Changes in the accounting for goodwill: Is Impairment more value relevant than Amortization?

Transcription

1 Changes in the accounting for goodwill: Is Impairment more value relevant than Amortization? Master Thesis Laye Mory Kourouma First supervisor: Dr. G. Georgakopoulos Second supervisor: Dr Sanjay Bissessur First draft: August Final draft: August Master Accountancy & Control Faculte Economie en Bedrijfskunde Universe van Amsterdam 1

2 Acknowledgements I would like to acknowledge everybody who gives me assistance in the completion of this master thesis. I am very grateful to my supervisor, Dr. G. Georgakopoulos, for his support, useful comments and input throughout all the thesis process. I also want to thank my second supervisor, Dr. S. Bissessur for his valuable comments. Finally, my special thank to the exam commtee for allowing me to finish my thesis. 2

3 Contents Pages Acknowledgements 2 Abstract 4 Introduction 5 Lerature Review 8 Research Design 12 Hypothesis 12 Description of the models 12 Periods Observed 15 Data Sources 15 Sample Selection 16 Results 21 Descriptive Statistics 21 Correlations 26 Regression Analyses 27 Conclusion 35 References 36 3

4 Abstract The principal aim of this thesis is to investigate whether goodwill impairment is more value relevant than goodwill amortization. To reach this objective, I use the Ohlson model (Ohlson, 1995) to test the hypothesis stating that goodwill impairment is more value relevant than goodwill amortization. To provide empirical evidence for the testing of the hypothesis, I examined 4 samples ( , , and ). The sample consist of European companies that only apply goodwill amortization and the three other samples consist of European companies that apply goodwill impairment. There are 4 samples because I segregate the sample period from 2005 to 2013 into 3 sub-periods. By doing so all the sub- periods will have the same length, this makes possible to analyze the period deeply by comparing the sub- periods whin this period wh each other but also to compare easily the sample for goodwill amortization ( ) wh the samples for goodwill impairment ( , , and ) because the period length of the samples are the same. The results show that goodwill impairment is more value relevant than goodwill amortization in the first six years after adoption of IFRS 3 and that this value relevance is decreasing as time past to the level that is not more relevant than goodwill amortization after nine years. Further, is remarkable that the contribution of goodwill impairment in the explanatory power decreases over the time while that of goodwill amortization increases. An unexpected finding is the higher significance and contribution of the variable net income in the explanatory power of both equations (1) and (2). 4

5 1. Introduction Goodwill is defined as "the excess of the cost of the acquired company over the sum of the amounts assigned to identifiable assets acquired less liabilies assumed" (APB Opinion No. 16, Paragraph 87). This is one of the many definions of a well-known and very old term in the academic and accounting world. Leake (1914, p. 81) pointed out that the word Goodwill has been in commercial use for centuries and ced a reference from the year Below, you can find some of the many definions of the goodwill which are almost the same. Goodwill is the excess of the cost of an acquired enterprise over the net of the amounts assigned to assets acquired and liabilies assumed (CICA, 2002). In his paper A polical Economy of SSAP22: Accounting for goodwill Bryer (1995, p. 286) defined goodwill as the difference between the market value of an enty at any point in time and the fair value (eher the replacement cost, recoverable amount or net realizable value) of s net assets, including any separately identifiable intangible assets (such as trademarks, copyrights, patent rights) Goodwill is the name accountants give to the difference between the price an acquirer pays for a business, and the value of the individual identifiable assets acquires (Higson, 1998, p. 141) Goodwill is the advantage or benef, which is acquired by an establishment, beyond the mere value of the capal stock, funds, or property employed therein, in consequence of general public patronage and encouragement, which receives from constant or habual customers, on account of s local posion, or common celebry, or reputation for skill or affluence, or punctualy, or from other accidental circumstances or necessies, or even from ancient partialies or prejudices. Goodwill is the value of a business or practice that exceeds the combined value of the net assets. Essentially, goodwill is the favor won by the management of a business from the public, and probabily that old customers will continue their patronage (Jerrold and Richards, 2005, p. 389) Goodwill is the value attributed to such intangible assets (among others) as reputation, welltrained workforce, good contacts whin the industry, favorable business location, and any 5

6 other unique features of the company for which another company would pay in excess of the value of net assets shown in the balance sheet (Barber, 2001). There are two different kinds of goodwill: Goodwill developed internally whin the firm and purchased goodwill, which is acquired by the firm during the purchase or takeover of another firm. Goodwill developed internally consist of diverse factors like a company s good name and reputation, an established customer base, dependable suppliers and employees, and the abily to obtain more value from assets than competors (Henderson and Peirson, 1988; Jennings, LeClere, and Thompson, 2001). Goodwill and s proper accounting treatment have been debated over the past years resulting in some accounting major changes worldwide. Both the FASB and IASB modified the rules to require impairment rather than amortization of goodwill. For example the issuance of SFAS No. 142 in June 2001 in the U.S. wh an effective date for fiscal years beginning after December 15, 2001; and the adoption of IFRS in 2005 in Europe. In 2005 the IASB (2005) adopted IFRS 3. This adoption completely changes the accounting treatment for goodwill in diverse countries around the world. Goodwill impairment replaces goodwill amortization because FASB (2001) and IASB (2005), respectively the accounting standard-setting bodies in the U.S. and Europe, claim that goodwill impairment is more value relevant than goodwill amortization for the market valuation of the companies. The new rules are meant to provide a better picture of the value of the goodwill on the financials. However, may be questionable whether these new rules are better than the old ones. The principal aim of the present thesis is to determine whether goodwill impairment is more value relevant than goodwill amortization. Therefore, the research question is the following: Is impairment of goodwill more value relevant than amortization of goodwill? The results show that goodwill impairment is more value relevant than goodwill amortization in the first six years after adoption of IFRS 3 and that this value relevance is decreasing as time past to the level that is not more relevant than goodwill amortization after nine years. Further, is remarkable that the contribution of goodwill impairment in the explanatory power decreases over the time while that of goodwill amortization increases. An unexpected finding is the higher significance and contribution of the variable net income in the explanatory power of both equations (1) and (2). 6

7 The remaining of this thesis is structured as follows. The following section, present a summary of the previous lerature. I then discuss my research method (including data collection, sample selection and variable definions), present the results and, finally, outline the thesis s conclusions limations, and future research opportunies. 7

8 2. Lerature Review Several studies have been conducted on the value relevance of goodwill impairment and amortization in different ways. Jennings, LeClere, and Thompson II (2001) examined whether excluding goodwill amortization from the computation of earnings increases the usefulness of earnings data to investors. They found that earnings before goodwill amortization explain significantly more of the cross-sectional variation in share prices than earnings after goodwill amortization. Further their analysis provided evidence that, even when disaggregated from the remainder of reported earnings, goodwill amortization provides no explanatory power for observed prices beyond that of earnings before goodwill amortization. Overall, their results indicate that when investors approximate share value by capalizing accounting earnings, perhaps as a prelude to further analysis, the goodwill amortization component of reported earnings can best be viewed as a source of noise. Thus, excluding goodwill amortization from corporate income statements under the new rules will not reduce the usefulness of earnings but, rather, may eliminate a source of noise in earnings as measured under previous standards. The paper of Jennings, LeClere, and Thompson II (2001) provide evidence that goodwill amortization is irrelevant to investors for their decision making process which is in line wh the FASB and IASB s point of view but contradictive to the results of Jennings et al. (1996). In their paper Goodwill and Amortization: Are they Value Relevant? Churyk and Chewning Jr. (2003) investigate if goodwill is valued as an economic resource by the equy markets and, if so, whether or not goodwill amortization and the market value of the company are related to each other. The results of their analysis of data from the years 1992 through 1996 indicate that the relation between goodwill and the market value of equy is posive for all years examined and statistically significant in four of the five years. These results are consistent wh their predictions and previous research and imply that investors value recorded goodwill as an asset. The relation between goodwill amortization and the market value of equy for the same period is negative, as predicted, and statistically significant in some years and in the analysis of data pooled across years. They interpret these findings to indicate that the market views goodwill as an economic resource wh a declining value, and the decline in value is associated wh the amortization recognized. The introduction of Financial Reporting Standard (FRS) 11 in the UK in 1998 (which allowed an annual impairment review as an alternative to capalization and subsequent systematic 8

9 amortization of goodwill) has been the reason for Li, Amel-Zadeh and Meeks (2010) to study the value relevance of goodwill impairment and the information content of impairment announcements. Their findings show that goodwill impairment leads to economically significant reductions in market value of the company. They also provide evidence that market react negatively to goodwill impairment announcements. The negative effect is greater for companies wh a higher proportion of assets carried as goodwill and for companies that release ltle information prior to the announcement. Their findings provide also a confirmation on prior evidences that capalized goodwill has a significant relation market value in the year of acquision, but this relation decreases thereafter; and that capalized goodwill is value irrelevant when is systematically amortized. The relation between accounting goodwill numbers and equy values are examined by Jennings et al. (1996). They find a negative relation between equy value and goodwill amortization meaning that investors value amortization as relevant to their decision making process but investors also consider amortization as an accounting number that decreases the value of goodwill, the value of the company and therefore also s share price. But they also recognize the weakness of their results indicating that the value decreasing of the goodwill is not the case for all companies or not as fast as assumed by the expected economic lifetime. This limation does not really impact the evidence provided by the paper that goodwill amortization is value relevant to investors for their decision making process. Using for their research, a model comparing the explanation power of the different earnings numbers, Moehrle et al. (2001) investigate the information content of goodwill accounting numbers. In contrast wh the results of Jennings et al. (2001), they show that there is no significant difference between the explanatory power of earnings numbers excluding or including amortization of goodwill. This means that amortization is not a disturbing factor but their finding that goodwill amortization is not relevant to investors is in line wh the results of Jennings et al. (2001), as including goodwill amortization does not increase the explanatory power of the model. In their paper The Impairment of Purchased Goodwill: Effects on the Market Value Li and Meeks (2006) investigate the value relevance of goodwill in the UK. In line wh the results of of Bugeja and Gallery (2006) they show that goodwill is value relevant but that this value decreases over time. Another finding of their research is that amortization is value irrelevant. 9

10 This confirms the claim of the IASB and FASB that the goodwill impairment is more value relevant than goodwill amortization. By using a valuation model, Chambers (2006) studies the effects of SFAS 142 on financial reporting; He examined both the value relevance of impairments and the effects of the elimination of goodwill amortization. In line wh objectives of FASB, He finds that financial reporting qualy has improved by using impairments but that the qualy of reporting is lower as consequence of the elimination of the amortization method. The main conclusion of Chambers (2006) is that concludes that the accounting qualy is increased because of the introduction of SFAS 142. Based on the Exposure Draft that the FASB issued on SFAS 142, Churyk (2005) examines whether the elimination of the amortization of goodwill is appropriate and compares market valuations of goodwill. Her results provide evidence that goodwill at acquision is almost never overvalued indicating that systematic amortization is not needed. But she shows that goodwill decrease in value in subsequent periods. Further, (Churyk, 2005) mentions that impairment of goodwill can be done when stock price decreases or when the book value of equy is greater than the market value of the company. Analyzing the effects of the adoption of IFRS on the Swedish stock market, Hamberg et al. (2006) show that goodwill is more persistent under the impairment method than under the amortization method. Further, the test also the value relevance of the reported goodwill by using a trading strategy in which they buy stocks wh relative high amortization costs and sell stocks wh relative low amortization costs. They find that abnormal returns are earned only if investors did not incorporate the changes already in prices, because of the longer persistence of goodwill under the impairment method of accounting. The investors see the higher persistence of goodwill as new information, but the results are not significant. Overall, their conclusion is that the introduction of IFRS led to more relevant accounting information. By comparing the association between goodwill accounting charges against income and firms economic investment opportunies in amortization and impairment regimes, Chalmers et al. (2011) examine in their article Does a goodwill impairment regime better reflect the underlying economic attributes of goodwill? the claim of Accounting standard-setting bodies that an impairment regime better reflects the underlying economic value of goodwill than 10

11 systematic amortization. Their results show that the association between firms goodwill charges against income and the firms investment opportunies is stronger during the IFRS regime than the AGAAP regime meaning that the claim is justified. In general, the results of the above studies indicate that goodwill impairment is more value relevant than goodwill amortization. This thesis contributes to the existing lerature by adding new features in several ways. First, extends and complements existing lerature by investigating the value relevance of the impairment and amortization of goodwill for the share price and the market value of the company. Second, this study adds insight to the ongoing debate over the impairment and amortization of goodwill by analyzing the impairment data of many European companies from the beginning of the effectiveness of IFRS 3 in 2005 until Third, the research in this thesis is international (the examined samples are made up of companies from different European countries). Fourth, spans the entire implementation period of goodwill impairment. 11

12 3. Research Design This section describes the research design used to investigate the research question. The first sub-section presents the tested hypothesis. The sub-section Models discuss the models used to test the hypothesis. Further, the data and samples selection is outlined. 3.1 Hypothesis The motivation of accounting standards setters such as the FASB (2001) and the IASB (2005) for issuing new accounting standards is the improvement of accounting information qualy. An example of new accounting standards issue is the adoption of IFRS 3 in The IASB (2005) introduces IFRS 3 because claims that goodwill impairment is more value relevant than goodwill amortization. Therefore, the main objective of this thesis is to investigate if this claim is justified. So the hypothesis of this thesis is as follows: H: Goodwill impairment is more value relevant than goodwill amortization. 3.2 Description of the Models Research on value-relevance investigate the association between a secury price-based dependent variable and a set of accounting variables ; if an accounting number is significantly related to the dependent variable, then is regarded as value relevant (Beaver, 2002, p.459). Most of this kind of studies makes use of an accounting-based valuation model like developed in Ohlson (1995) and s later various versions (Barth, et al., 2001). The Ohlson model (Ohlson, 1995) interprets the market value of a company as a function of the book value of equy and earnings. So, I used the Ohlson model (Ohlson, 1995) to test the above hypothesis. The original Ohlson model (Ohlson, 1995) is: MVE = 0 + β1bve + β2 β NI + ε In above equation: MVE represents the market value of the firm s equy, time t; BVE is the book value of the firm s equy, time t; and 12

13 NI is the net income of the firm, time t; The explanatory power of the model is the indicator of the value relevance. Investors will use accounting information to make decision if they think that is useful. Useful accounting information has a posive and strong wh the share price and market value of the firm meaning that the explanatory power of the model will be high in this case. But the explanatory power will be low if accounting information is not useful for investors to make decisions. Ohlson model (Ohlson, 1995) has the advantage is that allows other variables to be added to the basic model. This makes comparison between different models and different explanatory powers possible. The comparison helps to make a conclusion on the value relevance of accounting information in general and goodwill amortization and impairment in particular. For this reason, The Ohlson-model (Ohlson, 1995) is the appropriate model for the research question of this thesis, because is able to compare the value relevance of a model containing the variable goodwill amortization wh a model containing the variable goodwill impairment. The hypothesis of this thesis is tested using a modified basic Ohlson-model (Ohlson, 1995). To compare the value relevance of goodwill impairment wh that of goodwill amortization, I set up two models, one containing the variable goodwill amortization and the other containing the variable goodwill impairment. The equation of the model containing the variable goodwill amortization is as follows: MVE = 0 + β1bve + β2ni + β3 β AMORT + ε (1) In this equation (1), I defined AMORT as the amount of goodwill amortization time t. MVE, BVE and NI have been defined earlier in the original Ohlson model (Ohlson, 1995). To calculate the market value of the firms equy, I multiply the number of shares outstanding by the share price of three months after the fiscal year-end. By doing so, I want to be certain that the financial statements are published and that the accounting information used by investors to make their decision are available to them. The variables are not deflated by the number of shares outstanding to control for size differences because the samples consist of companies wh big European companies wh almost the same size, so the size problem can be neglected. Otherwise, the size problem could be a limation of this thesis. 13

14 The explanatory power of equation (1) should be higher if the variable goodwill amortization is value relevant to the investors for their decision making process. But the results of the research done by Moehrle et al. (2001) show that there is no significant difference between explanatory power of goodwill amortization and that of goodwill impairment, while the results of the research done by Jennings et al. (2001) show that there is a significant difference because the explanatory power of goodwill amortization decreases, meaning that goodwill amortization is a noise to the market value of equy. However, this comparison is outside the scope of this research. So, will not be developed in this thesis. The equation (1) of the model containing the variable goodwill amortization is compared to the equation of the model containing the variable goodwill impairment: MVE = 0 + β1bve + β2ni + β3 β IMP + ε In this equation (2), I defined IMP as the amount of goodwill impairment time t. MVE, BVE, NI and AMORT have been defined earlier in the original Ohlson model (Ohlson, 1995) and equation (1). Again, the number of outstanding shares will deflate all variables. The market value of the firms equy is calculated as the multiplication of the number of shares outstanding by the share price of three months after the fiscal year-end. The explanatory power of equation (2) should be higher if the variable goodwill impairment is value relevant to the investors for their decision making process. This is confirmed by Lapointe et al. (2009) but the results of the research done by Barskjö and Paananen (2006) show no evidence that the explanatory power of goodwill impairment is higher. The hypothesis states that goodwill impairment is more value relevant than goodwill amortization. To test the hypothesis, the explanatory powers of equations (1) and (2) are compared wh each other. If the hypothesis is not rejected, goodwill impairment is more value relevant than goodwill amortization but the rejection of the hypothesis means that goodwill impairment is not more value relevant than goodwill amortization. The significance of the variables AMORT and IMP is very important in the process of providing evidence for the hypothesis testing. The following sub-section describes the periods observed for the testing of the hypothesis. (2) 14

15 3.3 Periods observed For the empirical research, I choose to observe the period from 2002 to I divided this period in four sub-periods: The first sub-period is set up to observe European companies which only applied goodwill amortization. This sub-period is chosen because 2002 is the year in which all European companies use Euro, as reporting currency and 2004 is the last fiscal year before the implementation of IFRS 3. The second sub- period , the third period and the last sub- period are set up to observe European companies which only applied goodwill impairment. These sub- periods are chosen to evaluate goodwill impairment on the longest that exist. I segregate the sample period from 2005 to 2013 into 3 sub-periods because by doing so all the sub- periods will have the same length, this makes possible to analyze the period deeply by comparing the sub- periods whin this period wh each other but also to compare easily the sample for goodwill amortization wh the samples for goodwill impairment because the period length of the samples are the same. 3.4 Data Sources To conduct the empirical analysis of this thesis, I require data on book value of shareholders equy, goodwill, and net income from Compustat Global -Fundamentals Annual and data on share price, number of shares outstanding from Compustat Global - Secury Daily. The main advantage of these databases is that there is a lot of data available on companies worldwide. To calculate the market value of the equy, I multiply the share price by the number of shares outstanding. The database Compustat does not include the line em goodwill amortization/impairment but records the goodwill figures. I measure goodwill amortization/impairment figures as the difference between goodwill reported in successive years. 15

16 3.5 Sample selection Based on the period observed, I select the samples according to the following requirements: Companies from all industries are represented. Companies have to use euro as reported currency. Companies have to report a non-zero amount of Market value of shareholders equy, Book value of shareholders equy, Net Income and goodwill amortization/impairment. Companies selected for sub-period have to use amortization for the accounting of goodwill. Few firms report goodwill amortization; the sample for subperiod is limed to firms wh continuously declining goodwill. By doing so, is possible to measure goodwill amortization directly as the difference between goodwill reported in successive years. Companies selected for sub-period , and have to apply impairment for the accounting of goodwill. The fiscal year of companies selected consisted of 12 months. Firms wh fiscal year consisting of other months are excluded from all samples Firms wh missing amount of book value of equy, net income and goodwill amortization and impairment are excluded from all samples Firms wh negative book value of equy are excluded from all samples Firms applying goodwill impairment are excluded for the sample of the sub-period The samples comprise companies wh financial year-ends at December 31 in order to produce a near common valuation date across firms. Firms wh other than December 31 year ends are excluded from all samples. Firms not applying goodwill impairment are excluded for the sample of the subperiods , and

17 The above sample selection method resulted in a total sample of N = 332 for the sub-period , N = 698 for the sub-period , N = 1081 for the sub-period and N = 1010 for the sub-period I review the observations of the samples for extreme values but there is no extreme value, so I do not exclude observations from the samples. Distribution Sample per country per year Country N Austria Belgium Germany Spain Finland France Great Brain Greece Ireland Italy Luxemburg Netherlands Portugal Totaal

18 Distribution Sample per country per year Country N Austria Belgium Germany Spain Finland France Greece Ireland Italy Monaco Netherlands Portugal Sweden Totaal

19 Distribution Sample per country per year Country N Austria Belgium Swserland Cyprus Germany Spain Finland France Great Brain Greece Ireland Italy Luxemburg Monaco Netherlands Portugal Slovenia Sweden Totaal

20 Distribution Sample per country per year Country N Austria Belgium Swserland Cyprus Germany Spain Finland France Great Brain Greece Ireland Italy Luxemburg Monaco Netherlands Portugal Sweden Totaal Please refer to the attachment Data Master Thesis Laye Mory Kourouma of this thesis for more details on the samples used for the empirical research. These details cannot be include in this thesis because of the size of. 20

21 4. Results In this section, the empirical results (Descriptive Statistics, Correlations and Regression Analyses) are presented and discussed. The empirical analysis is based on 4 samples ( , , , ). 4.1 Descriptive Statistics Statistics Sample (A) Market Value Equy Book Value Equy Mean 3485 Mean 1644 Standard Error 490 Standard Error 215 Median 208 Median 115 Mode 8373 Mode Standard Deviation 8923 Standard Deviation 3912 Sample Variance Sample Variance Kurtosis 16 Kurtosis 15 Skewness 4 Skewness 4 Range Range Minimum 1 Minimum 542 Maximum Maximum Sum Sum Count 332 Count 332 Largest(1) Largest(1) Smallest(1) 1 Smallest(1) 542 Confidence Level (95,0%) 963 Confidence Level(95,0%)

22 Statistics Sample (B) Goodwill Amortization Net Income Mean 211 Mean 23 Standard Error 61 Standard Error 85 Median 3 Median 6 Mode 1 Mode 773 Standard Deviation 1109 Standard Deviation 1550 Sample Variance Sample Variance Kurtosis 183 Kurtosis 160 Skewness 12 Skewness 11 Range Range Minimum 0 Minimum Maximum Maximum 3626 Sum Sum 7619 Count 332 Count 332 Largest(1) Largest(1) 3626 Smallest(1) 0 Smallest(1) Confidence Level (95,0%) 120 Confidence Level(95,0%) 167 Statistics Sample (A) Market Value Equy Book Value Equy Mean 6639 Mean 3118 Standard Error 538 Standard Error 252 Median 1909 Median 855 Mode #N/B Mode 240 Standard Deviation Standard Deviation 6645 Sample Variance Sample Variance Kurtosis 33 Kurtosis 21 Skewness 5 Skewness 4 Range Range Minimum 8 Minimum 203 Maximum Maximum Sum Sum Count 698 Count 698 Largest(1) Largest(1) Smallest(1) 8 Smallest(1) 203 Confidence Level (95,0%) Confidence 1056 Level(95,0%)

23 Statistics Sample (B) Goodwill Impairment Net Income Mean 200 Mean 489 Standard Error 48 Standard Error 48 Median 6 Median 110 Mode 0 Mode 501 Standard Deviation 1278 Standard Deviation 1264 Sample Variance Sample Variance Kurtosis 181 Kurtosis 45 Skewness 11 Skewness 6 Range Range Minimum 6127 Minimum 3518 Maximum Maximum Sum Sum Count 698 Count 698 Largest(1) Largest(1) Smallest(1) 6127 Smallest(1) 3518 Confidence Level (95,0%) 95 Confidence Level (95,0%) 94 Statistics Sample (A) Market Value Equy Book Value Equy Mean 6246 Mean 5155 Standard Error 382 Standard Error 322 Median 1367 Median 1154 Mode #N/B Mode 9948 Standard Deviation Standard Deviation Sample Variance Sample Variance Kurtosis 16 Kurtosis 13 Skewness 4 Skewness 3 Range Range Minimum 2 Minimum 455 Maximum Maximum Sum Sum Count 1081 Count 1081 Largest(1) Largest(1) Smallest(1) 2 Smallest(1) 455 Confidence Level (95,0%) 749 Confidence Level (95,0%)

24 Statistics Sample (B) Goodwill Impairment Net Income Mean 107 Mean 436 Standard Error 24 Standard Error 51 Median 2 Median 75 Mode 1 Mode 5027 Standard Deviation 792 Standard Deviation 1674 Sample Variance Sample Variance Kurtosis 66 Kurtosis 86 Skewness 2 Skewness 3 Range Range Minimum Minimum Maximum Maximum Sum Sum Count 1081 Count 1081 Largest(1) Largest(1) Smallest(1) Smallest(1) Confidence Level (95,0%) 47 Confidence Level (95,0%) 100 Statistics Sample (A) Market Value Equy Book Value Equy Mean 7865 Mean 6282 Standard Error 468 Standard Error 408 Median 1889 Median 1421 Mode #N/B Mode Standard Deviation Standard Deviation Sample Variance Sample Variance Kurtosis 13 Kurtosis 13 Skewness 3 Skewness 3 Range Range Minimum 0 Minimum 2129 Maximum Maximum Sum Sum Count 1010 Count 1010 Largest(1) Largest(1) Smallest(1) 0 Smallest(1) 2129 Confidence Level (95,0%) 919 Confidence Level (95,0%)

25 Statistics Sample (B) Goodwill Impairment Net Income Mean 36 Mean 417 Standard Error 42 Standard Error 56 Median 0 Median 87 Mode 1 Mode 4842 Standard Deviation 1337 Standard Deviation 1796 Sample Variance Sample Variance Kurtosis 110 Kurtosis 38 Skewness 0 Skewness 2 Range Range Minimum Minimum Maximum Maximum Sum Sum Count 1010 Count 1010 Largest(1) Largest(1) Smallest(1) Smallest(1) Confidence Level (95,0%) 83 Confidence Level (95,0%) 111 Statistics Sample (A) Market Value Equy Book Value Equy Mean 6931 Mean 5054 Standard Error 263 Standard Error 205 Median 1700 Median 1131 Mode 833 Mode 4301 Standard Deviation Standard Deviation Sample Variance Sample Variance Kurtosis 19 Kurtosis 16 Skewness 4 Skewness 4 Range Range Minimum 0 Minimum 2129 Maximum Maximum Sum Sum Count 2789 Count 2789 Largest(1) Largest(1) Smallest(1) 0 Smallest(1) 2129 Confidence Level (95,0%) 515 Confidence Level (95,0%)

26 Statistics Sample (B) Goodwill Impairment Net Income Mean 78 Mean 443 Standard Error 22 Standard Error 31 Median 1 Median 89 Mode 1 Mode 501 Standard Deviation 1143 Standard Deviation 1629 Sample Variance Sample Variance Kurtosis 152 Kurtosis 62 Skewness 4 Skewness 0 Range Range Minimum Minimum Maximum Maximum Sum Sum Count 2789 Count 2789 Largest(1) Largest(1) Smallest(1) Smallest(1) Confidence Level (95,0%) 42 Confidence Level (95,0%) Correlations Correlation Sample Market Value Equy Book Value Equy Goodwill Amortization Net Income Market Value Equy 1,00 Book Value Equy 0,83 1,00 Goodwill Amortization 0,30 0,41 1,00 Net Income 0,15 0,02 0,73 1,00 Correlation Sample Market Value Equy Book Value Equy Goodwill Impairment Net Income Market Value Equy 1,00 Book Value Equy 0,91 1,00 Goodwill Impairment 0,15 0,15 1,00 Net Income 0,91 0,85 0,11 1,00 26

27 Correlation Sample Market Value Equy Book Value Equy Goodwill Impairment Net Income Market Value Equy 1,00 Book Value Equy 0,81 1,00 Goodwill Impairment 0,27 0,21 1,00 Net Income 0,68 0,58 0,24 1,00 Correlation Sample Market Value Equy Book Value Equy Goodwill Impairment Net Income Market Value Equy 1,00 Book Value Equy 0,75 1,00 Goodwill Impairment 0,01 0,02 1,00 Net Income 0,60 0,55 0,40 1,00 Correlation Sample Market Value Equy Book Value Equy Goodwill Impairment Net Income Market Value Equy 1,00 Book Value Equy 0,78 1,00 Goodwill Impairment 0,10 0,06 1,00 Net Income 0,68 0,59 0,27 1, Regression Analyses The results of the regression analyses on the samples are reported in Table A through E. The results of the regression on the sample are reported below: 27

28 Table A: Results Regression Analysis Sample Table A MVE = 0 + β1bve + β2ni + β3 β AMORT + ε (1) N Intercept BVE NI AMORT Coefficients 505,82 1,42 0,07-0,08 p-value 0,40 0,00 0,86 0,89 t statistic 0,85 9,28 0,18-0, Coefficients 192,32 1,39 5,74 4,26 p-value 0,68 0,00 0,00 0,01 t statistic 0,41 7,43 5,86 2, Coefficients 253,43 0,55 8,58 6,38 p-value 0,18 0,00 0,00 0,00 t statistic 1,34 5,87 14,99 12,75 Pooled 332 Coefficients 335,43 1,67 1,68 1,74 p-value 0,24 0,00 0,00 0,00 t statistic 1,17 19,48 5,85 3,95 0,71 Adj. R² Overall, the results on the variables BVE, NI and AMORT are consistent. Except for 2002 (where the p-values of net income (0.86) and amortization (0.89) are greather than 0.01), the results on the variables BVE, NI and AMORT are significant at the 1% level even in the pooled sample. This indicates that investors consider these accounting information numbers in their valuation of the firm. The results on the variables BVE, and NI are an expectation because the book value of equy and the net income are elements of the company s value that are the property of the shareholders. However the results on the variable AMORT is a surprise especially for the pooled sample. The coefficient on goodwill amortization for the pooled sample (1.74) is the highest compared to those on the book value of equy (1.67) and the net income (1.68), This would mean that a change in goodwill amortization expense has more effect on the share price and the value of the company than a change in book value of equy or in net income has. This is not in line wh the theory that higher expenses of goodwill amortization decrease the market value of a firm s equy. The reported adjusted R² of 71% is the explanatory power of the model. This means that the combination of the three variables (BVE, NI and AMORT ) in the model explains seventy-one percent of the share price or the market value of the company. 28

29 Table B presents the results of the regression on the sample Table B: Results Regression Analysis Sample Table B MVE = 0 + β1bve + β2ni + β3 β IMP + ε (2) N Intercept BVE NI IMP Coefficients 173,49 1,26 6,21 1,39 p-value 0,54 0,00 0,00 0,00 t statistic 0,61 15,57 14,78 4, Coefficients 711,39 1,03 6,42 0,50 p-value 0,03 0,00 0,00 0,17 t statistic 2,17 12,96 14,55 1, Coefficients 660,50 0,76 5,24 0,00 p-value 0,02 0,00 0,00 0,97 t statistic 2,35 10,12 13,85 0,03 Pooled 698 Coefficients 608,45 1,08 5,35 0,26 p-value 0,00 0,00 0,00 0,06 Adj. R² t statistic 3,13 21,49 20,43 1,87 0,89 In above table, the results on the variables BVE, and NI are all significant at the 1% level. However the results on the variable IMP is not consistent in the different years. The coefficient and the t-values on the variable IMP decrease from 2005 to 2007 while s p- values increase. The coefficient on the variable IMP (0.26) is not significant for the pooled sample indicating that a change in impairment expense results in a lower or no effect on the share price and the value of the company. This is not in line wh the theory that higher expenses of goodwill impairment result in lower market value of a firm s equy and share prices. The explanatory power of the model is the reported adjusted R² of 89%, which means that the combination of the three variables (BVE, NI and IMP ) in the model explains eighty-nine percent of the share price or the market value of the company. But is interesting to note that Net income has the highest coefficient for all the years in this sample inclusive the pooled sample. So the contribution of the variable net income is the biggest in the explanation. 29

30 Comparison between Table A and Table B shows the following findings: First of all, there are more than two times observations in the sample (N = 698) containing firms applying goodwill impairment than in the sample (N = 332) containing firms applying goodwill amortization. This difference can be explained by the fact that most of the European companies already apply goodwill impairment before 2005 the year in which goodwill impairment became effective by law. Secondly, in Table A, the coefficient of the variable AMORT increase through the examined year while in Table B the coefficient of the variable IMP decrease through the examined year meaning that goodwill amortization becomes more significant while goodwill impairment becomes less significant in the explanation of the share price and the value of the company. Finally, the adjusted R² of 89% reported in Table B is higher than the adjusted R² of 71% reported in Table A indicating that equation (2) explains better the share price and the market value of the company than equation (1) but this is due more to the variable net income than to the variables book value of equy and impairment. The results of the regression on the sample are presented below: 30

31 Table C: Results Regression Analysis Sample Table C MVE = 0 + β1bve + β2ni + β3 β IMP + ε (2) N Intercept BVE NI IMP Coefficients 788,64 0,76 1,30-0,16 p-value 0,02 0,00 0,00 0,59 t statistic 2,35 20,65 7,92-0, Coefficients 1920,87 0,65 4,18 1,65 p-value 0,00 0,00 0,00 0,05 t statistic 4,68 14,39 10,89 1, Coefficients 1353,67 0,59 3,46 3,37 p-value 0,00 0,00 0,00 0,00 t statistic 3,96 14,45 10,84 7,13 Pooled 1081 Coefficients 1356,05 0,73 2,27 1,07 p-value 0,00 0,00 0,00 0,00 Adj. R² t statistic 6,06 31,49 15,26 4,08 0,72 In 2008, Table C shows that the results on the variable IMP are insignificant (coefficient = , p-value = 0.59 > 0.01, t-value = -0.54). Further, the other results on the variables BVE, NI and IMP are significant at the 1% level. The coefficient of the variable BVE decrease through the examined year while the coefficients of the variable NI and IMP increase through the same period meaning that net income and goodwill impairment become more significant while good impairment become less significant in the explanation of the share price and the value of the company. It is also remarkable that compared to the sample (N = 698) the number of observations in the sample almost has increased twice to However the adjusted R² of 72% reported in Table C is lower than the adjusted R² of 89% reported in Table B indicating that the variables BVE, NI and IMP in the sample explain better the share price and the market value of the company than those in the sample Again, the variable net income has the highest coefficient. 31

32 Table D contains the regression statistics of the sample Table D: Results Regression Analysis Sample Table D MVE = 0 + β1bve + β2ni + β3 β IMP + ε (2) N Intercept BVE NI IMP Coefficients 2180,52 0,63 2,04-0,76 p-value 0,00 0,00 0,00 0,01 t statistic 4,66 15,32 7,59-2, Coefficients 2811,98 0,48 4,26-2,72 p-value 0,00 0,00 0,00 0,00 t statistic 5,06 9,04 8,47-4, Coefficients 2828,12 0,68 3,79-2,81 p-value 0,00 0,00 0,00 0,00 t statistic 4,60 11,84 7,82-3,43 Pooled 1010 Coefficients 2582,53 0,64 2,83-1,46 p-value 0,00 0,00 0,00 0,00 t statistic 8,12 23,22 12,94-5,95 0,63 Adj. R² The results on the variables BVE, NI and IMP are all significant at the 1% level. But the coefficients on the variable IMP are all negative, indicating that an increase in impairment expenses result in a lower market valuation. This is consistent wh the concept that higher impairment expenses lead to lower share price and market value of the company. Further, the adjusted R² of 63% reported in Table D is much more lower than the adjusted R² of 71% reported in Table A, the adjusted R² of 89% reported in Table B and the adjusted R² of 72% reported in Table C indicating that the variables BVE, NI and IMP in the sample do not explain better the share price and the market value of the company than those in the samples , and The results of the regression on the sample are presented below: 32

33 Table E MVE = 0 + β1bve + β2ni + β3 β IMP + ε (2) N Intercept BVE NI IMP Coefficients 173,49 1,26 6,21 1,39 (p-value) 0,54 0,00 0,00 0,00 (t statistic) 0,61 15,57 14,78 4, Coefficients 711,39 1,03 6,42 0,50 (p-value) 0,03 0,00 0,00 0,17 (t statistic) 2,17 12,96 14,55 1, Coefficients 660,50 0,76 5,24 0,00 (p-value) 0,02 0,00 0,00 0,97 (t statistic) 2,35 10,12 13,85 0, Coefficients 788,64 0,76 1,30-0,16 (p-value) 0,02 0,00 0,00 0,59 (t statistic) 2,35 20,65 7,92-0, Coefficients 1920,87 0,65 4,18 1,65 (p-value) 0,00 0,00 0,00 0,05 (t statistic) 4,68 14,39 10,89 1, Coefficients 1353,67 0,59 3,46 3,37 (p-value) 0,00 0,00 0,00 0,00 (t statistic) 3,96 14,45 10,84 7, Coefficients 2180,52 0,63 2,04-0,76 (p-value) 0,00 0,00 0,00 0,01 (t statistic) 4,66 15,32 7,59-2, Coefficients 2811,98 0,48 4,26-2,72 (p-value) 0,00 0,00 0,00 0,00 (t statistic) 5,06 9,04 8,47-4, Coefficients 2828,12 0,68 3,79-2,81 (p-value) 0,00 0,00 0,00 0,00 (t statistic) 4,60 11,84 7,82-3,43 Pooled 2789 Coefficients 1937,35 0,73 3,01-0,38 (p-value) 0,00 0,00 0,00 0,01 Adj. R² (t statistic) 11,78 42,60 25,47-2,75 0,68 33

34 Interesting at above table is the pooled sample because I have discussed the other parts of the table earlier. The results on the variables BVE, NI and IMP are all significant at the 1% level. But the coefficient on the variable IMP is negative, indicating that an increase in impairment expenses result in a lower market valuation. The adjusted R² of 68% reported in Table E is much more lower than the adjusted R² of 71% reported in Table A, meaning that the variables BVE, NI and IMP in the sample do not explain better the share price and the market value of the company than that in the samples But this conclusion has to be taken wh care because comparing two samples wh different period ( and ) and numbers of observations (N= 332 for and N= 2789 for ) is not fair. The hypothesis states that goodwill impairment is more value relevant than goodwill amortization. Based on the higher explanatory power of equation (2) in Table B over equation (1) in Table A, this hypothesis cannot be rejected. However, based on the increasing significance of the amortization variable and the decreasing significance of the impairment variable during the three years, is difficult to conclude that goodwill impairment is more value relevant than goodwill amortization. Based on the higher explanatory power of equation (2) in Table C over equation (1) in Table A, and the less significance of the impairment variable compared wh the more significance of the amortization variable in two of the three years of eher sample, I have to adm that very difficult to not reject the hypothesis. Because of the higher explanatory power of equation (2) in Table C over equation (1) in Table A, I have to conclude that goodwill impairment is more value relevant than goodwill amortization. The comparison of Table A wh Table D shows that the explanatory power of equation (1) R² of 71% in Table A is higher over the explanatory power of equation (2) R² of 63% in Table D, and the insignificance of the impairment variable compared wh the significance of the amortization variable in all the three years of eher sample, the hypothesis can be rejected. I conclude that goodwill impairment is not more value relevant than goodwill amortization. The comparison of Table A wh Table E shows that the explanatory power of equation (1) R² of 71% in Table A is higher over the explanatory power of equation (2) R² of 68% in Table E, and the insignificance of the impairment variable compared wh the significance of the amortization variable in the pooled sample of eher sample, the hypothesis can be rejected. I conclude that goodwill impairment is not more value relevant than goodwill amortization. 34

35 5. Conclusion The principal aim of this thesis is to investigate whether goodwill impairment is more value relevant than goodwill amortization. The results of this thesis show mixed evidence. Comparing the sample wh goodwill amortization to samples wh goodwill impairment, the results shows consistent wh my prediction that goodwill impairment is more value relevant than goodwill amortization in the first six years after adoption of IFRS 3 and that this value relevance is decreasing as time past to the level that is not more relevant than goodwill amortization after nine years. Further, is remarkable that the contribution of goodwill impairment in the explanatory power decreases over the time while that of goodwill amortization increases. This conclusion has to be taken wh care because the sample wh goodwill amortization has been analyzed for a shorter time (3 years) while the goodwill impairment is subject to a research period of 9 years. An unexpected finding is the higher significance and contribution of the variable net income in the explanatory power of both equations (1) and (2). A main limation of this thesis is the shorter time on which goodwill amortization is analyzed. Therefore, Future research can extend these tests of goodwill amortization and impairment wh amortization data before The size problem of the companies used in the samples could be another limation of this thesis. 35

36 6. References American Instute of Certified Public Accountants (AICPA), (1970). Accounting Principles Board Opinion No. 16 Business Combinations. AICPA, New York. Barksjö, J. and Paananen, M. (2006). Preliminary Evidence of the Effects of the Adoption of the Impairment-Only Approach to Goodwill Accounting in Sweden. Working Paper, Universy of Göteborg. Barth, M.E., Beaver, W.H., Landsman, W.R., (2001). The relevance of the value-relevance lerature for financial accounting standard setting. Journal ofaccounting and Economics, 31, (1 3), pp Bryer, R. A. (1995) A polical Economy of SSAP22: Accounting for goodwill. Brish Accounting Review 27, 4, pp Bugeja, M. and Gallery, N. (2006). Is Older Goodwill Value Relevant? Accounting & Finance, 46, 4, pp Chalmers et al. (2011). Does a goodwill impairment regime better reflect the underlying economic attributes of goodwill? Accounting and Finance, 51, 3, pp Chambers, D.J. (2006). Is Goodwill Accounting under SFAS 142 an Improvement over Systematic Amortization of Goodwill? Working Paper. Universy of Kentucky. Churyk, N.T. and Chewning Jr., E.G. (2003). Goodwill and Amortization: Are they Value Relevant? Academy of Accounting and Financial Studies Journal, 7, 2, pp Churyk, N.T. (2005). Reporting Goodwill: Are the New Accounting Standards Consistent wh Market Valuations? Journal of Business Research, 58, 11, pp