 (for example, Nelson, 1996 and Eccher et al. 1996). We interpret this as the market being aware that some fair s of energy trading assets and liabilities are, at best, estimates obtained from valuation models. Thus, these fair estimates are not relevant for valuation. 9 We thank a reviewer for this suggestion.

260 L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 This paper is subject to a few limitations. First, we have a small sample size due to mark-to-market accounting being proposed in EITF 98-10 and subsequently rescinded in EITF 02-3. Second, there may be other firms that engage in energy trading activities not captured by our search criteria. Finally, we do not have sufficient data to analyze fair s of energy trading assets and liabilities by specific types; our analyses are based on aggregate s of energy trading assets and liabilities. As fair s are being endorsed by SEC and FASB, future research may provide further evidence on the -relevance of fair s of various kinds of assets and liabilities across a wider spectrum of industries. Acknowledgements We thank James Myers, Sandeep Nabar, Jake Thomas and participants at the forum session of the 2004 AAA Annual Meeting for helpful comments and suggestions. We would also like to acknowledge helpful discussions with Gary Belitz, Thomas Sell and J. Kevin Vann of Williams Company on industry practices relating to mark-to-market accounting for energy contracts, and Bill Stanley on alternative external financing. This paper was formerly titled An empirical examination of mark-to-market accounting for energy contracts under EITF Issue No. 98-10. Data are available from the authors. Appendix A. CLECO POWER LLC Note 3 fair of financial instruments. The amounts reflected in Cleco Power s Balance Sheets at December 31, 2002, and 2001, for cash and cash equivalents, accounts receivable, accounts payable, and short-term debt approximate fair because of their short-term nature. Estimates of the fair of Cleco Power s long-term debt is based upon the quoted market price for the same or similar issues or by a discounted present analysis of future cash flows using current rates obtained by management for debt with similar maturities. The estimated fair of energy market positions is based upon observed market prices when available. When such market prices are not available, management estimates market at a discrete point in time by assessing market conditions and observed volatility. These estimates are subjective in nature and involve uncertainties. Therefore, actual results may differ from these estimates. Fair of financial instruments At December 31 2002 2001 Carrying (Thousands) Financial instruments not marked-to-market Estimated fair Carrying Estimated fair Long-term debt $361,260 $384,543 $336,260 $357,775 Original Estimated fair Original Estimated fair Financial instruments marked-to-market Energy market positions Assets $ 20,793 $ 24,457 $ 800 $ 799 Liabilities $ 32,652 $ 37,239 $ 3,984 $ 4,091 The financial instruments not marked-to-market are reported on Cleco Power s Balance Sheets at carrying. The financial instruments marked-to-market represent off-balance sheet risk because, to the extent Cleco Power has an open position, it is exposed to the risk that fluctuating market prices may adversely affect its financial condition or results of operations upon settlement. Original represents the fair of the positions at the time originated. Source: SEC EDGAR, Cleco Power LLC, 2002 10-K Annual Financials.

L.L. Eng et al. / J. Account. Public Policy 28 (2009) 251 261 261 References Ahmed, A., Kilic, E., Lobo, G., 2006. Does recognition versus disclosure matter? Evidence from -relevance of banks recognized and disclosed derivative financial instruments. The Accounting Review 81, 567 588. Barth, M., Beaver, W., Landsman, W., 1996. Value-relevance of banks fair disclosures under SFAS No. 107. The Accounting Review 71, 513 537. Beaver, W., Landsman, W., 1983. Incremental Information Content of Statement 33 Disclosures. Financial Accounting Standards Board, Stamford, CT. Beaver, W., McNichols, M., 2001. Do stock prices of property casualty insurers fully reflect information about earnings, accruals, cash flows, and development? Review of Accounting Studies 6, 197 220. Beaver, W., Ryan, S., 1985. How well do Statement No. 33 earnings explain stock returns? Financial Analysts Journal 41, 66 71. Bergstresser, D., Desai, M., Rauh, J., 2006. Earnings manipulation, pension assumptions and managerial investment decisions. Quarterly Journal of Economics 121, 157 195. Bernard, V., Ruland, R., 1987. The incremental information content of historical cost and current cost income numbers: Timeseries analyses for 1962 1980. The Accounting Review 62, 707 722. Bublitz, B., Frecka, T.J., McKeown, J.C., 1985. Market association tests and FASB Statement No. 33 disclosures: A reexamination. Journal of Accounting Research 23 (Supplement), 1 23. Eccher, A., Ramesh, K., Thiagarajan, S.R., 1996. Fair disclosures by bank holding companies. Journal of Accounting and Economics 22, 79 117. The Economist, 2005. The Ones that Get Away. July 30th, pp. 14, 65 66. Emerging Issues Task Force (EITF) of the Financial Accounting Standards Board (FASB), 1998. Accounting for contracts involved in energy trading and risk management activities. Emerging Issues Task Force Issue No. 98-10. Financial Accounting Standards Board, Norwalk, CT. Emerging Issues Task Force and (EITF) of the Financial Accounting Standards Board (FASB), 2000. Measuring the fair of energy related contracts in applying Issue No. 98-10. Emerging Issues Task Force Issue No. 00-17. Financial Accounting Standards Board, Norwalk, CT. Emerging Issues Task Force and (EITF) of the Financial Accounting Standards Board (FASB), 2002. Issues involved in accounting for derivative contracts held for trading purposes and contracts involved in energy trading and risk management activities. Emerging Issues Task Force Issue No. 02-3. Financial Accounting Standards Board, Norwalk, CT. Financial Accounting Standards Board (FASB), 2007a. Fair measurements. Statement of Financial Accounting Standards No. 157. Financial Accounting Standards Board, Norwalk, CT. Financial Accounting Standards Board (FASB), 2007b. The fair option for financial assets and financial liabilities. Statement of Financial Accounting Standards No. 159. Financial Accounting Standards Board, Norwalk, CT. Hanlon, M., Myers, J., Shevlin, T., 2003. Dividend taxes and firm valuation: a reexamination. Journal of Accounting and Economics 35, 119 153. Harris, T., Kemsley, D., 1999. Dividend taxation in firm valuation: new evidence. Journal of Accounting Research 17, 275 291. Haw, I., Lustgarten, S., 1988. Evidence on income measurement properties of ASR No. 190 and SFAS No. 33 data. Journal of Accounting Research 26, 331 352. Hodder, L., Hopkins, P., Wahlen, J., 2006. Risk-relevance of fair- income measures for commercial banks. The Accounting Review 81, 337 375. Holthausen, R., Watts, R., 2001. The relevance of -relevance literature for financial accounting standard setting. Journal of Accounting and Economics 31 (September), 3 75. Hopwood, W., Schaefer, T., 1989. Firm-specific responsiveness to input price changes and the incremental information in current cost income. The Accounting Review 64, 312 338. Lehavy, R., 2002. Reporting discretion and the choice of fresh start s in companies emerging from Chapter 11 bankruptcy. Review of Accounting Studies 7, 53 73. Lev, B., Li, S., Sougiannis, T., 2009. The usefulness of counting estimates for predicting cashflows and earnings. Working paper, New York University, University of Illinois. Liu, J., Thomas, J., 2000. Stock returns and accounting earnings. Journal of Accounting Research 38, 71 101. Lobo, G.J., Song, I.M., 1989. The incremental information in SFAS No. 33 income disclosures over historical cost income and its cash and accrual components. The Accounting Review 64, 329 343. Murdoch, B., 1986. The information content of FAS 33 returns on equity. The Accounting Review 61, 273 287. Noland, T., Pasewark, W., Strawser, J., 1998. An investigation of the accuracy of pre-implementation estimates required by SAB 74. Journal of Accounting and Public Policy 17, 227 244. Nelson, K., 1996. Fair accounting for commercial banks: an empirical analysis of SFAS No. 107. The Accounting Review 71, 161 182. Ohlson, J., 1995. Earnings, book s, and dividends in equity valuation. Contemporary Accounting Research 12, 661 687. Ray, R., 2002. Energy traders face rule changes. Tulsa World, E1, E2 (Nov. 6). Ray, R., 2003. Williams loses $814 million, Tulsa World, E1, E8 (May 14). Watts, R., 2003. Conservatism in accounting Part I: Explanations and implications. Accounting Horizons 17, 207 221. White, H., 1980. A heteroscedasticity-consistent covariance matrix estimator and a direct test for heteroscedasticity. Econometrica 48, 817 838.