«Leif Atle Beisland University of Agder Dissertation submitted to the Department of Accounting, Auditing and Law at the Norwegian School of Economics ...»
5.1 Disaggregation of the Book Value of Equity The book value of equity can be disaggregated into assets minus liabilities and from there into finer partitions of assets and liabilities. When partitioning assets and debt, we exclude financial companies such as banks, which shrinks the sample from 725 to 602 firm-year observations. Financial firms are excluded because they have a very special balance sheet relative to other firms. Specifically, assets and debt cannot easily be divided into operational and financial components for many financial firms.
In order to avoid too many variables, we choose this partition of the book value of equity:
where INT equals intangible assets and OOA is other operating assets, which includes property, plant and equipment, investments in associated companies, other long term operating assets, inventory and other short term operational assets, per outstanding shares. FA is financial assets, and FD is financial debt per share. To avoid an extra variable of minor significance, the minority interests are simply classified as financial debt (and not as equity because only the majority’s equity is valued by the stock market). OD is operational debt per share, both long term provisions such as deferred taxes and short term debt such as taxes payable.
Notice that INT is different from the control variable INTAN.39 Specifically, while INT is the reported intangible assets divided by the number of outstanding shares, the indicator variable INTAN is constructed from industry membership; see Table 1.
INT is intangible assets per share, OOA is other operating assets per share, FA is financial assets per share, FD is financial debt per share, and OD is operational debt per share; see Panel B of Table 1 for definitions of the rest of the variables. In the correlation matrix, the IFRS coefficients are found below and the NGAAP coefficients are found above the diagonal. The regression model is PRICE = α0 · IND + α11 · INT + α12 · OOA + α13 · FA + α14 · FD + α15 · OD + α2 · IFRS + α31 · EARN’ + α32 · LOSS + α33 · INTAN + α34 · TRAN + α35 · BETA + α36 · SIZE + α41 · INT · IFRS + α42 · OOA · IFRS + α43 · FA · IFRS + α44 · FD · IFRS + α45 · OD · IFRS + α51 · BOOK · EARN’ + α52 · BOOK · LOSS + α53 · BOOK · INTAN + α54 · BOOK · TRAN + α55 · BOOK · BETA + α56 · BOOK · SIZE + ε; see (1) - (3). IND is a vector of dummy variables for each industry. This means that there is one constant term for each industry, meaning that fixed industry effects are controlled for. The coefficients of IND are not reported. The set of control variables does not include BTM and MOM. BTM is already represented by BOOK. Including the lagged price (MOM) in the regression would change the specification to a regression of the price change on BOOK. The coefficient estimates are based on OLS, unless in the last regression model that utilizes feasible GLS in which HAC is taken into account in the coefficient estimates; it allows firm-specific heteroskedasticity and first-order autocorrelation (41 observation are lost because of only one observation in the panel). The condition number is a measure of multicollinearity. If it is above 20, there is some troublesome multicollinearity; if it is above 30, there is severe multicollinearity; see Belsley, Kuh and Welsch (1980). The condition number with control variables equals 33.33 or 84.00, which indicate severe multicollinearity. As suggested by the average variance inflation factors of the test variables, some of the collinearity stems from these variables, making inferences about their coefficients somewhat arbitrary. To build the arbitrariness into statistical inferences, bootstrapped standard deviations with 1000 replications are employed when calculating t- and p-values in the OLS regressions. Thus, for collinear variables BOOT produces higher standard deviation than the HAC standard deviations. One asterisk * means statistical significance at the 10% level, two asterisks ** means significance at the 5% level, and three asterisks *** means significance at the 1% level, tested two-sided. Due to lacking observations, the sample is reduced from 602 to 517 when control variables are employed in the model.
sets OOA and 17.1% financial assets FA. These assets are financed by 35.8% equity BOOK, 36.8% financial debt FD and 27.4% operational debt OD. The balance sheet is also split on the two reporting regimes IFRS and NGAAP. The binary correlations between the variables are presented in Panel B of Table 6; IFRS observations are below and NGAAP observations above the diagonal.
Panel C gives the regression results. First notice that there is severe multicollinearity involving the five test variables INT · IFRS, OOA · IFRS, FA · IFRS, FD · IFRS and OD · IFRS.
The variance inflation factors show that three of the five test variables are collinear, as they are above 10. This means that the loadings on these coefficients could be somewhat ‘arbitrary’, possibly making statistical inferences based on HAC standard deviations inadequate.
We utilize instead standard deviations directly, incorporating the possible within sample arbitrariness of collinear coefficient estimates, and we do this by bootstrapping.40 Thus, we draw, for example, 1000 new samples from our sample with each drawing containing the same number of observations, and then estimate the coefficient in each sample; see Green (2008, pp. 596-598). Based on these 1000 observations of each OLS coefficient, we calculate the standard deviation. This procedure typically increases the standard deviation of the coefficient of variables involved in multicollinearity, relative to the HAC standard deviations. Thus, we have thereby adjusted for increased uncertainty in the coefficient estimates – and hence improved the statistical reliability of t- and p-values. In addition, we have employed specifications with and without control variables to catch possible instabilities.
Collinearity increases the variance inflation factor of a variable and thereby the standard deviation of its OLS coefficient. This means that collinear variables typically have high standard deviations and low t-values, making them more insignificant. Still, the bootstrapping procedure might better capture within sample ‘arbitrariness’ in the loading of the coefficients.
ing to IFRS than NGAAP; this holds for all three types of assets INT, OOA and FA. Furthermore, the regression coefficients of debt, FD as well as OD, are also more responsive, since they are more negative. This is consistent with the conclusion in subsection 4.1, in which the response coefficient of the book value of equity BOOK was found to be higher according to IFRS than NGAAP. The most significant difference is related to financial debt FD. The difference between IFRS and NGAAP is estimated at about -1.2 according to the OLS regression model with control variables, suggesting more weight on the response coefficient according to IFRS, as it is more negative and larger in absolute value. One reason could be that more financial instruments are reported at fair value.
5.2 Disaggregation of Earnings Earnings can be disaggregated into revenue minus expenses and then into finer partitions of
revenue and expenses:
(8) EARN = REV - OCOST - NFC + TEARN - TAX, in which REV is operational revenue, OCOST is operational expenses, NFC is net financial expenses, TEARN is transitory or non-recurring earnings, and TAX is the tax expense – payable as well as deferred, all variables are deflated by the previous year’s stock price. Notice that TEARN is different from the control TRAN, but collinearity between them may cause problems.41 Panel A of Table 7 presents descriptive statistics regarding these variables – for the total sample and the two subsamples according to reporting regime.
Transitory earnings are defined in Table 1. While TEARN is transitory earnings divided by ingoing market value of equity, TRAN is and indicator variable for ‘extreme’ TEARN.
REV is operational revenue per share, OCOST is operational expenses per share, OEARN = REV - OCOST, NFC is net financial expenses per share, TEARN is transitory or non-recurring earnings per share, and TAX is taxes per share; all these variables have been price-deflated; the other variables is defined in Panel B of Table 1.
In the correlation matrix, the IFRS coefficients are found below and the NGAAP coefficients are found above the diagonal. The regression model is RET = β0 · IND + β11 · OEARN + β12 · NFC + β13 · TEARN + β14 · TAX + β2 · IFRS + β31 · BETA + β32 · SIZE + β33 · BTM + β34 · MOM + β35 · LOSS + β36 · INTAN + β37 · TRAN + β41 · OEARN · IFRS + β42 · NFC · IFRS + β43 · TEARN · IFRS + β44 · TAX · IFRS + β51 · EARN · BETA + β52 · EARN · SIZE + β53 · EARN · BTM + β54 · EARN · MOM + β55 · EARN · LOSS + β56 · EARN · INTAN + β57 · EARN · TRAN + ε; see also the ERC given by (4) - (6). IND is a vector of dummy variables for each industry.
This means that there is one constant term for each industry, meaning that fixed industry effects are controlled for. The coefficients of IND are not reported. Since the Breusch-Pagan test for heteroskedasticity (H) and the Arallano-Bond test of autocorrelation (AC) detect significant HAC, we employ Newey-West standard deviations when calculating the t- and p-values; see White (1980) and Newey and West (1997). One asterisk * means statistical significance at the 10% level, two asterisks ** means significance at the 5% level and three asterisks *** means significance at the 1% level, tested two-sided. The condition number is a measure of multicollinearity. If it is above 20, there is some troublesome multicollinearity and if it is above 30, there is severe multicollinearity;
see Belsley, Kuh and Welsch (1980). OEARN is utilized in the regression instead of OREV and OCOST, because these variables according to Panel B are highly collinear. We observe that even though the condition number is high in the regressions with control variables, about 40.1, the variance inflation factors show that test variables are not collinear. Accordingly, we use HAC standard deviations and not the bootstrapped ones; see Table 6.
In the total sample, percentage of OCOST relative to REV is 96.9%, NFC is 1.9%, TEARN is
-0.4% and TAX is 0.5%. This means that earnings EARN is only 0.3%, on average. Panel B presents the binary correlation matrix involving the variables in (8); IFRS correlations are below the diagonal and NGAAP correlations are above the diagonal. Notice that the NGAAP
than the IFRS correlations. Further, notice that REV and OCOST are collinear, suggesting that we should replace them with operational earnings OEARN = REV - OCOST, in the following regression analyses.
Panel C of Table 6 gives the regression results. First, we observe that there is no problematic collinearity involving test variables. We therefore employ HAC standard deviations to examine statistical significance. Second, the coefficients of the test variables are not significant when employing OLS, which is consistent with the second regression model in Table 5. If we utilize GLS, allowing for panel specific heteroskedasticity and first order autocorrelation, we observe that three of the coefficients become significant at the 5% level, which is consistent with the last regression model in Table 5. The response coefficients of the operating earnings OEARN and transitory earnings TEARN are significantly higher according to NGAAP than according to IFRS.
A major source for the finding in subsection 4.2 that IFRS earnings are less value relevant than NGAAP earnings, is that the value relevance of transitory earnings TEARN is smaller under IFRS. Thus, the result could be explained by transitory items being relatively more common under IFRS than under NGAAP, because of more gains and losses due to measurement at fair value; compare e.g. Hann, Heflin and Subramanayan (2007) and Stunda and Typpo (2004).
It is more surprising that the operating earnings OEARN is less responsive under IFRS, given more recognition of intangible assets and the results obtained by Gjerde, Knivsflå and Sættem (2008). However, in subsection 5.1, we find that intangible assets recognized in the balance
less weight on the corresponding signal in the income statement, i.e. OEARN; compare Penman (1998). Thus, more recognition of intangible assets might be detrimental to the value relevance of earnings, contrary to our expectation, due to the positive and dominating balance sheet effect.
6. Conclusions This study compares the response coefficients of book values and earnings under respectively IFRS and NGAAP. The two accounting regimes are in many respects similar, but IFRS generally allows more measurement at fair value and in practice recognizes more intangible assets than NGAAP. We expected IFRS to exhibit higher response coefficients in relation to the balance sheet (equity), though the hypothesis and the test is two-sided. The effect on earnings response coefficients was considered to be more ambiguous. However, we indicated that nonrecurring gains and losses might contribute to earnings response coefficients being lower under IFRS than under NGAAP.
We find evidence that the response coefficient of book equity is higher under IFRS than under NGAAP, as initially expected – in any case it is not lower. Disaggregation of the book value of equity suggests that the difference is related to all items in the balance sheet, but most to financial debt and operating assets. This is consistent with the view that more recognition of intangible assets and measurement at fair value are improving the value relevance of the balance sheet.
Furthermore, we find some evidence that the earnings response coefficient is lower under IFRS than under NGAAP. Among the main sources of a significant difference is less weight
ate ‘noise’ in earnings and depress earnings response coefficients; see also e.g. Hann, Heflin and Subramanayam (2007) and Stunda and Typpo (2004). From a policy perspective, our results suggest that IASB should consider introducing a clearer distinction between recurring and non-recurring earnings in the format of the income statement. This is also on the IASB’s agenda (Economist, 20. September 2008, p. 81).
- 284 Literature Aboody, D., J. Hughes, and J. Liu, 2002, “Measuring Value Relevance in a (Possibly) Inefficient Market,” Journal of Accounting and Economics, 40, pp. 965-986.
Aboody, D., and B. Lev, 1998, “The Value Relevance of Intangibles: The Case of Software Capitalization,” Journal of Accounting Research, 36, pp. 161-191.
Ahmed, A. S., and C. Takeda, 1995, “Stock Market Valuations of Gains and Losses for Commercial Banks’ Investment Securities: An Empirical Analysis,” Journal of Accounting and Economics, 20, 207-255.
Baker, M., and J. Wurgler, 2006, “Investor Sentiment and the Cross-Section of Stock Returns,” Journal of Finance, 61, pp. 230-264.
Barth, M. E., 1994, “Fair Value Accounting: Evidence form Investment Securities and the Market Valuation of Banks,” Accounting Review, 69, pp. 1-25.
Barth, M. E., W. H. Beaver and W. H. Landsman, 1996, “Value-Relevance of Banks’ Fair Value Disclosures under SFAS No. 107,” Accounting Review, 71, pp. 513-537.
Barth, M., and S. Kallapur, 1996, “The Effects of Cross-Sectional Scale Differences on Regressions in Empirical Accounting Research,” Contemporary Accounting Research, 19, pp. 1-48.
Basu, S., 1997, “The Conservatism Principle and the Asymmetric Timeliness of Earnings,” Journal of Accounting and Economics, 24, pp. 3-37.
Beaver, W. H., 1998, “Financial Reporting: An Accounting Revolution.” Third Edition, Prentice Hall.
Beaver, W. H., and M. Venkatachalam, 2003, “Differential Pricing of Components of Bank Loan Fair Values,” Journal of Accounting, Auditing & Finance, 18, pp. 41-67.
Belsley, D. A., E. Kuh, and R. E. Welsch, 1980, “Regression Diagnostics: Identifying Influential Data and Sources of Collinearity,” Wiley.
Bradshaw, M. T., and R. Sloan, 2002, “GAAP versus The Street: An Empirical Assessment of Two Alternative Definitions of Earnings,” Journal of Accounting Research, 40, pp. 41Brown, S., K. Lo, and T. Lys, 1999, “Use of R2 in Accounting Research: Measuring Changes in Value Relevance over the Last Four Decades,” Journal of Accounting and Economics, 28, pp. 83-115.
Cairns, D., 2006, “The Use of Fair Value in IFRS,” Accounting in Europe, 3, pp. 5-22.
Capkun, V., A. Cazavan-Jeny, T. Jeanjean and L. A. Weiss, 2008, “Earnings Management and Value Relevance during the Mandatory Transition from Local GAAP to IFRS in Europe,” Paper posted at SSRN: http://www.ssrn.com/abstact=1125716.
Carhart, M. M., 1997, “On Persistence in Mutual Fund Performance,” Journal of Finance, 52, pp. 87-82.
Carroll, T. J., T. J. Linsmeier and K. R. Petroni, 2003, “The Reliability of Fair Values versus Historical Cost Information: Evidence from Closed-End Mutual Funds,” Journal of Accounting, Auditing & Finance, 18, pp. 1-23.
Chow, G., 1960, “Test of Equality between Sets of Coefficients in Two Linear Regressions,” Econometrica, 28, pp. 591-605.
Collins, D. W., E. L. Maydew, and I. S. Weiss, 1997, “Changes in the Value Relevance of Earnings and Book Values of the Past Forty Years,” Journal of Accounting and Economics, 24, pp. 39-67.
Cramer, J. S., 1987, “Mean and Variance of R2 in Small and Moderate Samples,” Journal of Econometrics, 35, pp. 253-266.
Daske, H., L. Hail, C. Leuz, and R. Verdi, 2007, “Mandatory IFRS Reporting Around the
World: Early Evidence on the Economic Consequences.” Paper posted at SSRN:
- 285 Dichev, I. D., 2007, ”On the balance sheet based model of financial reporting.” Paper posted at SSRN: http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1009673.
Easton, P. D., and T. S. Harris, 1991, “Earnings as an Explanatory Variable for Returns,” Journal of Accounting Research, 29, pp. 19-36.
Easton, P., and G. Sommers, 2003, “Scale and the Scale Effect in Market-Based Accounting Research,” Journal of Business Finance & Accounting, 30, pp. 25-55.
Eccher, E. A., K. Ramesh, and S. R. Thiagarajan, 1996, “Fair Value Disclosures by Bank Holding Companies,” Journal of Accounting and Economics, 22, 79-117.
Elliot, J. A., and J. D. Hanna, 1996, “Repeated Accounting Write-Offs and the Information Content of Earnings,” Journal of Accounting Research, 34, pp. 135-155.
Fama, E. F., and K. R. French, 1992, “The Cross-Section of Expected Returns,” Journal of Finance, 47, pp. 427-465.
Francis, J., and K. Schipper, 1999, “Have Financial Statements Lost Their Relevance?” Journal of Accounting Research, 37, pp. 319-352.
Ghosh, A., G. Zhaoyang, and P. J. Jain, 2005 “Sustained Earnings and Revenue Growth, Earnings Quality, and Earnings Response Coefficients,” Review of Accounting Studies, 10(1), pp. 33-57.
Gjerde, Ø., K. H. Knivsflå, and F. Sættem, 2008, “The Value Relevance of Adopting IFRS:
Evidence from 145 NGAAP Restatements,” Forthcoming in the Journal of International Accounting, Auditing and Taxation, 17.
Green, W. H., 2008, “Econometric Analysis,” Sixth Edition, Pearson Prentice Hall.
Gu, Z., 2007, “Across-Sample Incomparability of R2s and Additional Evidence on Value Relevance Changes Over Time,” Journal of Business Finance & Accounting, 34, pp.
Hamberg, M., J. Novak, and M. Paananen, 2008, “The Impact of IFRS Adoption of Goodwill Recognition and Stock Market Valuation,” Working Paper, Göteborg University, Sweden.
Hann, R. N., F. Heflin, and K. R. Subramanayam, 2007, “Fair-Value Pension Accounting,” Journal of Accounting & Economics, 44, pp. 328-358.
Hayn, C., 1995, “The Information Content of Losses,” Journal of Accounting and Economics, 20, pp. 125-153.
Henning, S., B. Lewis, and W. Shaw, 2000, “Valuation of the Components of Purchased Goodwill,” Journal of Accounting Research, 38, pp. 375-386.
Horton, J., and G. Serafeim, 2008, “Market Reaction & Valuation of IFRS Reconciliation Adjustments: First Evidence from the UK,” Paper posted at SSRN: http://www.ssrn.
Jennings, R., J. Robinson, R. B. Thompson II, and L. Duvall, 1996,”The Relation between Accounting Goodwill Numbers and Equity Values,” Journal of Business Finance & Accounting, 23, pp. 513-533.
Khurana, I. K., and M. Kim, 2003, “Relative Value Relevance of Historical Cost vs. Fair Value: Evidence from Bank Holding Companies,” Journal of Accounting and Public Policy, 22, pp. 19-42.
Kothari, S. P., 2001, Capital Markets Research in Accounting. Journal of Accounting & Economics 31, pp. 105-231.
Lev, B., 1989, “On the Usefulness of Earnings and Earnings Research: Lessons and Directions from Two Decades of Empirical Research,” Journal of Accounting Research, 27, pp. 153-192.
Lev, B., and T. Sougiannis, 1996, “The Capitalization, Amortization, and Value-Relevance of R&D,” Journal of Accounting and Economics, 21, pp. 107-138.
- 286 Lev, B. and P. Zarowin, 1999, “The Boundaries of Financial Reporting and How to Extend Them,” Journal of Accounting Research, 37, pp. 353-385.
Li, G., 1985, “Robust Regression,” In Hoaglin, D. C., F. Mosteller, and J. W. Turkey (eds.), “Exploring Data Tables, Trends, and Shapes,” Wiley, pp. 281-340.
Nelson, K. K., 1996, “Fair Value Accounting for Commercial Banks: An Empirical Analysis of SFAS No. 107,” Accounting Review, 71, pp. 161-182.
Newey, W., and K. West, 1987, “A Simple Positive Semi-Definite, Heteroscedasticity and Autocorrelation Consistent Covariance Matrix,” Econometrica, 55, pp. 703-708.
Nissim, D., 2003, “Reliability of Banks’ Fair Value Disclosure of Loans,” Review of Quantitative Finance and Accounting, 20, pp. 355-384.
Nissim, D, and S. H. Penman, 2008, “Principles for the Application of Fair Value Accounting,” Center for Excellence in Accounting and Security Analysis, White Paper Number Two.
Park, M. S., T. Park, and B. T. Ro, 1999, “Fair Value Disclosures for Investment Securities and Bank Equity: Evidence from SFAS No. 115,” Journal of Accounting, Auditing & Finance, 14, pp. 347-370.
Penman, S. H., 1998, “Combining Earnings and Book Value in Equity Valuation,” Contemporary Accounting Research, 15, pp. 291-342.
Penman, S. H., 2007, “Financial Reporting Quality: Is Fair Value a Plus of a Minus?”, Accounting and Business Research, Special Issue, pp. 33-44.
Petroni, K., and J. Wahlen, 1995, “Fair Value of Equity and Debt Securities and Share Prices of Property-Liability Insurers,” Journal of Risk and Insurance, 62, pp. 719-737.
Stunda, R. A, and E. Typpo, 2004, “The Relevance of Earnings and Funds Flow from Operations in the Presence of Transitory Earnings,” Journal of Real Estate Portfolio Management, 10, pp. 37-46.
Venkatachalam, M., 1996, “Value-Relevance of Banks’ Derivatives Disclosures,” Journal of Accounting and Economics, 22, pp. 327-355.
White, H., 1980, “A Heteroscedastic-Consistent Covariance Matrix Estimator and a Direct Test of Heteroscedasticity,” Econometrica, 48, pp. 817-838.
Whittington, G., “Fair Value and the IASB/FASB Conceptual Framework Project: An Alternative View,” ABACUS, 44, pp. 139-168.
Wooldridge, J. M., 2008, “Introductory Econometrics: A Modern Approach,” Fourth Edition, South-Western.
Zhang, X. J., 2000, “Conservative Accounting and Equity Valuation,” Journal of Accounting and Economics, 29, pp. 125-149.