This article was downloaded by: [Florida State University] On: 25 December 2014, At: 03:22 Publisher: Routledge Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK European Accounting Review Publication details, including instructions for authors and subscription information: http://www.tandfonline.com/loi/rear20 An international comparison of income statement and balance sheet information: Germany, Japan and the US a Ervin L. Black & John J. White a b Brigham Young University b Elmhurst College Center for Business and Economics Elmhurst IL USA 60126-3296 Published online: 06 Aug 2010. To cite this article: Ervin L. Black & John J. White (2003) An international comparison of income statement and balance sheet information: Germany, Japan and the US, European Accounting Review, 12:1, 29-46, DOI: 10.1080/0963818022000001127 To link to this article: http://dx.doi.org/10.1080/0963818022000001127 PLEASE SCROLL DOWN FOR ARTICLE Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of the Content. Any opinions and views expressed in this publication are the opinions and views of the authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon and should be independently verified with primary sources of information. Taylor and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use of the Content. Downloaded by [Florida State University] at 03:22 25 December 2014 This article may be used for research, teaching, and private study purposes. Any substantial or systematic reproduction, redistribution, reselling, loan, sublicensing, systematic supply, or distribution in any form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http:// www.tandfonline.com/page/terms-and-conditions European Accounting Review 2003, 12:1, 29–46 Downloaded by [Florida State University] at 03:22 25 December 2014 An international comparison of income statement and balance sheet information: Germany, Japan and the US Ervin L. Black Brigham Young University John J. White Elmhurst College Manuscript first received: September 2000. Manuscript accepted: March 2002 ABSTRACT As accounting regulators develop international standards, it is helpful for them to understand the use of financial statements by investors in different countries. In this study we compare the value relevance of earnings relative to book value of equity in Germany, Japan and the US owing to the size of their capital markets and differences in accounting systems and institutional structures. Accounting in Germany and Japan is conservative, creditor-oriented and tax-based, with institutional structures that rely on bank financing and close relationships between capital providers and investee firms, while the US is used as a basis for comparison. This study hypothesizes that book values are more value relevant than earnings in Germany and Japan while earnings are more value relevant in the US, because capital providers in Germany and Japan are more concerned with balance sheet measures such as liquidity. Also, accounting characteristics such as conservatism and tax conformity may lead to greater value relevance of the balance sheet compared to the income statement in those countries. Our results provide evidence that book value of equity is more value-relevant than earnings, particularly in Germany, with mixed results for Japan. In the US, positive earnings are more value relevant than book value of equity, but not negative earnings. In summary, we provide evidence concerning the relative value relevance of summary measures of the balance sheet versus the income statement. Standard-setters can then look at potential causes for these differences and determine if proposed accounting standards will have a desired outcome. Addresses for correspondence Ervin L. Black, Assistant Professor, 529 TNRB, School of Accounting and Information Systems, Marriott School of Management, Brigham Young University, Provo, UT 84602, USA. E-mail: erv_black@byu.edu John J. White, Assistant Professor, Center for Business and Economics, Elmhurst College, Elmhurst, IL 60126-3296, USA. Copyright # 2003 European Accounting Association ISSN 0963-8180 print=1468-4497 online DOI: 10.1080=0963818022000001127 Published by Routledge Journals, Taylor & Francis Ltd on behalf of the EAA 30 European Accounting Review Downloaded by [Florida State University] at 03:22 25 December 2014 1. INTRODUCTION In this study we compare the value relevance of earnings relative to book value of equity in Germany, Japan and the US. Previous studies have found evidence of differences in the value relevance of earnings across countries (Alford et al., 1993; Ali and Hwang, 2000; Collins et al., 1997; Hall et al., 1994; Harris et al., 1994). However, these studies have not compared the value relevance of earnings (a summary income statement measure) to book value of equity (a summary balance sheet measure). We examine Germany, Japan and the US owing to the size of their capital markets and differences in accounting systems and institutional structures. Accounting in Germany and Japan is conservative, creditor-oriented and tax-based, with institutional structures that rely on bank financing and close relationships between capital providers and investee firms. The US is used as a basis for comparison because it has more of an investor focus and companies in the US have been accused of a focus on the income statement. Diversity in cultures and political systems across the globe has resulted in differences in accounting systems. These differences can impede many different aspects of international economic activity such as stock exchange listings by firms outside their home countries if investors in those countries are unable to interpret financial statements prepared using foreign generally accepted accounting principles (GAAP). For example, US stock exchanges are bound by strict Securities and Exchange Commission (SEC) regulations that forbid foreign companies from listing in the United States unless foreign firms prepare US GAAP financial statements or reconcile their financial statements to US GAAP. Currently a movement is under way to reduce the diversity in international accounting systems and thereby alleviate the resulting problems. US stock exchanges such as the New York Stock Exchange are lobbying the Securities and Exchange Commission (SEC) to change its rules, and the SEC has made some concessions along these lines. In addition, the International Organization of Securities Commissions (IOSCO) has agreed to accept with many provisos a core set of international standards for cross-border listing for all member nations since the completion in 1998 of new standards by the International Accounting Standards Board (IASB). Listed companies in the European Union are required to prepare consolidated accounts in accordance with International Accounting Standards (IASs) by 2005. Other major countries including Japan, Canada and Australia are also supporting this initiative. By examining the relative usefulness of accounting data from different statements – the income statement and balance sheet – our study provides information to those formulating international accounting standards. As regulators continue the task of standardizing accounting rules internationally they will have a clearer understanding of the relative usefulness of balance sheet Downloaded by [Florida State University] at 03:22 25 December 2014 An international comparison of Germany, Japan and the US 31 disclosures versus income statement disclosures in the three largest capital markets: Germany, Japan and the US. There are two potential causes for a difference in the value relevance of balance sheet disclosures and income statement disclosures. First, investor characteristics: investors in some countries, owing to institutional and cultural factors, prefer one type of information versus another. For example, in Germany, investors primarily consist of large institutions who are both debt and equity providers. Thus, in this first case, income statement disclosures, regardless of the accounting rules, may not provide the information needed by these investors, such as that provided in the balance sheet disclosures about the liquidity and solvency of a company. Second, accounting characteristics: a country’s generally accepted accounting principles may be structured such that more information relevant to investors is found in the balance sheet than in the income statement. In this case a change in accounting standards could affect this relationship. Standard-setters can only affect the second cause, at least in the short term. Thus, to the extent that there are differences in the value relevance of disclosures across countries, regulators need to use caution as they prepare new standards that are used across many different countries. What may be better for one country may not have the same impact on users in other countries. For example, if new accounting standards are implemented that improve the income statement at the expense of the balance sheet, will investors be better off if the information that they relied on was principally found in the balance sheet? While we do not attempt to answer this question, or to distinguish between these two potential causes, we do provide information about the relative value relevance of summary measures of the balance sheet versus the income statement. Standard-setters can then look at potential causes for these differences and determine if proposed accounting standards will have the desired outcome. Because of accounting and institutional differences along several dimensions (Gray, 1988; Nobes, 1983, 1998; Nair and Frank, 1980; Mueller, 1968), accounting in Germany and Japan differs from that in the US. Thus, we hypothesize that the relative value relevance of earnings and book value differs in these three countries. Because of greater creditor orientation, greater conservatism, a higher level of tax conformity, lower diversity of shareholders, and the greater use of provisions to smooth earnings in Germany and Japan it is hypothesized that book value of equity exhibits greater value relevance relative to earnings in Germany and Japan. However, in the US earnings are expected to be relatively more value relevant than book value of equity. Our results generally provide support for the hypotheses in Germany and are robust to different research methods used. However, in Japan the results are less robust and seem to hold only for firms with negative earnings and=or which are small. In the US, earnings are shown to be more value relevant for positive earnings and large firm samples. 32 European Accounting Review Downloaded by [Florida State University] at 03:22 25 December 2014 2. HYPOTHESIS DEVELOPMENT While previous studies have found evidence of differences in the value relevance of earnings across countries, we focus on three countries in this study: Germany, Japan and the US. Germany and Japan are used because they have similar corporate governance structures and accounting systems (Nobes, 1998). The US is used as a benchmark for comparison purposes because it provides an example of a marketoriented institutional structure based on the Anglo-Saxon=British accounting model, which differs significantly from that used in Germany and Japan. German accounting is often cited in accounting literature as a prime example of the Continental model of accounting development. German accounting is characterized by its basis in code law, its emphasis on financial reporting conformity with tax accounting, conservatism, and a focus on creditors and the balance sheet. Government has a significant influence on accounting in Japan. There are three sources of accounting regulation in Japan: the Commercial Code, the Securities and Exchange Law, and the tax laws and regulations. The first Commercial Code was enacted in 1890 in Japan and was heavily influenced by the German accounting model. It was oriented towards creditors and tax collection. Historically shareholders were considered insiders while creditors were considered the outsiders (Fujita, 1991) and thus accounting was geared towards creditor protection. Taxation, creditor interests and the conservative culture in Japan may cause earnings to be understated. One study, based on data during the 1980s, found that Japanese earnings were understated by 33.9% relative to US GAAP earnings (Aron, 1991).1 In Japan, more emphasis is placed on growth in sales and market share (Radebaugh and Gray, 1997). The use of historical cost and conservatism in Japanese accounting is similar to German accounting. In fact, Radebaugh and Gray (1997) state that Japanese accounting is a more extreme form of Germanic accounting. In both Germany and Japan there is a system of relationship-oriented capital providers who are more highly concentrated than those in the US and whose focus is on long-term business relationships. Table 1 shows the relative percentages of equity ownership in the US, Germany and Japan. In both Germany Table 1 Percentage of outstanding corporate equity held by various sectors in Germany, Japan and the US Banks Other financial institutions Non-financial firms Individuals Foreign Government US Germany Japan 0.2 44.3 15.0 36.3 4.2 0.0 10.3 20.0 42.1 14.6 8.7 4.3 13.3 22.5 31.2 22.4 10.1 0.5 Source: Stephen D. Prowse, ‘Corporate finance and government: an international perspective’, Southwest Economy, September=October 1997, Federal Reserve Bank of Dallas. Downloaded by [Florida State University] at 03:22 25 December 2014 An international comparison of Germany, Japan and the US 33 and Japan individual ownership of shares is less widespread than it is in the US, although public ownership is greater in Japan than in Germany. In Germany, a significant amount of capital is provided by a system of ‘universal banks’. They supply debt and equity capital. As Table 1 shows, 10.3% of German equity is owned by banks compared to only 0.2% in the US. There is also a large system of cross-holdings wherein companies own stakes in each other. This is evidenced by the data in Table 1 which show that 42.1% of the equity in German firms is owned by other non-financial businesses. The institutional structure in Japan is similar to that of Germany. In Japan, shareholders tend to be banks, suppliers and customers. These ‘stable shareholders’ are interested in maintaining business ties so shares are held on a longterm basis. There is also a large system of cross-holdings whereby companies hold stakes in each other, also similar to Germany. Japanese firms have a heavy reliance on bank debt and, in many cases, banks may also be the largest shareholders. These banks are referred to as ‘main banks’ in Japan and Table 1 shows that 13.3% of the equity in Japan is held by banks. Institutional shareholders exhibit influence as associates who hold shares as a means of doing business. Cooke and Kikuya (1992) state that 75% of the shares held on the Tokyo Stock Exchange may be held for purposes of long-term business relationships. The involvement of banks and the long-term nature of share ownership mean that there is less focus on short-term earnings and also on conservatism and creditor protection. In conclusion, there are many similarities between the German and Japanese institutional structures. Both are characterized by groups of dedicated permanent owners. Their aims are the perpetuation of the enterprise and the building of the corporate position as opposed to the maximization of period-by-period profit. This varies from the US model in that corporate governance in Germany and Japan is relationship-based while in the US it is market-based (Kester, 1997). The bank-based system focuses on close, long-term relationships while the marketbased system focuses on short-term returns (Franks and Mayer, 1997). Book value of equity is a summary measure of the balance sheet – the net of assets less liabilities available to common shareholders. It is the net (long-term, over the life of the company) accumulated wealth of the shareholders as measured by accounting rules. Earnings, on the other hand, are the profits that a firm has earned during the most recent period and are not necessarily a reflection of long-term wealth accumulation, but provide information about the company’s recent operations. Thus, in Germany and Japan which have a longer-term focus, we expect that book value of equity is more value relevant than earnings, while the opposite relation is expected in the US. Formally stated our research hypotheses are: H1: The relative value relevance of the book value of equity is greater than earnings in Germany and Japan. H2: The relative value relevance of earnings is greater than book value of equity in the US. 34 European Accounting Review 3. RESEARCH METHODOLOGY Data The sample of countries selected for this study consists of Germany, Japan and the US. Firm-specific data for each of these countries is obtained from the Global Vantage database. Global Vantage is the international version of Compustat. Earnings before extraordinary items, book value, market value of equity, SIC code, level of consolidation, accounting standard and fiscal year come from the Global Vantage database. Downloaded by [Florida State University] at 03:22 25 December 2014 Sample selection criteria The sample selection criteria employed in this study are as follows. First, firms in the SIC codes from 4800 to 4999 (Regulated Industries: Communications and Electric, Gas and Water Utilities) and from 6000 to 6999 (Financial Services) are excluded.2 Second, only firms reporting accounting data using domestic standards and full consolidation are used except in the case of Japan where a full consolidation and a parent-only sample are used owing to the small number of firms using full consolidation. Third, any firm reporting a negative book value is excluded. Data on level of consolidation, accounting standard, earnings before extraordinary items, book value, market value of equity and total assets are collected during the thirteen-year period 1986–98. A firm-year is included if all such variables are available for a given fiscal year. In addition, outliers are removed using a truncating procedure to reduce the potential of results being affected by a few observations. Observations, which fall in the 1st or 99th percentiles, are deleted from the sample for the following variables: earnings before extraordinary items (earnings), book value of equity, and market value of equity. Testing the relative value relevance of earnings and book values The first hypothesis states that the value relevance of book value of equity is greater than earnings in Germany and Japan. The second hypothesis states that earnings are relatively more value relevant than book value of equity in the US. This implies that the proportion of variation in market value of equity explained by book value of equity is relatively larger than that explained by earnings in Germany or Japan, but earnings is relatively more explanatory in the US. The basic model used to test our hypotheses is equation (1) that has been used in numerous accounting studies including Barth et al. (1998), Burgstahler and Dichev (1997) and Collins et al. (1997). This model is particularly useful for our study because the two independent variables are the summary measures from the balance sheet and the income statement: MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit ð1Þ An international comparison of Germany, Japan and the US 35 where: MVEit ¼ Market value of equity at the end of the third month following the end of fiscal year t, for firm i. BVEit ¼ Book value of equity at the end of fiscal year t, for firm i. Downloaded by [Florida State University] at 03:22 25 December 2014 EARNit ¼ Earnings before extraordinary items for fiscal year t, for Firm i. Two statistical methods are used to assess the relative value relevance of earnings and book value of equity. The first is a methodology derived and used by Biddle et al. (1995) in assessing the relative information content of net income, cash flows and net sales for stock returns. This method uses a Wald chi-square statistic adjusted for heteroskedasticity to compare the ability of the two independent variables in equation (1) to explain a dependent variable under the null hypothesis that the explanatory power of both variables is equal. Thus, rejection of the null indicates that one of the independent variables has greater explanatory power than the other.3 (The Appendix provides a detailed explanation of this test statistic.) The variable with greater explanatory power is derived by comparing R-squares of the two competing independent variables in the following equations: MVEit ¼ b0 þ b1 EARNit þ eit ð1aÞ MVEit ¼ b0 þ b1 BVEit þ eit ð1bÞ These Wald chi-square tests are performed on equation (1) for each of the three country samples and the R-squares are compared for equations (1a) and (1b) for each country. A second method used to test for relative value relevance is to assess the proportions of variation in the dependent variable, market value of equity, which are accounted for by book values and earnings, respectively. This method uses standardized regression coefficients (Johnson et al., 1987) to assess the relative importance of independent variables in a regression equation. Each variable is standardized to mean 0 and variance of 1, then estimates the equation (1) regression using the standardized variables for each country sample. Next, the coefficients are compared using an F-test. Hayn (1995) and Burgstahler and Dichev (1997) provide evidence that the value relevance of earnings is affected by the sign; negative earnings are less value relevant than positive earnings. We control for this effect by separating each country sample into positive and negative earnings samples. In addition, we control for size by separating the sample into quartiles based on size in each country sample. To test the sensitivity of the results to macroeconomic conditions and changes in accounting standards tests are also conducted by year. 36 European Accounting Review 4. RESULTS This section discusses the results of the study. It begins with descriptive statistics and then discusses the results of hypothesis tests. Descriptive statistics Downloaded by [Florida State University] at 03:22 25 December 2014 We split the samples into four groups: Germany; Japan–Consolidated; Japan– Parent Only; and the United States. Sample sizes for each of the four samples are shown in Table 2, Panel A. The German sample has 1,673 firm-year observations, Japan–Consolidated has 2,187 firm-year observations, Japan–Parent Only has 7,962 firm-year observations and the US has 24,321 firm-year observations. Table 2 Descriptive statistics for selected variables (in $ millions) Panel A: Means and standard deviation Variable Market value of equity Assets Earnings Book value of equity Country Germany N ¼ 1,673 Japan– Consolidated N ¼ 2,187 Japan– Parent Only N ¼ 7,962 United States N ¼ 24,321 706.6 (1,341.6) 1,262.9 (3,021.9) 28.2 (74.6) 309.0 (634.8) 2,503.3 (3,885.1) 4,747.2 (8,705.4) 58.9 (135.8) 1,232.9 (2,035.0) 975.1 (1,150.3) 1,150.2 (1,754.1) 17.5 (28.4) 357.2 (400.8) 1,059.3 (2,418.2) 1,233.7 (3,046.6) 53.4 (139.1) 427.4 (888.4) Panel B: t-Tests for mean differences across country samples (t-statistic and p-value) Variable Country Germany vs US Market value of equity Assets Earnings Book value Japan– Consolidated vs US Japan– Parent Only vs US 7 9.7 (0.0001) 17.1 (0.0001) 4.2 (0.0001) 0.37 (0.70) 7 12.7 (0.0001) 7 7.2 (0.0001) 18.8 (0.0001) 1.6 (0.12) 18.4 (0.0001) 3.0 (0.0026) 38.5 (0.0001) 9.7 (0.0001) Notes: Market value of equity is measured at the end of the third month following the fiscal year end. Assets is total assets at fiscal year end. Earnings is income before extraordinary items for the fiscal year. Book value of equity is the book value of common=ordinary equity at fiscal year end. All of these values are in $ millions. Downloaded by [Florida State University] at 03:22 25 December 2014 An international comparison of Germany, Japan and the US 37 The average market value of equity is much larger for the Japan–Consolidated sample, as are assets, earnings and book value of equity compared to the US, Germany and Japan–Parent Only samples. t-Tests for mean differences (Table 2, Panel B) of Germany, and the two Japan samples compared to the US show that there is a significant difference in the average size of these variables across countries, with the exception of earnings in Germany and the US. In summary, these tables indicate that there are differences among the four country groups. For Germany, market value, earnings and book value are smaller while assets are very similar to the other countries. For Japan–Consolidated, all are larger with the exception of earnings. The US is larger across all of the three countries. Japan–Parent Only is smaller. In order to test for any effects of size on value relevance, the regressions used to test the hypotheses are run after the sample is split into quartiles based on total assets. These results are reported in the tables along with the results for the pooled sample and individual year samples. Value relevance of book value of equity versus earnings This section discusses the test results of hypotheses for the relative value relevance of earnings and book value of equity within countries. Table 3 provides evidence confirming the hypotheses. For the German sample, the Wald chi-square statistic is significant at a p-value of 0.0005 and the book value of equity regression has a higher R-square than the earnings regression. This indicates that in Germany book value of equity has more value relevance than earnings in explaining market value of equity. Similar results are found for the Japanese sample groups; see Table 3. There is a significant difference in the explanatory power of earnings and book value, Table 3 Wald w2 test results for relative value relevance MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit MVEit ¼ b0 þ b1 EARNit þ eit ð1Þ ð1aÞ MVEit ¼ b0 þ b1 BVEit þ eit ð1bÞ Country R2 Earnings Equation (1a) R2 Book value of equity Equation (1b) w2 Test statistic Equation (1) p-Value Germany Japan–Consolidated Japan–Parent Only United States 0.576 0.571 0.542 0.714 0.700 0.648 0.604 0.654 12.29 3.35 44.85 53.28 0.0005 0.0671 < 0.0001 < 0.0001 38 European Accounting Review Downloaded by [Florida State University] at 03:22 25 December 2014 particularly in the Japan–Parent Only sample. For the Japan–Consolidated sample the difference is only marginally significant at the 0.0671 level. The R-squares for book value of equity are greater than earnings in both of these samples. Thus, the evidence supports the hypothesis that in Japan book value of equity is more value relevant than earnings. In the US sample, the second hypothesis is supported. Again, the chi-square statistic is significant at less than a 0.0001 level. However, the R-square for earnings is greater than that for book value of equity. This indicates that earnings in the US, during this period of time, are more value relevant than book value of equity. Sensitivity analysis To assess the sensitivity of these results to the test method used we test the two hypotheses using standardized regression coefficients. These results confirm and support the results using the Biddle et al. (1995) tests; see Table 4. The standardized regression coefficients on book value of equity are statistically greater than the earnings coefficient at less than a 0.0001 level for Germany and the Japanese groups. In the US, the standardized regression coefficient on earnings is greater than the coefficient on book value of equity. Hayn (1995) finds that the value relevance of earnings depends on the sign of earnings. We separate the groups into positive and negative earnings samples and perform tests of value relevance on each of these positive and negative earnings samples within each country group sample. Table 5 provides results of these tests. The results of these tests support H1 for Germany, but not in all the Japanese sub-samples. In Germany, book value of equity is significantly more value relevant than earnings for both the positive and negative earnings samples. It Table 4 Standardized regression coefficients test for relative value relevance MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit ð1Þ Country Standardized regression coefficient Earnings Standardized regression coefficient Book value of equity F-Value (H0: Earnings ¼ Book value of equity) p-Value Germany Japan–Consolidated Japan–Parent Only United States 0.288 0.379 0.362 0.562 0.616 0.541 0.513 0.336 79.38 31.07 73.10 393.50 0.0001 0.0001 0.0001 0.0001 An international comparison of Germany, Japan and the US 39 Table 5 Wald w2 test results for relative value relevance sensitivity analysis: negative and positive earnings MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit MVEit ¼ b0 þ b1 EARNit þ eit MVEit ¼ b0 þ b1 BVEit þ eit Downloaded by [Florida State University] at 03:22 25 December 2014 Country N ð1Þ ð1aÞ ð1bÞ w2 Test R2 Book p-Value R2 Earnings Equation (1a) value of statistic equity Equation (1) Equation (1b) Germany Positive earnings Negative earnings 1,424 0.604 249 0.109 0.694 0.801 6.16 23.52 0.0131 < 0.0001 Japan–Consolidated Positive earnings Negative earnings 1,849 0.701 338 0.073 0.660 0.429 1.47 23.25 0.2245 < 0.0001 Japan–Parent Only Positive earnings Negative earnings 7,363 0.588 599 0.070 0.605 0.423 3.59 20.21 0.0583 < 0.0001 19,726 0.747 4,595 0.096 0.650 0.568 140.02 65.64 < 0.0001 < 0.0001 United States Positive earnings Negative earnings should be noted that the R-square for negative earnings is lower than for positive earnings in these samples. In Japan, however, book value of equity is more significant than earnings only in the negative earnings samples. In the positive earnings samples for the Japan– Consolidated sample there is no statistical difference in explanatory power between earnings and book value of equity. For the Japan–Parent Only, positive earnings sample, book value of equity is only marginally significant and the book value of equity has a higher R-squared. In the US sample, only the positive earnings group supports H2; earnings are more value relevant than book value of equity if earnings are positive. However, if earnings are negative, book value of equity is more value relevant at significant levels. The test results on the negative earnings samples in all countries are most likely due to the transitory nature of negative earnings. Transitory earnings should not be very value relevant; thus, book value of equity becomes relatively more value relevant compared to earnings when earnings are negative with more transitory components. Another sensitivity test performed is to check if macroeconomic events impact the results. We do this by examining each country sample by year. Table 6, 40 European Accounting Review Table 6 Wald w2 test results for relative value relevance sensitivity analysis: yearly tests MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit MVEit ¼ b0 þ b1 EARNit þ eit MVEit ¼ b0 þ b1 BVEit þ eit R2 Book value of equity Equation (1b) w2 Test statistic Equation (1) p-Value Panel A: Germany 1986 27 0.867 1987 33 0.889 1988 72 0.649 1989 108 0.848 1990 138 0.511 1991 141 0.631 1992 138 0.541 1993 175 0.354 1994 189 0.508 1995 188 0.447 1996 215 0.523 1997 200 0.658 1998 49 0.652 0.844 0.881 0.845 0.919 0.641 0.734 0.759 0.792 0.645 0.787 0.861 0.841 0.660 1.47 0.17 20.68 31.92 12.03 7.24 15.00 81.61 6.53 37.77 30.75 15.75 0.01 0.225 0.679 < 0.0001 < 0.0001 0.0005 0.0071 0.0001 < 0.0001 0.0106 < 0.0001 < 0.0001 < 0.0001 0.9206 Panel B: Japan–Consolidated 1986 3 0.186 1987 13 0.760 1988 49 0.645 1989 85 0.706 1990 100 0.754 1991 94 0.787 1992 84 0.326 1993 101 0.200 1994 108 0.618 1995 109 0.674 1996 696 0.577 1997 739 0.541 1998 6 0.284 0.608 0.500 0.633 0.649 0.737 0.768 0.707 0.801 0.825 0.861 0.782 0.662 0.739 1.9E29 3.22 0.10 2.47 0.11 0.14 21.49 93.33 18.58 28.63 26.44 6.67 5.84 < 0.0001 0.0728 0.7503 0.1164 0.7435 0.7039 < 0.0001 < 0.0001 < 0.0001 < 0.0001 < 0.0001 0.0098 0.0157 Panel C: Japan–Parent Only 1986 759 0.509 1987 810 0.607 1988 807 0.574 1989 799 0.587 1990 794 0.629 1991 801 0.587 1992 793 0.571 1993 766 0.484 1994 739 0.569 1995 735 0.588 0.617 0.684 0.684 0.671 0.701 0.661 0.746 0.773 0.724 0.780 10.31 6.90 19.51 8.74 9.17 8.32 44.40 77.71 36.32 57.18 0.0013 0.0086 < 0.0001 0.0031 0.0025 0.0039 < 0.0001 < 0.0001 < 0.0001 < 0.0001 Downloaded by [Florida State University] at 03:22 25 December 2014 Year N R2 Earnings Equation (1a) ð1Þ ð1aÞ ð1bÞ An international comparison of Germany, Japan and the US 41 Table 6 (continued ) R2 Earnings Equation (1a) R2 Book value of equity Equation (1b) 0.795 0.778 n.a. 0.785 0.754 n.a. 0.13 0.39 n.a. 0.7165 0.5322 n.a. Panel D: United States 1986 1,703 0.731 1987 2,113 0.741 1988 2,049 0.807 1989 1,926 0.813 1990 1,862 0.801 1991 1,866 0.733 1992 1,906 0.782 1993 1,965 0.723 1994 1,956 0.802 1995 1,907 0.694 1996 1,922 0.755 1997 1,763 0.713 1998 1,383 0.615 0.687 0.683 0.684 0.680 0.635 0.683 0.727 0.719 0.725 0.683 0.708 0.690 0.608 2.68 3.66 12.94 20.49 29.79 2.08 5.23 0.03 12.53 0.11 3.67 0.64 0.05 0.1015 0.0557 0.0003 < 0.0001 < 0.0001 0.1490 0.0222 0.8590 0.0004 0.7346 0.0555 0.4240 0.8310 Year Downloaded by [Florida State University] at 03:22 25 December 2014 1996 1997 1998 N 94 63 2 w2 Test statistic Equation (1) p-Value Panel A, provides results for Germany. Only the years with lower sample size (1986, 1987 and 1998) are insignificant. In all other years, 1988–97, book value of equity is more value relevant than earnings. The results are much more variable for the Japanese sample. In the Japanese samples, Table 6, Panels B and C, there are some years with insignificant differences between earnings and book value of equity. However, in all cases, when there is a significant difference book value of equity has a higher R-square than earnings; earnings are never statistically more significant than book value of equity. For the Japan–Consolidated sample from 1987–91, there is no statistical difference between earnings and book value of equity. However, from 1992–98 book value of equity is significantly more explanatory than earnings for the Japan–Consolidated samples. In the Japan–Parent Only samples the results are more robust; book value of equity is more explanatory in all years except 1996 and 1997, in which there is no statistical difference. In the US (Table 6, Panel D), earnings generally have higher R-squares than the book value of equity. In yearly samples with significant chi-square statistics earnings are more value-relevant than book value of equity. These results confirm the second hypothesis. However, the US results are driven by a few years: 1987–90, 1992, 1994 and 1996 are least significant at a 0.06 level or higher. As a final sensitivity we examine the effect of size on the results by partitioning each of the four groups into four size quartiles and performing the Biddle et al. (1995) tests. These results are found in Table 7. 42 European Accounting Review Table 7 Wald w2 test results for relative value relevance sensitivity analysis: size quartiles MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit MVEit ¼ b0 þ b1 EARNit þ eit MVEit ¼ b0 þ b1 BVEit þ eit R2 Earnings Equation (1a) R2 Book value of equity Equation (1b) 414 415 416 414 0.097 0.144 0.134 0.471 0.237 0.375 0.201 0.581 11.00 36.40 4.33 5.18 0.0009 < 0.0001 0.0375 0.0229 Japan–Consolidated 1st quartile 545 2nd quartile 545 3rd quartile 546 4th quartile 546 0.383 0.413 0.490 0.479 0.267 0.410 0.459 0.505 1.95 0.01 0.39 0.27 0.1629 0.9237 0.5313 0.6005 Japan–Parent 1st quartile 2nd quartile 3rd quartile 4th quartile 1,989 1,990 1,990 1,989 0.120 0.141 0.309 0.453 0.165 0.114 0.184 0.453 11.23 3.10 24.36 0.00 0.0008 0.0782 < 0.0001 0.9906 United States 1st quartile 2nd quartile 3rd quartile 4th quartile 6,047 6,047 6,048 6,048 0.058 0.175 0.335 0.608 0.218 0.231 0.353 0.501 276.25 18.37 1.28 85.55 < 0.0001 < 0.0001 0.2575 < 0.0001 Country Size quartile Downloaded by [Florida State University] at 03:22 25 December 2014 ð1Þ ð1aÞ ð1bÞ Germany 1st quartile 2nd quartile 3rd quartile 4th quartile N w2 Test statistic Equation (1) p-Value Size does not affect the tests of hypotheses in Germany; book value of equity is more value relevant than earnings in each of the size quartiles. However, there are not significant differences in explanatory power in any of the size quartiles for the Japan–Consolidated samples. In the Japan–Parent Only samples the first quartile is the only sample that confirms the hypothesis. In the other quartiles the Rsquares are either opposite what is hypothesized or insignificant. In the US, earnings are only more value relevant than book value of equity for firms in the fourth quartile. There is no statistical difference for firms in the third quartile. Book value of equity is more value relevant than earnings for smaller firms in the first and second quartiles. This is consistent with findings by Black (1999), which indicates that firms in growth and start-up stages, that are likely to be small, have limited value relevance of earnings. The US results in the pooled sample appear to be driven by the large firms. An international comparison of Germany, Japan and the US 43 Downloaded by [Florida State University] at 03:22 25 December 2014 To summarize, the overall results are consistent with the main hypotheses, but must be tempered by the sensitivity analyses: 1 The results for Germany are robust. Book value of equity is more value relevant for both positive and negative earnings, in all years with sufficient sample sizes, and for all quartiles. 2 For Japan the results are much less clear. The Japan–Parent Only sample results are robust across all years, but seem to be driven by smaller firms. For the Japan–Consolidated sample, the results show that book value of equity is more value relevant than earnings only for negative earnings in later years. Also, when size is factored in there is no significant difference in any of the size quartiles. 3 In the US the results seem to be driven by positive earnings firms, later years and large companies. The results are not robust for small firms or for negative earnings firms. 5. CONCLUSIONS AND IMPLICATIONS This study theorizes that balance sheet information, as measured by book value of equity, is more value relevant than income statement information, as measured by earnings, in Germany and Japan owing to accounting factors and institutional factors. The accounting factors include a creditor orientation, conservatism, tax conformity and the use of provisions to smooth earnings. Institutional factors primarily reflect differences in capital providers, particularly the greater role of banks as suppliers of both debt and equity capital and the close relation between suppliers of capital and the firms in which they invest. Our results generally support this hypothesis, H1, in Germany. For German firms, using various sensitivity tests, we provide evidence that book value of equity is more value relevant, as reflected by higher R-squares and standardized regression coefficients. However, in Japan the results are less robust and seem to hold only for firms with negative earnings and=or which are small. In the US, earnings are shown to be more value relevant in the pooled, positive earnings, and large firm samples, supporting H2. The results are unaffected by test statistics. However, for negative earnings samples and smaller firms book value of equity appears to be more value relevant than earnings. Thus, the pooled results are driven by large, positive earnings firms. These results imply that the value relevance of accounting information is different across countries (and across type of firm in the US and Japan). Our tests do not try to distinguish whether one accounting system provides more valuerelevant information than another. What we do show is that, given a set of accounting standards, cultural norms and regulatory environment the relative valuation implications of earnings and book value of equity (two summary measures) differ in these countries. Thus, to the extent that international accounting standard-setting bodies impact the set of accounting standards used 44 European Accounting Review in a country they can have some impact on the relative usefulness of accounting disclosures. It should be noted, however, that standard-setting bodies have little, if any, control over cultural norms and regulatory environments. Therefore, changing standards, by itself, may not have the desired effect of improving value relevance if these cultural and regulatory effects are still present. APPENDIX Wald chi-square relative information content tests by country Downloaded by [Florida State University] at 03:22 25 December 2014 (1) BE has equal value relevance with ( ¼ R) EARN Run regressions: MVEit ¼ b0 þ b1 BVEit þ b2 EARNit þ eit ð1Þ (a) Information content of BVE Define X as an (n 2) matrix with a column of 1s to the left of column BVE. Define Y as an (n 1) matrix with columns EARN. Define M as a (2 2) matrix ¼ Y0Y 7 Y0X(X0X)1X0Y, with elements mij. Set BVE ¼ m11*b1^2 þ m22*b2^2 þ 2*m12*b1*b2 (b) Information content of EARN Define X as an (n 2) matrix with a column of 1s to the left of columns for EARN. Define Y as an (n 1) matrix with columns BVE. Define M as a (2 2) matrix ¼ Y0Y 7 Y0X(X0X)1X0Y, with elements mij. Set EARN ¼ m11*b1^2 þ m22*b2^2 þ 2*m12*b1*b2 (c) Use Wald chi-square test to test the following restrictions: BVE ¼ R EARN If a significant chi-square value is obtained then compare R2 from the following regressions to determine which variable has the greatest value relevance: MVEit ¼ b0 þ b1 EARNit þ eit ð1aÞ MVEit ¼ b0 þ b1 BVEit þ eit ð1bÞ NOTES 1 It should be noted that, more recently, Japanese companies have been covering up disasters by realizing capital gains on sale of old assets. This may cause earnings to be overstated, but would similarly reduce the value relevance of the net income disclosure. 2 These firms are excluded owing to the unique nature of their financial statements. For example, regulated industries earnings are in many cases controlled, as are other variables in the financial statements. An international comparison of Germany, Japan and the US 45 Downloaded by [Florida State University] at 03:22 25 December 2014 3 Tests of the relative value-relevance hypotheses are performed using Wald chi-square tests on the significance of the squared coefficients. This methodology is appropriate asymptotically when the disturbances could be heteroskedastic and assumes that the error terms, ei, from equations (1a) and (1b) are vectors of unobserved random disturbances that are normally distributed with mean zero and unknown covariance matrix. Two different matrices represent subsets of predictor variables; each consists of a subset of columns of the independent variables. The object is to test whether one subset of predictor variables is significantly better than the other for the purpose of explaining the dependent variable, MVE. In general when comparing the quality of two sets of predictors, X1 and X2, the matrix Y1 consists of the columns of X that are not in X1, and similarly for Y2. B1 and B2 are subsets (1 and 2, above) of regression coefficients for Y1 and Y2, respectively. Thus, the null hypothesis becomes in general form: H0 : B01 Y10 ½In X1 ðX10 X1 Þ1 X10 Y1 B1 ¼ B02 Y20 ½In X2 ðX20 X2 Þ1 X20 Y2 B2 ð2Þ This non-linear hypothesis is of quadratic form in the regression coefficients. The lefthand side of the hypothesis represents the bias due to omitting Y1 from the regression, keeping only X1, which is a quadratic form in the omitted regression coefficients B1. The right-hand side represents the bias when Y2 is omitted. The test result is then computed using the estimated coefficients and their heteroskedasticity-adjusted variance–covariance matrix. The Wald test is asymptotically valid and is appropriate when specific hypotheses are prespecified: as is the case in this paper. REFERENCES Alford, A., Jones, J., Leftwich, R. and Zmijewski, M. (1993) ‘The relative informativeness of accounting disclosures in different countries’, Journal of Accounting Research, 31 (Supplement): 183–223. Ali, A. and Hwang, L. S. (2000) ‘Country-specific factors related to financial reporting and the value relevance of financial accounting data’, Journal of Accounting Research, 38(1) Spring: 1–21. Aron, P. H. (1991) ‘Japanese P=E ratios in an environment of increasing uncertainty’, in Choi, F. D. S. (ed.) Handbook of International Accounting. New York: John Wiley. Barth, M. E., Beaver, W. H. and Landsman, W. R. (1998) ‘Relative valuation roles of equity book value and net income as a function of financial health’, Journal of Accounting and Economics, 25: 1–34. Biddle, G. C., Seow, G. and Siegel, A. (1995) ‘Relative versus incremental information content’, Contemporary Accounting Research, 12 (Summer): 1–23. Black, E. L. (1999) ‘Which is more value relevant: earnings or cash flows? A life-cycle examination’, Working Paper, Brigham Young University, Provo, UT. Burgstahler, D. C. and Dichev, I. D. (1997) ‘Earnings, adaptation and equity value’, Accounting Review, 72 (April): 187–215. Collins, D. W., Maydew, E. and Weiss, I. (1997) ‘Changes in the value relevance of earnings and book values over the past forty years’, Journal of Accounting and Economics, 24: 39–68. Cooke, T. and Kikuya, M. (1992) Financial Reporting in Japan Regulation, Practice and Environment. Oxford: Blackwell. Franks, J. and Mayer, C. (1997) ‘Corporate ownership and control in the UK, Germany, and France’, in Chew, D. H. (ed.) Studies in International Corporate Finance and Downloaded by [Florida State University] at 03:22 25 December 2014 46 European Accounting Review Governance Systems: A Comparison of the US, Japan and Europe. Oxford: Oxford University Press. Fujita, Y. (1991) An Analysis of the Development and Nature of Accounting Principles in Japan. New York: Garland. Gray, S. J. (1988) ‘Towards a theory of cultural influence on the development of accounting systems internationally’, Abacus, 24: 1–15. Hall, C., Hamao, Y. and Harris, T. (1994) ‘A comparison of relations between security market prices, returns and accounting measures in Japan and the United States’, Journal of International Financial Management and Accounting, 5: 47–73. Harris, T. S., Lang, M. and Moller, H. (1994) ‘The value relevance of German accounting measures: an empirical analysis’, Journal of Accounting Research, 32 (Autumn): 187–209. Hayn, C. (1995) ‘The information content of losses’, Journal of Accounting and Economics, 20: 125–53. Johnson, A. C., Johnson, M. B. and Buse, R. C. (1987) Econometrics Basic and Applied. New York: Macmillan. Kester, C. (1997) ‘Governance, contracting, and investment horizons: a look at Japan and Germany’, in Chew, D. H. (ed.) Studies in International Corporate Finance and Governance Systems: A Comparison of the US, Japan and Europe. Oxford: Oxford University Press. Mueller, G. G. (1968) ‘Accounting principles generally accepted in the United States versus those generally accepted elsewhere’, International Journal of Accounting Education and Research, Spring: 91–103. Nair, R. D. and Frank, W. (1980) ‘The impact of disclosure and measurement practices on international accounting classifications’, Accounting Review, July: 426–50. Nobes, C. W. (1983) ‘A judgmental international classification of financial reporting practices’, Journal of Business Finance and Accounting, 10: 1–19. Nobes, C. W. (1998) ‘Towards a general model of the reasons for international differences in financial reporting’, Abacus, 34(2): 162–87. Radebaugh, L. H. and Gray, S. (1997) International Accounting and Multinational Enterprises, 4th edn. New York: John Wiley.