IFRS and the value relevance of
accounting numbers
M.A. LANDEWEER
Student number: 1335340
University of Groningen
MSc BA, Faculty of Management and Organization
Supervisor 1: H. Gonenc
ABSTRACT
In this study I investigate the impact of switching to International Financial Reporting System (IFRS ) on value-relevance under the Edwards-Bell-Ohlson valuation model (EBO model) and the return-earnings model. I use data from France, Germany, and The Netherlands over the time period 1999-2005. The results for the French sample show that book value of equity is relatively more value rele vant than earnings in all years. Earnings do become more important through the years, which confirm my expectations. However, I find that book value is not of much relevance relatively to earnings for Germany and the Netherlands in the years before 2005. The value relevance of both book value and earnings increases through the years. In 2005, I find evidence that the Netherlands has the highest total value relevance (79%), followed by France (77%), and finally Germany (69%).
The incremental value relevance of earnings, book value, and value relevance common to earnings and book value show a clear converging pattern for Germany and The Netherlands in 2005. The incremental value relevance of earnings seems a bit low for France and the incremental value relevance of book value a bit high. Consistent with my expectations, the value relevance common to earnings and book value is almost equal for all three countries in 2005.
TABLE OF CONTENTS
1. INTRODUCTION... 4
2. THEORETICAL MODELS ... 5
3. RELATED RESEARCH ... 7
3.1EBO VALUATION MODEL...7
3.2RETURN MODEL...8
4. ACCOUNTING SYSTEMS ... 9
4.1HISTORICAL BACKGROUND...9
4.1.1 Relative importance government in standard -setting process... 9
4.1.2 The providers of capital... 9
4.1.3 The influence of taxation...10
4.2DEVELOPMENTS IN THE EU ...10 4.2.1 EU Directives...10 4.2.2 IFRS. ...11 4.2.3 U.S. GAAP...12 4.3HYPOTHESES...12 5. METHODOLOGY...13
5.1TESTS AND ANALYSIS...13
5.2DATA AND SAMPLE SELECTION...14
6. RESULTS EBO VALUATION MODEL ...18
6.1ALL YEARS...18
6.2POSITIVE VERSUS NEGATIVE EARNINGS...19
6.3SAMPLES PER YEAR...22
6.3.1 Coefficients and relative value relevance. ...22
6.3.2 Incremental value relevance...23
6.3.3 Summary. ...28
7. RESULTS RETURN MODEL ...29
7.1ALL YEARS...29
7.2SAMPLES PER YEAR...30
8. CONCLUSION AND DISCUSSION ...32
1. INTRODUCTION
Harmonization is a hot item. Many countries over the world already accept the International Accounting Standards (IAS). The Securities and Exchange Commission (SEC) is even trying to minimize the differences between the General Accepted Accounting Standard of the United States (US GAAP) and the IAS.
In Europe the initial step was made in 1957 with the establishment of the European Community (EC). An important goal of the EC was to reach an economic equal level playing field within the Community, with harmonization of the accounting systems as an important feature (Haller and Keple r, 2002). The first initiatives in order to derive this harmonization did not lead to a satisfactory outcome. Many authors (e.g. King and Langli, 1998; Ballas and Hevas, 2005; Arce and Mora, 2002) conclude that there are significant differences in value relevance of accounting numbers between the different European Union (EU) countries, even after the initiatives towards harmonization. Because of the rapid globalization of capital markets from 1990 on, the demand for harmonization grew quickly. In 2002, the decision to require listed firms of the EU to comply with IAS (later International Financial Reporting Standards, IFRS) by 2005 was approved by the Parliament and Council of the European Union (Jermakowicz and Gornik-Tomaszewski, 2006).
25 Countries implement a completely new accounting system at the same time; this makes the EU the perfect environment for me to investigate whether the value relevance of accounting numbers is influenced by the accounting system.
In this paper I investigate whether the va lue relevance of accounting numbers changes when firms switch their accounting systems to IFRS. I use the Edwards-Bell-Ohlson valuation model (the EBO model) and a return-earnings model for analyses on data from listed firms from France, Germany, and The Netherlands. These countries have very distinct accounting systems originally, with distinct historical backgrounds. Germany employs the most conservative, creditor oriented, and tax-based system and The Netherlands on the other side has no connection to tax-law and a market oriented system.
with different accounting systems, differ in their usefulness for firm valuation (Graham and King, 2000).
A possible problem of the EBO model is that it suffers from a scale effect, because large security prices tend to be mechanically related to large book value and large earnings per share and vice versa (Dumontier and Raffournier, 2002). Therefore, I also use a return regression which is not affected by potentially serious scale problems, because stock data and accounting numbers are all scaled by beginning-of-period stock prices.
The paper is organized as follows. Section 2 discusses the EBO valua tion model and the return model. Section 3 outlines recent research using the same methodology as I. Section 4 addresses the historical backgrounds of the different accounting systems. Section 5 describes the methodology used. Section 6 presents the results of the EBO valuation model. Section 7 presents the results of the return-earnings model and section 8 concludes.
2. THEORETICAL MODELS
A well known finance theory is the dividend discount model (DDM). Under the DDM the value of a stock is equal to the discounted present value of the sum of next period’s dividends plus next period’s stock price or the discounted present value of all future dividends. (Ross, Westerfield and Jaffe, 2005, pp.112). To overcome practical difficulties of the DDM Ohlson (1995) redefined the model using accounting variables:
Where:
Pt = the market value, or price, of the firm’s equity at date t BV = (net) book value of equity at date t
= abnormal earnings r = cost of equity capital
Et = the expected value operator conditioned on the date t information
The EBO model is based on only three assumptions (Ohlson, 1995):
- Present value (PV) of expected dividends determines the market value. Combined
with the assumption of risk neutrality.
- The accounting system follows a clean surplus relation (CSR); this means that
book value of equity only changes through the income statement, net capital investments, and dividends.
- A linear model frames the stochastic time-series behavior of abnormal earnings
Because of the second assumption, firm value is theoretically not influenced by differences in accounting systems. This holds as long as the systems will follow a CSR in the future. When applying the model to finite horizons the model can be affected by bias (conservatism). These two features make it plausible that accounting values, from countries with different accounting systems, differ in their usefulness for firm valuation (Graham and King, 2000).
The EBO valuation model can also be used as a theoretical basis for a relationship between stock returns and earnings, to be more specific, between return and earnings level and earnings change. The advantage of a return model is that it overcomes potentially serious scale problems, which the EBO valuation model does face. For a detailed derivation of the model see Dumontier and Raffournier (2002). The empirical form of the restated EBO model is given by the following equation:
Where:
Rj,t = rate of return on security j for period t
xj,t = reported earnings of firm j over period t
? xj,t = earnings change between period t and period t-1
µ2j,t = error term
3. RELATED RESEARCH
Accounting and finance researchers have been studying the association between accounting numbers and capital market values for decades. In this section I will discuss recent research with study designs similar to mine. First, I will discuss research focused on stock price in relation to earnings and book value and second, research focused on returns in relation to earnings level and earnings change.
3.1 EBO valuation model
King and Langli (1998) examine the relationship between stock price and book value per share and earnings per share across Germany, Norway, and the United Kingdom (UK). The explanatory power (R-square) of German accounting numbers turns out to be about 40 percent. The relationship is a bit stronger in Norway with 60 percent. The accounting numbers of the UK show the strongest relationship with stock prices with an R-square of 60 percent. Book values explain more than earnings in Germany and Norway.
Ballas and Hevas (2005) use a slightly different approach; they investigate whether regulation effects or industry effects are greater. Using data from France, Germany, The Netherlands, and the UK, they conclude that accounting variables used in the EBO model do have va lue implications. When they use accounting variables to forecast market values they find that industry-specific valuation multip les reduce forecasting error more than country-specific ones.
Arce and Mora (2000) use data from Belgium, France, Germany, Italy, The Netherlands, Switzerland, Spain, and the UK. The y find evidence of significant differences in the stock market valuation of accounting data, mainly explained by differences in reporting philosophies in Europe.
Lin and Chen (2005) investigate the incremental value relevance of the reconciliation of accounts from the Chinese Accounting Standards (CAS) to the IAS. They find that earnings and book values determined under CAS are more relevant accounting information for the purpose of determining the stock price than IAS. However, this might also be due to the specific characteristics of the Chinese market.
directives. The magnitude of the relations is comparable to that of advanced markets. The incremental information content of lagged book value is greater than that of current earnings.
3.2 Return model
Strong (1993) uses UK data to study on the explanatory power of annual earnings for annual stock returns. They find evidence that both earnings levels and earnings differences have significant explanatory power. Changes in earnings turn out to be more important than the level of earnings.
Joos and Lang (1994) investigate the financial statement effects of differences in accounting measurement practice in France, Germany, and the UK, before and after the implementation of the Fourth and Seventh EU directives. Their analysis is based on univariate analyses of financial ratios, returns regressions, and price regressions. They find evidence that measurement practices in Germany are more conservative than those in the UK (and France in the case of net income). However, they find no evidence that accounting data from the UK has a higher association with share price. They also find no evidence that the differences were reduced following implementation of the European Union directives.
Harris, Lang, and Möller (1994) examine the effects of accounting differences in the United States (US) and Germany. The correlation between 18- month returns and annual earnings is generally similar for the US and Germany. The coefficient applied to earnings in Germany is larger than in the US, consistent with a more conservative measurement approach in Germany.
Looking at individual French stocks and portfolios, Dumontier and Labelle (1998) find that stock returns are more linked to earning changes than to earning leve ls. Moreover, they conclude that accounting data prepared under the French accounting principles are not less value-relevant than under US or UK GAAP. Finally they find evidence of a noise-in-earnings effect, this means value-relevant events are not integrated into earnings exactly when they occur.
4. ACCOUNTING SYSTEMS
As stated before unbiased accounting and clean surplus accounting are crucial accounting characteristics for the EBO model. In conservative accounting systems value changes are reflected asymmetrically, value declines are recognized more quickly than value increases. CSR is violated if changes in book value can by-pass income. To find out how conservative the different accounting systems are, I will look at their historical backgrounds. First, I will discuss the historical backgrounds of France, Germany, and The Netherlands, second, I will briefly discuss the developments in the EU to understand the background of the IFRS, and finally, I will use the historical backgrounds to derive my hypotheses.
4.1 Historical background
The accounting systems of France, Germany, and The Netherlands have very distinct historical background s. Following Joos and Lang (1994), three factors are distinguished that caused differences in accounting practice across EU countries: relative importance government in the standard-setting process, the providers of capital, and the influence of taxation.
4.1.1 Relative importance government in standard-setting process:
Government bodies have the tendency to satisfy government needs when setting accounting rules. Private sector bodies are more likely to pursue integration of emerging accounting ideas into the existing accounting system. If private sector bodies exert more influence on determining accounting standards than regulatory requirements, accounting is more likely to address the needs of capital providers. (Ali and Hwang, 2000)
Germany has a Roman Law tradition of highly codified and prescriptive regulations. The financial accounting rules of France are set by government bodies as well (Joos and Lang, 1994). The Dutch GAAP is set by government bodies and is strongly influenced by the relationship of the Dutch profession with the academia and the research conducted therein (private sector bodies) (Ballas and Hevas, 2005).
4.1.2 The providers of capital
and market-oriented systems. In bank-oriented systems, banks fulfill most of the capital needs of a firm. Banks in these systems have concentrated and long-term debt and equity holdings and often have direct access to company information. This implies a focus on conservative accounting and on the balance sheet. Furthermore, no emphasize on public financial statements is needed. Companies in market-oriented systems on the other hand, have numerous diverse investors. Investors do not have easy access to the firm and have to rely on financial accounting disclosures (Ali and Hwang, 2000).
Germany and France are historically bank-oriented. These countries differ in the respect that France also has a focus on uniform accounting and reporting practices and the provision of detailed information to assist the gove rnment in managing the economy. The Netherlands is not as market-oriented as the U.S. and the UK, but in comparison with Germany and France, The Netherlands is market-oriented. (Ali and Hwang, 2000)
4.1.3 The influence of taxation
The accounting systems of some countries reflect tax laws, which are in turn influenced by political, economic, and social objectives. The value relevance of these accounting numbers may be jeopardized, since tax rules do not primarily focus on the needs of capital market participants (Ali and Hwang, 2000). Companies might use tax-minimizing techniques and furthermore, have the incentive to report lower income (Joos and Lang, 1994).
In France and Germany the influence of the tax law has traditionally been strong. In The Netherlands however, a very low alignment exists between tax and financial statements. (Alford et al, 1994)
4.2 Developments in the EU
4.2.1 EU Directives
substantial requirements on information which has to be provided by means of notes (Joos and Lang, 1994). The Seventh Directive determines the identification of groups, scope of group accounts and obligation to prepare, audit and publish group financial statements as well as consolidation-related methods (Haller and Kepler, 2002).
The political parties of the Member States were not ready for this, mainly due to the differing relationships of financial accounting in respect to income and corporate tax determination. When the United Kingdom and Ireland became members in 1973 it became even more difficult, because they introduced the Anglo-Saxon philosophy. To overcome these conflicting interests and views the Member States agreed on a compromise through the incorporation of a considerable number of options from the directives. In the end the proposed degree of comparability and equivalence of financial statements across Europe was not achieved (Haller and Kepler, 2002).
As discussed in the previous section, Joos and Lang (1994) investigated the influence of these directives on the value relevance of accounting numbers of France, Germany, and the United Kingdom. Consistent with Haller and Kepler (2002), they find no change in value relevance after the implementation of the EU directives. Other studies found significant differences between European countries as well (e.g., Arce and Mora, 2002; Ballas and Hevas, 2005; King and Langli, 1998)
4.2.2 IFRS
Since the beginning of the 1990s the conditions in Europe have changed. Increasing cross-border merger and acquisition activities is the most important reason. Trade and investment rates also have grown rapidly. In 1999, Western Europe represented 73% of transaction value of all deals carried out worldwide. In order to fund these trade and investment activities European companies had to gain access to liquid capital markets. Alongside this, rapid globalization caused an increase in demand, especially by investors, for internationally accepted financial information (Haller and Kepler, 2002).
into the International Accounting Standards Board, a highly professional organization supported by industry and governments throughout the world, comparable to the U.S. Financial Accounting Standards Board (FASB). The mandate of the IASB is to produce a single set of high-quality, understandable, and enforceable International Financial Reporting Standards (IFRS). The IFRS include existing IAS as well as standards the IASB issued (Jermakowicz and Gornik-Tomaszewski, 2006).
4.2.3 U.S. GAAP
European companies operating in the U.S. capital market still face the problem of reconciliation to U.S. GAAP. Therefore, in 2002, both the FASB and the IASB pledged to make existing IAS and U.S. GAAP fully compatible as soon as possible. Many standards are already in line with U.S. GAAP and improvements are still made. Even now, the SEC does not accept IFRS financial statements without reconciliation. They want IFRS to be more rigorously interpreted and applied, which involves more uniform auditing procedures, enforcement mechanisms, and regulatory environments worldwide. The SEC proposed a road map toward equivalence between IFRS and U.S. GAAP, this should happen as soon as 2007, but no later than 2009 (Jermakowicz and Gornik-Tomaszewski, 2006).
4.3 Hypotheses
As stated before, the directives did not provide the intended harmonization; an interesting
the focus is on the balance sheet, therefore I expect their book value of equity to be more value relevant than under IFRS and Dutch GAAP, before 2005.
This leads to the following hypotheses:
Hypothesis 1 Before the year 2005 accounting numbers under the Dutch GAAP will be the most value relevant, followed by the French GAAP, and the German GAAP will be the least value relevant.
Hypothesis 2 Before the year 2005 book value will be the most important in Germany and France. In The Netherlands earnings will be more value relevant than in France and Germany.
Hypothesis 3 The value relevance of the accounting numbers from France, Germany, and The Netherlands will converge in the year 2005.
Even before 2005 there were some spontaneous efforts by `global players’ to adopt accounting methods that will improve communication with users in other countries (Canibano and Mora, 2000). This could imply that even before 2005 the value relevance of accounting numbers from France, Germany, and The Netherlands were starting to converge.
5. METHODOLOGY
5.1 Tests and analysis
Following other studies (e.g., Arce and Mora, 2002; Collins, Maydew, and Weiss, 1997; El Shamy and Kayed, 2005; Lin and Chen, 2005; King and Langli, 1998) I will use a simplified version of the EBO valuation model. This leads to the following equations:
Pit = a0 + a1 Eit + a2 BVit + uit (3)
Pit = b0 + b1 Eit + uit (4)
Pit = c0 + c1 BVit + uit (5)
Where:
Pit = firm i’s stock price at the end of year t Eit = earnings per share for firm i during period t
BVit = book value of equity per share for firm i at the end of period t
To compare the explanatory power that earnings and book value have for price, I follow Collins et al. (1997). They decompose total explanatory power into three parts:
R2b¦ e = R2b,e- R2e incremental explanatory power (R2) book value = total R2 - R2
earnings
R2e¦ b = R2b,e- R2b incremental R2 earnings = total R2 - R2 book value
R2com = R2b,e- R2e¦ b - R2b¦ e R2 common to book value and earnings = total R2 – incremental R2 book value - incremental R2 earnings
The relative explanatory power will be assessed as well, that is, whether book value or earnings have greater explanatory power. The sample will be divided in sub-samples by country and by profitability (negative versus positive earnings). Accounting numbers and stock prices at the end of the fiscal year will be used.
An additional analysis will be performed on returns. Returns will be regressed on earnings and change in earnings using the following equations:
Where:
Rj,t = rate of return on security j for period t
xj,t = reported earnings (per share) of firm j over period t
? xj,t = earnings (per share) change between period t and period t-1
uit = error term
Pjt-1 = price per share of firm j at time t-1
5.2 Data and sample selection
TABLE 1 Sample selection
Sample selection Number of Number of Number of
French firms German firms Dutch firms
Amadeus 626 541 120
Missing data 429 363 33
Final sample 197 178 87
As can been seen in table 1 quite some companies were excluded because of missing data, this was almost entirely due to missing data of the year 2005. The reason for this is that I collected the data in 2006, therefore Amadeus did not have financials of 2005 available for most firms yet.
In total 1379 firm years are left for France, 1246 for Germany, and 609 for The Netherlands. For the return model 1182 firm years for France, 1068 for Germany, and 522 for The Netherlands are left. Table 2 presents the descriptive statistics for all years per country. The mean and median stock prices are the highest in France. The Netherlands shows the lowest stock prices on average. The mean and median of the German stock prices is in between. French book values have a high mean in comparison to German book values. However, the standard deviation of the French book values is also much higher. France shows the highest return on average, followed by Germany.
In table 4 the correlation statistics for the variables used in the Return- model are presented. Compared to the correlations between the EBO- model variables, the Return-model correlations are quite small. This means no multicollinearity problems are to be expected in the Return- model regressions.
TABLE 2 Descriptive statistics
Variable Firm - Mean Median Standard Minimum Maximum
years deviation
France
Price 1379 95.63 32.01 334.95 0.07 6025.00
Earnings per share 1379 6.29 1.64 27.54 -84.44 315.03
Book value per share 1379 69.50 17.91 235.22 -4.11 2169.33
Return 1182 0.15 0.07 0.63 -0.90 8.23
Earnings level 1182 0.05 0.06 0.17 -1.98 1.41
Earnings change 1182 0.03 0.01 0.24 -2.26 2.95
Germany
Price 1246 37.27 16.30 70.59 0.40 788.54
Earnings per share 1246 2.24 1.05 10.85 -47.19 242.74
Book value per share 1246 25.52 11.65 50.28 -29.22 784.39
Return 1068 0.11 0.03 0.56 -0.93 4.00
Earnings level 1068 0.03 0.06 0.40 -4.12 3.95
Earnings change 1068 0.07 0.01 0.78 -4.37 17.27
The Netherlands
Price 609 18.07 13.71 15.97 0.19 165.18
Earnings per share 609 1.01 0.90 2.38 -21.85 11.93
Book value per share 609 9.13 6.22 9.15 -8.60 76.88
Return 522 0.08 0.03 0.51 -0.89 4.97
Earnings level 522 0.04 0.07 0.19 -1.65 1.04
Earnings change 522 0.05 0.01 0.43 -1.82 4.99
Notes: Stock price is the price in euros at the end of the book year. Earnings per share is the reported net income divided by the number of shares outstanding at the end of the book year. Book value per share is the end of year book value of equity divided by number of shares outstanding. Return is the yearly return in percentage terms. Earnings level is the earnings per share dividend by the beginning of the period stock price. Earnings change is the change in earnings per share divided by beginning of the period stock price.
TABLE 3
Correlation statistics variables EBO-model
Spearman
Variable Stock Price Earnings per share Book value per share
France
Stock Price 1.000 0.682 0.807
Earnings per share 0.549 1.000 0.669
Book value per share 0.581 0.773 1.000
Germany
Stock Price 1.000 0.585 0.697
Earnings per share 0.180 1.000 0.659
Book value per share 0.506 0.315 1.000
The Netherlands
Stock Price 1.000 0.713 0.687
Earnings per share 0.484 1.000 0.669
Book value per share 0.606 0.545 1.000
Notes: The table reports the average of the correlation coefficients from the yearly samples.
TABLE 4
Correlation statistics Return-model
Spearman
Variable Return Earnings level Earnings change
France Return 1.000 0.435 0.288 Earnings level 0.259 1.000 0.442 Earnings change 0.127 0.177 1.000 Germany Return 1.000 0.425 0.275 Earnings level 0.165 1.000 0.423 Earnings change 0.082 0.310 1.000 The Netherlands Return 1.000 0.471 0.306 Earnings level 0.364 1.000 0.397 Earnings change 0.198 0.376 1.000
6. RESULTS EBO VALUATION MODEL
In this section I will first discuss the results of the average of all years, second the sample where positive and negative earnings are partitioned, and finally the results per year.
TABLE 5
Regression results all years
Sample Earnings Book Adjusted Incr. R2 Incr. R2 R2
value R2 Earnings Bookvalue Common
France -0.509 0.830 0.759 0.012 0.381 0.366 (0.421) (0.000) 5.371 0.378 (0.003) 0.78511 0.747 (0.000) Germany 2.353 0.592 0.455 0.088 0.110 0.257 (0.182) (0.136) 4.266 0.345 (0.011) 0.345 0.367 (0.013) The Netherlands 2.753 0.661 0.463 0.067 0.144 0.252 (0.129) (0.068) 4.515 0.319 (0.029) 0.961 0.396 (0.008)
Notes: The table reports the average of the coefficient estimates and P-values from the yearly cross-sectional regressions. The regressions are estimated using White heteroskedasticity-consistent standard errors and covariance. P-values are in parentheses. Per country the outcomes of the following regressions are shown:
Pit = a0 + a1 Eit + a2 BVit + eit (3)
Pit = b0 + b1 Eit + eit (4)
Pit = c0 + c1 BVit + eit (5)
Incr. R2 Earnings is the incremental value relevance of earnings over book value of equity: R2e¦ b = R2b,e
- R2b. Incr. R2 Book value is the incremental value relevance of book value of equity over earnings:
R2b¦ e = R2b,e - R2e. R2 Common is the value relevance common to both earnings and book value of
equity: R2com = R2b,e - R2e¦ b - R2b¦ e
6.1 All years
level, except for the regression of formula 3 on the German sample. None of the earnings coefficients of formula 3 are significant. In formula 4 the earnings coefficients are significant at the 5% significance level for all countries. The relative value relevance (R-square) for the French sample is the highest. Book value clearly has higher value relevance than earnings. In Germany the relative value relevance of book value and earnings are much closer together. In The Netherlands book value has a slightly higher R-square than earnings. The total value relevance is slightly higher for the Dutch sample than for the German sample.
The incremental value relevance of earnings and book value is also presented in table 5. Germany has the highest incremental value relevance of earnings over book value. France shows the highest incremental value relevance of book value over earnings. France shows the highest common and total value relevance.
6.2 Positive versus negative earnings
Collins, Pincus, and Xie (1999) argue that including book value of equity into the valuation model eliminates the negative price-earnings relation for loss firms. To see whether this holds for my data as well, the samples of the three countries are partitioned into firms with negative earnings and firms with positive earnings. The results are presented in table 6.
TABLE 6
Regression results positive and negative earnings
Sample Firm - Constant Earnings Book F-statistic Adjusted
years value R2 France 28 4.134 0.046 1.272 568.036 0.881 Negative Earnings (0.254) (0.320) (0.000) (0.000) 25.547 -6.186 44.851 0.237 (0.213) (0.189) (0.237) 3.096 1.231 1108.330 0.884 (0.195) (0.000) (0.000) France 169 30.659 0.941 0.681 252.136 0.744 Positive Earnings (0.001) (0.468) (0.067) (0.000) 39.550 6.440 242.058 0.551 (0.001) (0.002) (0.099) 32.595 0.744 474.801 0.731 (0.001) (0.000) (0.000) Germany 38 1.711 -1.446 1.372 25.896 0.497 Negative Earnings (0.266) (0.332) (0.007) (0.030) 10.584 -2.746 10.150 0.156 (0.079) (0.157) (0.136) 4.606 1.512 41.906 0.446 (0.266) (0.003) (0.011) Germany 140 15.410 4.031 0.413 110.907 0.512 Positive Earnings (0.081) (0.146) (0.209) (0.000) 21.787 5.474 167.793 0.442 (0.002) (0.016) (0.000) 16.766 0.861 96.474 0.341 (0.109) (0.017) (0.000) The Netherlands 16 2.217 -1.114 1.023 9.855 0.312 Negative Earnings (0.306) (0.340) (0.184) (0.119) 5.816 -2.339 2.495 0.074 (0.025) (0.283) (0.241) 3.741 1.084 17.065 0.421 (0.282) (0.159) (0.114) The Netherlands 71 7.900 4.516 0.484 52.600 0.488 Positive Earnings (0.002) (0.080) (0.172) (0.000) 9.579 6.518 74.382 0.426 (0.003) (0.004) (0.000) 10.434 0.948 55.898 0.391 (0.000) (0.010) (0.000)
Notes: The table reports the average of the coefficient estimates and P-values from the yearly cross-sectional regressions. The regressions are estimated using White heteroskedasticity-consistent standard errors and covariance. P-values are in parentheses. Per sample the outcomes of the following regressions are shown:
Pit = a0 + a1 Eit + a2 BVit + eit (3)
Pit = b0 + b1 Eit + eit (4)
Book value of equity shows the same results, when earnings are included the coefficient is insignificant and when book value is the only independent variable the coefficient is positive and significant at the 5% significance level. These findings are inconsistent with the findings of Collins et al. (1999).
The Netherlands shows slightly different results. In the negative earnings sample all coefficients are insignificant. The sample size is the smallest of all samples, which most likely influenced the results. In the positive earnings sample the coefficient of earnings is positive and significant when book value is included. The coefficient of book value is insignificant when earnings are included and significant at the 1% significance level when book value is the only independent variable. The relative value relevance of earnings is higher than the value relevance of book value of equity.
Table 7 shows the incremental value relevance of the positive and negative earnings samples. In the negative earnings sample of France the stock price is indeed almost entirely dependent on book value. The positive earnings sample of The Netherlands shows that the incremental value relevance of earnings and book value are almost the same. The incremental value relevance common to both earnings and book value is quite high for all positive earnings samples. The positive earnings samples show higher R-squares for earnings compared to the all years sample (see table 5). The incremental value relevance of book value is lower; this could indicate that book value is indeed important for negative earnings samples. The common value relevance of the positive earnings samples is higher for all countries.
TABLE 7
Incremental value relevance of positive and negative earnings samples
R2 R2 R2 R2
Earnings Bookvalue Common Total
France Negative -0.002 0.645 0.239 0.881
Positive 0.013 0.193 0.538 0.744
Germany Negative 0.050 0.340 0.106 0.497
Positive 0.171 0.070 0.271 0.512
The Netherlands Negative -0.109 0.238 0.183 0.312
Positive 0.097 0.062 0.329 0.488
Notes: Negative means the negative earnings sample and positive means the positive earnings sample.
Incr. R2 Earnings is the incremental value relevance of earnings over book value of equity: R2e¦ b = R2b,e
- R2b. Incr. R2 Book value is the incremental value relevance of book value of equity over earnings:
R2b¦ e = R2b,e - R2e. R2 Common is the value relevance common to both earnings and book value of
6.3 Samples per year
Table 10 shows the regression results per year for all three countries. I will first discuss the coefficients, the significance and the relative explanatory power of the regressions per country and thereafter I will analyze the incremental explanatory power.
6.3.1 Coefficients and relative value relevance
For France the coefficient on book value of equity is significant at the 1% significance level in all years. The coefficient is between 0.7 and 1.0 in the years prior to 2005 and goes down to 0.5 in 2005. The earnings coefficient however, is insignificant and negative for most of the years prior to 2005 when book value is included. In 2005 the coefficient is positive, but still insignificant. Both earnings and book value of equity show positive and significant (at the 5% significance level) coefficients when they are the only independent variables. The relative value relevance (R-square) of earnings declines in the years up to 2004 and in 2005 the value relevance increases to 45%. The value relevance of book value of equity does not change much across the years. The value relevance when both earnings and book value of equity are included does not change much either, in 2005 the total value relevance is 77%. In all years the relative value relevance of book value is higher than the value relevance of earnings.
For Germany the coefficient on book value of equity is significant at the 1% significance level in all years except for 2003. The coefficient changes from year to year and in 2005 is has a value of 0.6. When book value is included, the earnings coefficient is significant for the years 2002, 2003, and 2005 at the 1% significance level. The total value relevance increases from 29% in 1999 to 69% in 2005. The relative value relevance of earnings increases from 8% to 61% over the years. Book value of equity shows an increase in value relevance from 30% in 1999 to 56% in 2005. This means that earnings have become more important in comparison to book value through the years.
TABLE 8
Regression results per year; relative and incremental information content earnings and book value
Year Firm years Constant Earnings Book F-statistic Adjusted Earnings Adjusted Book Adjusted Incr. R2 Incr. R2 R2
value R2 R2 value R2 Earnings Bookvalue Common
TABLE 8 (continued)
Year Firm years Constant Earnings Book F-statistic Adjusted Earnings Adjusted Book Adjusted Incr. R2 Incr. R2 R2
value R2 R2 value R2 Earnings Bookvalue Common
The Netherlands 1999 87 13.367 4.234 0.504 10.351 0.179 6.030 0.165 0.987 0.132 0.047 0.014 0.118 (0.000) (0.083) (0.357) (0.000) (0.001) (0.048) 2000 87 12.964 2.412 0.526 15.480 0.252 3.778 0.197 0.805 0.201 0.051 0.055 0.146 (0.000) (0.068) (0.121) (0.000) (0.001) (0.008) 2001 87 10.684 0.271 0.698 15.398 0.251 0.754 0.042 0.734 0.253 -0.003 0.209 0.045 (0.000) (0.677) (0.001) (0.000) (0.197) (0.000) 2002 87 4.783 0.882 0.783 43.045 0.494 1.930 0.230 0.920 0.460 0.034 0.265 0.196 (0.000) (0.011) (0.000) (0.000) (0.003) (0.000) 2003 87 6.076 1.741 0.749 55.038 0.557 3.717 0.335 0.948 0.509 0.048 0.221 0.287 (0.000) (0.065) (0.000) (0.000) (0.000) (0.000) 2004 87 6.088 4.904 0.577 109.312 0.716 7.519 0.629 1.078 0.573 0.143 0.086 0.486 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) 2005 87 6.845 4.829 0.787 161.907 0.789 7.875 0.632 1.256 0.641 0.149 0.158 0.483 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)
Notes: The regressions are estimated using White heteroskedasticity-consistent standard errors and covariance. P-values are in parentheses. Per sample the outcomes of the following regressions are shown:
Pit = a0 + Eit + a2 Bvit + eit 3
Pit = b0 +a1 Eit + eit 4 Pit = c0 + c1 Bvit + eit 5
Incr. R2 Earnings is the incremental value relevance of earnings over book value of equity: R2e¦ b = R2b,e - R2b. Incr. R2 Book value is the incremental value
relevance of book value of equity over earnings: R2b¦ e = R 2
b,e - R 2
e. R 2
6.3.2 Incremental value relevance
The incremental value relevance is given in table 8 and graphically in figure 1. Figure 1 contains one cumulative line graph per country where the different contributions to total value relevance of incremental value relevance of earnings, book value, and the common value relevance over the years are shown.
The total value relevance of France does not change much over the years. However, the different contributio ns of earnings, book value and common value relevance do. The incremental value relevance of book value increases (from 10% to 58%) at the expense of common value relevance (from 65% to 14%) until 2002. In the years 2003 to 2005 it is the other way around: common value relevance increases to 42% at the expense of incremental value relevance of book value (to 33%). The incremental value relevance of earnings is near zero in the years prior to 2004. In 2004 and 2005 the incremental value relevance of earnings is about 2.5%.
In Germany the total value relevance increases over the years. The year 2002 shows a rapid increase in total value relevance, almost entirely caused by an increase in incremental value relevance of earnings and common value relevance. In 2003 a decrease of all variables can been seen. The last two years common value relevance increases to 49%.
The Netherlands shows a large increase in total value relevance. To a large extend caused by an increase from 12% to 48% in value relevance common to both earnings and book value. Furthermore, incremental value relevance of book value also increases from 1% to 16%. The incremental value relevance of earnings shows an increase from 2002 on, and in 2005 it reaches the highest level with 15%.
Figure 2 contains the development of the incremental value relevance of earnings, book value and common value relevance over the years. The graph of incremental value relevance of earnings shows an increase for The Netherlands. Germany and France show no clear pattern. In 2005 however, it seems that the relevance is higher than the years before for all countries. Germany and The Netherlands come closer together.
FIGURE 1
Total and incremental value relevance of earnings and book value
France 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 1999 2000 2001 2002 2003 2004 2005 year R-square
Earnings Book value Total
Germany 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 1999 2000 2001 2002 2003 2004 2005 year R-square
Earnings Book value Total
The Netherlands 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 1999 2000 2001 2002 2003 2004 2005 year R-square
Earnings Book value Total
FIGURE 2
Common and incremental explanatory power of earnings and book value
Incremental value relevance of Earnings
0.000 0.050 0.100 0.150 0.200 0.250 0.300 1999 2000 2001 2002 2003 2004 2005 year R-square
France Germany The Netherlands
Incremental value relevance of Book Value
0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 1999 2000 2001 2002 2003 2004 2005 year R-square
France Germany The Netherlands
Common value relevance
0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 1999 2000 2001 2002 2003 2004 2005 year R-square
France Germany The Netherlands
The value relevance common to earnings and book value increases for both Germany and The Netherlands. France acts differently again, by showing a downward movement in the years 1999 to 2002. In 2003, however the common value relevance starts to increase. In 2005 the common value relevance of all three countries seem to converge.
6.3.3 Summary
I expected The Netherlands to provide the most value relevant accounting numbers followed by France and finally Germany for the years prior to 2005. I expected book value to be the most important in Germany followed by France. Earnings would be more relevant for The Netherlands. In 2005 I expected the value relevance of the accounting numbers from the different countries to converge.
The results for France show that the total value relevance is relatively high in the beginning and does not change much during the years. The incremental value relevance of book value is a bit lower than I expected in 1999, but increases in the following years. After 2002 the incremental value relevance of book value declines. Earnings are not of much relevance in all years.
For Germany the results prior to 2005 are not as expected. Book value is of less importance than expected. Earnings are more value relevant than I expected in the years 2002 and 2003. However, in 2005 the common and incremental explanatory power of earnings and book value seem to converge towards The Netherlands.
The accounting numbers of The Netherlands show lower total value relevance than I expected in 1999. Thereafter, the total value relevance increases to a level higher than France and Germany. The value relevance of earnings decreases the first two years, but after that it increases. The value relevance of book value increases over the years, except for 2004 where a decrease can be seen.
7. RESULTS RETURN MODEL
In this section the regression results of the return model are shown. First, I will discuss the results of the sample with all years together and second, the results of the regressions per year.
7.1 All years
Table 9 presents the average of the yearly cross-sectional regressions of the return- model. When both earnings level and earnings change are included only France shows significant coefficients at the 1% significance level. R-square is rather low with 12%. The Netherlands shows a higher R-square, namely 22%. When using formula 7, The Netherlands is the only country which shows a significant coefficient at the 1% significance level. The R-square is the highest as well with 16%. When only earnings change is used, none of the countries show significant coefficients.
TABLE 9
Regression results all years
Sample Earnings Earnings Adjusted Earnings Adjusted Earnings Adjusted
level change R2 level R2 change R2
France 0.837 0.162 0.117 0.640 0.081 0.246 0.020 (0.086) (0.074) (0.123) (0.225) Germany 0.209 0.008 0.020 0.188 0.023 0.066 0.000 (0.203) (0.603) (0.205) (0.383) The Netherlands 0.830 -0.073 0.217 0.526 0.162 0.314 0.095 (0.160) (0.423) (0.068) (0.433)
Notes: The table reports the average of the coefficient estimates and P-values from the yearly cross-sectional regressions. The regressions are estimated using White heteroskedasticity-consistent standard errors and covariance. P-values are in parentheses. Per sample the outcomes of the following regressions are shown:
(6)
(7)
(8)
Where: Rj,t is the rate of return on security j for period t , xj,t is the reported earnings of firm j over
7.2 Samples per year
Table 10 shows the regression results per year for all three countries. For France the earnings level coefficient is significant at the 1% significance level for all years except for the years 2000, 2003, and 2004. The coefficient for earnings change is slightly insignificant in the years 2004 and 2005, but significant at the 5% significance level for the remaining years. The total value relevance is quite low, except for 2002 with an R-square of 24%. In 2005 the total value relevance equals 11%.
For Germany the coefficient of the earnings level is significant at the 1% significance level for the years 2000 and 2001. However, the coefficient is ins ignificant for the remaining years. The coefficient of the earnings change is insignificant for all years. It is dangerous to draw conclusions from the value relevance because of the many insignificant coefficients. What can be seen is that the values are very low.
In the Dutch sample the coefficient of the earnings level is insignificant for 2003 and 2005. The remaining years show significant coefficients at the 5% significance level. The coefficient on the earnings change is only significant for the years 2000 and 2003 at the 0.01 level, showing a negative and a positive relationship respectively. The R-squares are rather high compared to France and Germany; with the highest level in 2003 with 44%. In 2005 however, no value relevance can be found.
TABLE 10
Regression results per year
Year Firm Constant Earnings Earnings Adj. Earnings Adj. Earnings Adj.
years level change R2 level R2 change R2
France 2000 197 0.273 0.461 0.832 0.006 0.075 -0.004 0.219 -0.002 (0.000) (0.109) (0.044) (0.481) (0.383) 2001 197 -0.125 1.226 -0.364 0.104 0.679 0.070 0.087 0.000 (0.000) (0.000) (0.000) (0.003) (0.197) 2002 197 -0.164 1.358 -0.478 0.238 0.860 0.144 -0.035 -0.004 (0.000) (0.000) (0.002) (0.034) (0.558) 2003 197 0.193 0.201 0.393 0.049 0.374 0.025 0.477 0.047 (0.000) (0.227) (0.016) (0.013) (0.026) 2004 197 0.232 0.935 0.471 0.194 0.966 0.141 0.507 0.061 (0.000) (0.147) (0.150) (0.178) (0.160) 2005 197 0.158 0.842 0.120 0.110 0.886 0.108 0.221 0.017 (0.000) (0.033) (0.229) (0.026) (0.023) Germany 2000 178 0.030 0.278 -0.019 0.003 0.264 0.009 0.090 -0.002 (0.528) (0.056) (0.711) (0.036) (0.175) 2001 178 -0.094 0.512 -0.089 0.086 0.445 0.085 0.079 0.001 (0.002) (0.000) (0.306) (0.016) (0.380) 2002 178 -0.229 0.099 -0.004 0.021 0.094 0.026 0.028 0.006 (0.000) (0.201) (0.869) (0.101) (0.122) 2003 178 0.413 0.120 0.062 0.000 0.057 -0.003 0.032 -0.003 (0.000) (0.296) (0.254) (0.552) (0.590) 2004 178 0.214 0.180 0.053 0.008 0.185 0.013 0.072 -0.003 (0.000) (0.336) (0.809) (0.350) (0.733) 2005 178 0.270 0.067 0.045 0.000 0.085 0.005 0.093 0.001 (0.000) (0.326) (0.666) (0.176) (0.295) The Netherlands 2000 87 -0.176 2.461 -0.825 0.208 0.199 0.192 0.534 0.004 (0.001) (0.000) (0.010) (0.000) (0.383) 2001 87 -0.168 0.722 -0.209 0.175 0.571 0.273 0.355 0.055 (0.000) (0.030) (0.572) (0.000) (0.031) 2002 87 -0.309 0.557 0.060 0.206 0.521 0.201 0.001 -0.012 (0.000) (0.002) (0.251) (0.004) (0.985) 2003 87 0.274 -0.150 0.725 0.441 0.507 0.019 0.707 0.448 (0.000) (0.656) (0.000) (0.149) (0.000) 2004 87 0.186 0.885 -0.020 0.271 0.869 0.280 0.392 0.087 (0.001) (0.001) (0.918) (0.000) (0.325) 2005 87 0.302 0.506 -0.172 0.000 0.493 0.010 -0.102 -0.011 (0.000) (0.269) (0.784) (0.257) (0.871)
8. CONCLUSION AND DISCUSSION
In this study I investigate the impact of switching to IFRS on value-relevance under the EBO valuation model and the return-earnings model. I use data from France, Germany, and The Netherlands to conduct the research on. The French, German, Dutch, and IFRS accounting systems have completely different historical backgrounds, which are expected to cause differences in the value relevance of the accounting numbers under the different systems.
The results show fewer differences between Germany and The Netherlands than I expected. France on the other hand does show substantial differences in value relevance in comparison with Germany and The Netherlands. In France the relevance of book value is relatively more important than earnings in all years. Total value relevance does not change much and earnings become more important towards 2005, conform to my expectations. Germany shows lower relevance of book value than I expected. The value relevance does increase through the years, but so does the relevance of earnings. The Netherlands shows an increase of value relevance of earnings and book value as well. The total explanatory power is lower than I expected, but increases through the years to a level just above France and Germany. The total value relevance of especially France and The Netherlands does not differ much in 2005. The Netherlands shows the highest total value relevance (79%), followed by France (77%), and finally Germany (69%).
The incremental value relevance of earnings, book value, and value relevance common to earnings and book value show a clear converging pattern for Germany and The Netherlands. The incremental value relevance of earnings seems a bit low for France and the incremental value relevance of book value a bit high. The value relevance common to earnings and book value is almost equal for all three countries in 2005.
Collins et al. (1997) found a decline in value relevance of earnings in the US. They also found that much of this decline could be explained by the increasing frequency and magnitude of one-time items, the increasing frequency of negative earnings, and changes in average firm size and intangible intensity across time. Differences in these items between France, Germany, and The Netherlands could explain the obtained results. The positive earnings sample of France already showed a higher value relevance of earnings. However, further research is needed to draw conclusions.
sample of all years together you can see that the earnings level is the most important explanatory variable of returns for all the countries. The Netherlands shows the highest value relevance followed by France. Germany shows the lowest explanatory power.
According to Dumontier and Raffournier (2002) many other authors found low R-squares for the return model. They claim that there should be a lag in the inclusion of new information into earnings, and stock prices should be more prompt than earnings in reflecting new information. To correct for this lag effect several authors have expanded both the returns and earnings window by regressing multiple-year returns on multiple- year earnings. The results improved significantly. Unfortunately, I am not able to apply this methodology to my data, since there is no data for the years after 2005 yet. Therefore, I have to leave this for future research. Some of the samples I used were rather small, especially for The Netherlands. In the future more countries could be used and a longer time period in order to improve the results.
REFERENCES
Alford, A., Jones, J., Leftwich, R., and M. Zmijewski (1993) “The relative informativeness of accounting disclosures in different countries”, Journal of accounting research, 31, pp. 183-223
Ali, A. and L. Hwang (2000) “Country-specific factors related to financial reporting and the value relevance of accounting data”, Journal of accounting research, 38(1), pp.1-21
Arce, M. and A. Mora (2002) “Empirical evidence of the effect of European accounting differences on the stock market valuation of earnings and book value”, The European
accounting review, 11(3), pp. 573-599
Ballas, A.A., and D.L. Hevas (2005) “Differences in the valuation of earnings and book value: Regulation effects or industry effects?”, The International Journal of Accounting, 40, pp. 363-389
Bartov, E., Goldberg, S.R. and M. Kim (2005) “Comparative value relevance amo ng German, U.S., and International Accounting Standards: A German stock market perspective”, Journal
of accounting, auditing, and finance, 20(2), pp. 95-119
Canibano, L. and A. Mora (2000) “Evaluating the statistical significance of de facto accounting harmonization: a study of European global players”, The European accounting
review, 9(3), pp. 349-369
Collins, D.W., Maydew, E.L., and I.S. Weiss (1997) “Changes in the value-relevance of earnings and book values over the past forty years”, Journal of Accounting and Economics, 24, pp. 39-67
Collins, D.W., Pincus, M., and H. Xie (1999) “Equity valuation and negative earnings: the role of book value of equity”, The accounting review, 74(1), pp. 29-61
Dumontier, P. and B. Raffournier (2002) “Accounting and capital markets: a survey of the European evidence”, The European accounting review, 11(1), pp. 119-151
El Shamy, M.A., and M.A. Kayed (2005) “The value relevance of earnings and book values in equity valuation: an international perspective, the case of Kuwait”, International journal of
commerce and management, 14(1), pp. 68-79
Gornik-Tomaszewski, S.G., and E.K. Jermakowicz (2001) “Accounting-based valuation of Polish listed companies”, Journal of international financial management and accounting, 12(1), pp. 50-74
Graham, R.C. and R.D. King, (2000) “Account ing practices and the market valuation of accounting number: evidence from Indonesia, Korea, Malaysia, the Philipines, Taiwan, and Thailand”, The international journal of accounting, 35(4), pp. 445-470
Haller, A., and J. Kepler (2002) “Financial accounting developments in the European Union: past events and future prospects”, The European Accounting Review, 11(1), pp. 153-190
Harris, T.S., Lang, M. and H.P. Möller (1994) “The value relevance of German accounting measures: an empirical analysis”, Journal of accounting research, 32(2), pp. 187-209
Jermakowicz, E.K. and S. Gornik-Tomaszewski (2006) “IFRS and you, what are the implications of the European accounting revolution?”, Strategic finance, 87(9), pp. 43-48
Joos, P. and M. Lang (1994) “The effects of accounting diversity: evidence from the European Union”, Journal of accounting research, 32, pp.141-168
King, R.D. and J.C. Langli (1998) “Accounting diversity and firm valuation”, The
international journal of accounting, 33(5), pp. 529-567
Lin, Z.J. and F. Chen (2005) “Value relevance of international accounting standards harmonization: evidence from A- and B-share markets in China”, Journal of international
Ohlson, J.A. (1995) “Earnings, book values, and dividends in equity valuation”,
Contemporary accounting research, 11(2), pp.661-687
Ross, S.A., Westerfield, R.W., and J.Jaffe, 2005, Corporate finance, New York: McGraw-Hill Companies, seventh edition.
Strong, N. (1993) “The relation between returns and earnings: evidence for the UK”,
