Amsterdam Business School
The value relevance and reliability of goodwill in relation with
CFO equity incentives: empirical evidence from the US
Name: Mohamed Monji Student number: 10433589
Thesis supervisor: dr. G. Georgakopoulos Date: 20 June 2016
Word count: 14400
MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by student Mohamed Monji who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
The aim of this thesis is to explore the effect of SFAS 141 and 142 on the reporting of goodwill. To be more specific, I examine the impact of these accounting standards on both the value relevance and the reliability of reported goodwill amounts for US firms during the period 2002-2007. On the basis of a modified Ohlson (1995) model I find that goodwill is significantly associated with market values during the period 2002–2007. This result indicates that goodwill under SFAS 141 and 142 is value relevant. Furthermore, I explore whether firms with CFOs who have high equity incentives are more likely to overstate goodwill compared to those who have low equity incentives. However, I do not find enough evidence to conclude that firms with CFOs who have high equity incentives overstate goodwill. This result suggests that CFOs’ equity incentives do not diminish the reliability of reported goodwill.
Table of Contents
1 Introduction ... 5
2 Background ... 7
2.1 Intangible Assets: Goodwill ... 7
2.2 Accounting for Goodwill – SFAS 141 and 142 ... 9
2.3 The Relevance and Reliability of Financial Information ... 11
3 Literature Review and Hypothesis Development ... 12
3.1 The Value Relevance of Financial Information ... 12
3.1.1 Previous Literature on the Value Relevance of Goodwill……...………….12
3.2 The Reliability of Financial Information ... 14
3.2.1 The Agency Theory and Positive Accounting Theory ....………...………….14
3.2.2 Previous Literature on Equity Incentives to Manipulate Financial Information……….. 16
3.2.3 Discretion in the Accounting for Goodwill under SFAS 141 and 142..…...18
3.2.4 Previous Literature on the Reliability of Goodwill... ....………...…………...20
3.3 Hypothesis Development ... 22
4 Research Methodology ... 23
4.1 Sample and Data Selection ... 23
4.2 Research Design ... 25
5 Results ... 29
5.1 Descriptive Statistics and Correlation Analysis ... 29
5.2 Regresision Results and Analysis ... 32
6 Conclusion ... 35
1. Introduction
According to Cañibano, Garcia-Ayuso and Sánchez (2000) we are moving to a more knowledge- and technology-based economy in which intangible assets play a crucial role. The importance of intangible assets is also noticed by Corrado and Hulten (2010). In their research they show that the rate of investment in intangible assets more than doubled in the US during the period 1948-2007, while the rate of investment in tangible assets declined below the rate of intangible assets during this period. Fig. 1 depicts these rates of investment in both types of assets. Based on their results Corrado and Hulten conclude that there is a major change in the composition of investments (tangible vs. intangible) by firms and they emphasize the increasing importance of intangible assets.
Fig. 1. The investment rates in tangible assets and intangible assets in the US during the period 1948-2007. Retrieved from “How Do You Measure a “Technological Revolution”?,” by C. A. Corrado and C. R. Hulten, 2010, American Economic Review, 100, p. 101.
Moreover, the fact that Forbes' list of the top ten world’s most valuable brands is dominated by companies which operate in the technology industry confirms the claims of Cañibano et al. (2000) and that of Coraddo and Hulten (2010). Currently, there are five technology companies in this top ten: Apple, Google, Microsoft, IBM and Facebook.
So it is clear that intangible assets are very important and have a crucial role in today’s knowledge- and technology-based economy. One type of intangible assets that increased in importance and proportion is goodwill. For instance, the study of Hayn & and
Hughes (2006, p. 224) shows that goodwill as percentage of total assets increased in the US, on average, from 10.7% in 1998 to 16.8% in 2001. The increased significance of goodwill makes it even more important that this type of intangible asset should be reported in an appropriate way in the financial statements. This was also noticed by the Financial
Accounting Standards Board (2001a, 2001b) and as a response they introduced SFAS 141 –
Business Combinations and SFAS 142 - Goodwill and Other Intangible Assets in 2001 to
improve the financial reporting of business combinations and intangible assets, respectively, and therefore the reporting of goodwill.
Motivated by the increased importance and significance of goodwill and the issuance of SFAS 141 and 142 by the FASB, I want to investigate the effect of the adoption of SFAS 141 and 142 on the reporting of goodwill. To be more specific, I investigate in the first place whether the reported goodwill is associated with market values, i.e. whether goodwill is value relevant. Therefore, my first research question is specified as follows: Is reported goodwill
value relevant under SFAS 141 and 142? If goodwill is value relevant it would mean that
goodwill is perceived by investors as relevant, but only reliable to some extent (Barth et al., 2001). Therefore, in the extension of the value relevance of goodwill, I want to investigate the reliability of reported goodwill in further detail. Specifically, I examine the reliability of goodwill by looking whether the amount of goodwill is likely to be overstated by firms with CFOs who have high equity incentives compared to those who have low equity incentives. So the second research question is specified as follows: Is reported goodwill overstated by firms
with CFOs who have high equity incentives?
The research is based on a sample of US firm that reported positive amounts of
goodwill during the period 2002-2007. On the basis of a modified Ohlson (1995) model I find that goodwill is significantly associated with market values during the period 2002–2007. So I find evidence that goodwill reported under SFAS 141 and 142 is value relevant. However, I do not find enough evidence to infer that firms with a CFO who has high equity incentives are more likely to overstate goodwill compared to firms with a CFO who has low equity incentives. So this result suggests that the reliability of goodwill is not impaired by equity incentives and therefore goodwill is reliable in relation to equity incentives.
The objective of this thesis is to contribute to the existing knowledge regarding the reporting of goodwill. The focus of this research is to deepen and extend the knowledge on the value relevance and reliability of goodwill. There are previous studies that examined the value relevance of goodwill, but most of these studies explored the value relevance of goodwill before the adoption of SFAS 141 and 142 (e.g. Chauvin & Hirschey, 1994;
Jennings, Robinson, Thompson & Duvall, 1996; McCarthy & Schneider, 1995). A research that explores the value relevance of goodwill under SFAS 141 and 142 is the study of Chen, Kohlbeck and Warfield (2004). Their study provides evidence that goodwill under SFAS 141 and 142 is value relevant, however, the evidence is preliminary and limited, because they only focus on reported goodwill for the year 2002 (i.e. in an early stage of SFAS 141 and 142).
Furthermore, most of the existing literature examined the reliability of goodwill indirectly by exploring the reliability of goodwill impairments (e.g. Beatty & Weber, 2006; Jarva, 2009; Ramanna & Watts, 2012). The aim of this thesis is, however, to explore the reliability of goodwill by focusing on the recorded value of goodwill. More interesting, I explore the effect of CFO equity incentives on the reliability of goodwill. Kallapur and Kwan (2004) suggest in their paper that compensation based on equity could give managers
incentives to overstate intangible assets; however, they only suggest this and do not further investigate it. Last but not least, my research focuses on CFO equity incentives (instead of CEO equity incentives) and according to Indjejikian and Matějka (2009) the research of CFO incentives and compensation is limited. So my research will also contribute to the existing knowledge in this regard. To my best knowledge, this combination of studying the reliability of goodwill in relation with CFO equity incentives has not been done before.
The rest of this thesis is structured as follows. In the next section, I present the
background to the accounting for goodwill before and after SFAS 141 and 142. In section 3 I review existing literature in order to develop the hypotheses for this research. The research methodology (including the sample selection procedures and research design) is described in section 4. In section 5 the results are presented and analyzed. Finally, in section 6 I provide a conclusion.
2. Background
2.1 Intangible Assets: Goodwill
Cañibano, Garcia-Ayuso and Sánchez (2000) argue that we are moving to a more knowledge- and technology-based economy in which intangible assets are becoming important. The increasing importance of intangible assets is also noticed by Corrado and Hulten (2010). They show in their research that the rate of investment in intangible assets more than doubled
in the US during the period 1948-2007 and conclude that there is a major change in the composition of investments with the emphasis on intangible investments rather than tangible investments. Furthermore, intangible assets are becoming the main value creators and sources to gain a competitive advantage for businesses (Cañibano, et al., 2000; Zéghal & Maaloul, 2011). So it is clear that intangible assets are very important and have a crucial role in today’s knowledge- and technology-based economy.
The FASB (2001b) also noticed the increased importance of intangible assets by mentioning that “analysts and other users of financial statements, as well as company managements, noted that intangible assets are an increasingly important economic resource for many entities and are an increasing proportion of the assets acquired in many
transactions” (p. 5). One type of the intangible assets which is increasing in importance and proportion is goodwill. Goodwill represents the ability to gain an above “average” return on investment as a result of, for example, synergies, a unique location, an excellent reputation or customer’s loyalty and other intangible assets that cannot be recognized as separate assets from goodwill (Cañibano et al., 2000; FASB, 2001a; Jennings, LeClere & Thompson, 2001).
As stated above, goodwill is increasing in importance and proportion. For instance, McCarthy and Schneider (1995, p. 6) show that during the period 1988-1992 the total amount of recognized goodwill increased from $89 billion to $158 billion in the US (an increase of approximately 78%). Furthermore, Hayn and Hughes (2006, p. 224) show the increased weight of goodwill on the balance sheet in the US. They show that in 1988 goodwill was 10.7% of total assets; while this percentage increased to 16.8% in 2001.
In light of the findings above it is important to note that until 2001, the accounting for goodwill was addressed by APB Opinion No. 16, Business Combinations and APB Opinion No. 17, Intangible Assets. Under Opinion 16, there were two methods to account for business combinations, namely the pooling method (only when some criteria are met) and the purchase method. The main difference between both methods is that under the pooling method no goodwill is recognized, while under the purchase method goodwill (the difference between the purchase price and the fair value of the net assets) is capitalized and amortized over a maximum of 40 years according to Opinion No. 17 (Ayers, Lefanowicz & Robinson, 2000; Ramanna, 2008). It is thus likely that companies prefer the pooling method, because the purchase method leads to amortization expenses that negatively impact net income. Lys and Vincent (1995, p. 353) analyzed the acquisition of NCR by AT&T in 1991 and conclude that AT&T paid an amount between $50 and $500 million to qualify for the pooling method. They paid this huge amount in order to avoid a reduction in EPS due to amortization
expenses if they accounted for the acquisition under the purchase method. The research of Lys & Vincent confirms that firms prefer the pooling method above the purchase method. In light of this conclusion Ayers et al. (2000) analyze the impact of the pooling method on the financial statements. They show that due the pooling method an amount of $267 billion (approximately two-thirds of the total acquisition price) of assets is not recognized on the financial statements during the period 1992-1997.
So it makes sense that due the existence of the pooling method a lot of goodwill is not recognized on the balance sheets, despite its importance in today’s new knowledge-based and intangible-intensive economy. The FASB acknowledged the problems with the pooling method and responded to the increased importance of intangible assets with the issuance of two new accounting standards in 2001, namely SFAS 141 – Business Combinations and SFAS 142 - Goodwill and Other Intangible Assets, to improve the reporting of intangible assets and thus goodwill.
2.2 Accounting for Goodwill – SFAS 141 and 142
As stated earlier, the FASB issued in 2001 SFAS 141 and SFAS 142 to improve the reporting of goodwill. The main reasons of the FASB (2001a, 2001b) to issue these new standards are: (1) the existence of two accounting methods (pooling and purchase) which lead to different accounting treatment of similar business combinations and (2) the fact that goodwill
amortization is not regarded as useful by financial statement users. But what are the changes in accounting for intangible assets, especially for goodwill, in comparison with the old standards Opinion No. 16 and 17?
Under the standard SFAS 141 (FASB, 2001a) all business combinations are
accounted for by using the purchase method, so the doubtful pooling method of Opinion No. 16 is eliminated. This purchase method under SFAS 141 means that the acquiring entity should allocate the purchase price to the acquired assets (including intangible assets, whether or not recognized in the balance sheet of the acquired entity) and liabilities at their fair value. The excess of the purchase price over the fair value of the net assets acquired is recognized as goodwill. The consequence of the elimination of the pooling method under SFAS 141 is the increase of the amount of goodwill on the balance sheets. So the importance and proportion of goodwill is likely to increase even more due the adoption of this new standard SFAS 141. The independent valuation consultancy, Intangible Business, issued a report in 2007 to evaluate the effects of SFAS 141. Intangible Business (2007, p. 5) shows that in the period
2002-2006 a total of 212 acquisitions were reported by the 100 largest US companies with a total acquisition price of $1,033 billion. An amount of $490 billion of the total acquisition price (which is approximately 48%) was allocated to goodwill. This report emphasizes the increasing proportion and importance of the goodwill on the balance sheet due to the new standard SFAS 141. According to the FASB (2001a) this new standard SFAS 141 “will improve financial reporting because the financial statements of entities that engage in business combinations will better reflect the underlying economics of those transactions” (p. 3).
The FASB issued at the same time SFAS 142 to replace Opinion No.17. One of the reasons to issue SFAS 142 was because financial statement users do not regard goodwill amortization expenses, as required by Opinion No.17, as useful (FASB, 2001b). Under SFAS 142 (FASB, 2001b) goodwill is now tested for impairment and this is done at the reporting unit level; that is why the total amount of goodwill acquired must be assigned to reporting units. There is an impairment loss if the carrying amount of goodwill exceeds its implied fair value. In order to determine this possible impairment there is an impairment test of two steps: (1) step one is comparing the fair value of a reporting unit with its carrying amount (including goodwill) to identify a potential impairment. If this carrying amount exceeds the fair value the second step is necessary to measure the probable amount of impairment. (2) Step two is comparing the implied fair value1 of the reporting unit goodwill with the carrying amount of the goodwill. If the carrying amount of the goodwill exceeds the implied fair value of the goodwill, then the excess of it is recognized as impairment. Furthermore, it is important to note that SFAS 142 prohibits the reversal of the impairments. Last but not least, SFAS 142 requires that goodwill shall be tested for impairment at least annually and between annual tests only in certain circumstances. The FASB (2001b) claims that SFAS 142 “will improve financial reporting because the financial statements of entities that acquire goodwill and other intangible assets will better reflect the underlying economics of those assets” (p. 7).
So since the replacement of Opinion No. 16 and 17 in 2001, SFAS 141 and 142 are now the main accounting standards that address the recognition and subsequent measurement of goodwill. The main changes due to the introduction of SFAS 141 and 142 are, as discussed before, the elimination of the pooling method and the elimination of the amortization of
1
According to the FASB (2001b) the implied fair value of reporting unit goodwill shall be determined by allocating the fair value of a reporting unit to the reporting unit its assets and liabilities at their fair value. The excess of the fair value of a reporting unit over the amounts assigned to the assets and liabilities is considered to be the implied fair value of reporting unit goodwill. It follows in essence the same process as determining the goodwill in a business combination.
goodwill. Under SFAS 141 now only the purchase method is allowed and SFAS 142 introduced impairment testing for goodwill as a replacement for the amortization. As previously quoted the FASBbelieves that SFAS 141 and 142 will improve the financial reporting of business combinations and intangible assets and therefore the reporting of goodwill.
2.3 The Relevance and Reliability of Financial Information
As previously noted the FASBaims with the adoption of SFAS 141 and 142 to improve the financial reporting of business combinations and intangible assets and therefore the reporting of goodwill. However, in order to evaluate whether there is an improvement in the financial reporting of goodwill it is in the first place necessary to understand the objective of financial reporting. According to the Conceptual Framework of the FASB (2010) “the objective of financial reporting is to provide financial information about the reporting entity that is useful to existing and potential investors, lenders, and other creditors in making decisions about providing resources to the entity” (p.1). So, in short, the goal of financial reporting is to provide useful financial information to capital providers.
Now the objective of financial reporting is clear, it is in the second place important to know when financial information is useful. The FASB (2010) mention in their Conceptual Framework that financial information is useful when it is relevant and faithfully represented (i.e. reliable). These concepts of relevance and faithfully representation (referred to as reliability) are fundamental characteristics of useful information. According to the FASB (2010) financial information is relevant when it “is capable of making a difference in the decisions made by users” (p.17). In the extension of this they provide further information about when information is relevant by stating that “financial information is capable of making a difference in decisions if it has predictive value, confirmatory value, or both” (p.17). The other fundamental characteristic, as stated earlier, is the reliability of financial information. Financial information is reliable if it is complete, neutral and free form error (FASB, 2010).
On the basis of the above, it is clear that in order for SFAS 141 and 142 to improve the reporting of goodwill the financial information provided about goodwill must be relevant and reliable. In order to explore the relevance and reliability of goodwill (which is in essence the aim of this thesis) in the next section the concept of value relevance, the agency theory and the positive accounting theory are discussed and related literature is reviewed.
3. Literature Review and Hypothesis Development
3.1 The Value Relevance of Financial Information
In the extant literature an important concept that is related to the relevance and reliability of financial information is value relevance. This concept of value relevance is according to Barth, Beaver and Landsman (2001) an operationalization of the fundamental characteristics relevance and reliability. Financial information is considered to be value relevant when it is associated with equity market values (e.g. stock price, stock return or market capitalization) (Barth et al., 2001; Wyatt, 2008). However, Barth et al. (2001) and Wyatt (2008) note that when an accounting number is associated with market values (i.e. it is value relevant) than it is considered to be relevant, but only reliable to some extent.
Value relevance tests are done on the basis of a valuation model, e.g. the valuation model of Ohlson (1995). The aim of these value relevance tests is to determine whether there is an association between stock data and accounting amounts by focusing on the significance of the coefficients of these accounting amounts (Barth et al., 2001; Holthausen and Watts, 2001). For example, Collins, Maydew and Weiss (1997) focused in their research on the value relevance of earnings and the book value of equity, while Kallapur and Kwan (2004) investigated the value relevance of brand assets.
So it is clear that value relevance test are an interesting way to explore whether accounting amounts in the financial statements are relevant and to some degree reliable. Therefore these value relevance tests are also interesting for analyzing whether the
implementation of SFAS 141 and 142 leads to relevant and reliable goodwill. But, first it is important to explore existing literature on the value relevance of goodwill.
3.1.1 Previous Literature on the Value Relevance of Goodwill
The aim of this section is to provide a review of the existing literature on the value relevance of goodwill. One of the previous researches that explored the association between equity values and goodwill amounts is the research of McCarthy and Schneider (1995). They focus in their study on the association between the market capitalization and the book value of goodwill for firms in the US during the period 1988-1992. It is important to note here that during this period the old standards Opinion No.16 and 17 were effective and thus goodwill was only recognized under the purchase method and was subject to amortization. They found
that reported goodwill is value relevant, indicated by the positive and significant coefficients on goodwill. On the basis of this result McCarthy and Schneider conclude that the market uses goodwill information in the valuation of a firm and therefore perceives goodwill as an asset.
The study of Jennings, Robinson, Thompson and Duvall (1996) also investigates the value relevance of goodwill in the US. In their research they assess the association between the market capitalization and reported goodwill during the period 1982-1988. Their results also show a positive and significant coefficient on goodwill and therefore suggest, in line with McCarthy and Schneider (1995), that goodwill is perceived as an asset by investors. Jennings et al. (1996) also explore in their research the association between market capitalization and goodwill amortization numbers. Their results provide only very little evidence that goodwill amortization numbers are value relevant. So from these results it can be suggested that because the amortization amounts of goodwill are not considered to be relevant and reliable (i.e. not value relevant) it is likely that this has also an effect on the value relevance of the goodwill amount on the balance sheet, because that is used as the basis for the amortization. This result of Jennings et al. and the suggestion are interesting because under the new standards SFAS 141 and 142 this amortization approach is eliminated and a new method, the impairment approach, is introduced.
In light of the above, the study of Hirschey and Richardson (2002) is interesting in two ways. In their research they examine the effect of goodwill write-off announcements on stock prices in the US during the period 1992-1996.In this period goodwill write-off were accounted for under SFAS 121 - Accounting for the Impairment of Long-Lived Assets and for
Long-Lived Assets to Be disposed; however because this standard is completely superseded in
2001 and not very relevant for this thesis it is not explored here in detail.As stated earlier, this research is interesting in two ways. In the first place, if there is a significant effect of the goodwill write-off announcement on stock prices it indicates write-offs (i.e. impairments), in comparison to amortization, provide more useful information to investors. In the second place, if goodwill write-off announcements have a significant effect on stock prices this gives indirect evidence that goodwill amounts are valuable for investors and thus value relevant. The results of Hirschey and Richardson show that the effect of goodwill write-off
announcements on stock price is negative and significant, it namely leads to a 2-3% adverse effect on stock value.On the basis of these results it is clear that goodwill write-offs
usefulness of these goodwill write-off announcements gives therefore also indirect evidence that the underlying goodwill is also relevant and to some degree reliable, thus value relevant.
A study that focuses on the value of relevance of goodwill on the basis of the new standard SFAS 142 is the study of Chen, Kohlbeck and Warfield (2004). They explore in their research the effect of the implementation of SFAS 142 on the value relevance of goodwill as reported the year 2002. Firstly, they assess the association between the market value of equity and goodwill including all adoption and year 1 impairments (i.e. SFAS 142 practice). Secondly, they look at the association between the market value of equity and goodwill excluding all the impairments (as indication for the pre-SFAS 142 practice where there was no impairment). Their results show for both specifications a significant and positive coefficient on goodwill; however the coefficient for the SFAS 142 specification is greater than the coefficient of the pre-SFAS specification, respectively 2.74 and 1.14. On the basis of these results there is preliminary evidence that goodwill under SFAS 142 is value relevant and that it provides even more useful information than under the old standards.
3.2 The Reliability of Financial Information
As explained earlier, value relevant financial information is considered to be relevant, but only reliable to some extent. On the basis of the discussed literature on the value relevance of goodwill amounts it is clear that goodwill is perceived as relevant by investors and to be measured reliable to some degree. Thus the evidence whether goodwill amounts are reliable is limited, therefore it is important to explore the reliability of reported goodwill in further detail. Remember that according to the FASB (2010) reliable information is complete, neutral and free form error. The reliability of financial information will be explored on the basis of the agency theory (Jensen & Meckling, 1976) and the positive accounting theory (Watts & Zimmerman, 1986, 1990). After exploring these theories previous literature on the reliability of goodwill amounts will be discussed.
3.2.1 The Agency Theory and Positive Accounting Theory
Jensen & Meckling (1976) set out the agency theory by first defining an agency relationship as “a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf which involves delegating some decision making authority to the agent” (p. 308). So within an agency relationship there are two
parties the principal and the agent who should act on behalf of the principal. However, as Jensen & Meckling notice both parties have their own interests and therefore it is possible that the agent doesn’t act in the best interest of the principal (also called the agency problem).
In nowadays large companies there is, in general, a separation of ownership and control. This separation of ownership and control results in a typical agency relationship between the owners/shareholders (i.e. the principals) and the management (i.e. the agents) (Jensen & Meckling, 1976). So this creates the possibility that an agency problem exists between shareholders and the management; where the management acts in favor of their own interest rather than in the interest of the shareholders. The management of large companies exists of different managers, however the two most important managers are the chief
executive officer (CEO) and the chief financial officer (CFO). In this thesis the focus lies on the CFO, because it is the CFO who is responsible for implementing accounting standards, making decisions about accounting choice and preparing the financial statements (Balsam, Irani, & Yin, 2012; Indjejikian & Matejka, 2009; Loyeung & Matolcsy, 2015). In other words, it is the CFO who is responsible for the financial reporting of a company.
Another important accounting theory, which is related to the agency theory, is the positive accounting theory (Watts & Zimmerman, 1986, 1990). The fact that financial
reporting, to a large extent, is based on judgments and estimates (FASB, 2010; Fields, Lys, & Vincent, 2001) gives management discretion in how to apply accounting standards. The positive accounting theory tries to provide insight about how management uses this
discretion. It assumes that accounting choice can be used in two ways (Watts & Zimmerman, 1986, 1990). On one hand, the accounting discretion could be used to provide unbiased and reliable information to increase the wealth of shareholders. However, managers could also use accounting choice to provide biased and unreliable information in order to maximize their own interests. In the existing literature the latter is also referred to as earnings management2, this term suggest that it only concerns earnings; but for the purpose of this thesis I use this term to refer to the manipulation of financial information in general.
On the basis of the agency theory and the positive accounting theory it is clear that a CFO (as an agent in the agency-relationship with the shareholders) has incentives to
maximize his own interests and the discretion in accounting creates the opportunity to fulfill
2
Healy and Wahlen (1999) provide an definition of earnings management, namely “earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers” (p. 386).
this. These incentives derive according to Watts and Zimmerman (1978) from the fact that a manager’s compensation, among other things, exists of incentive compensation, e.g. bonuses, stock and stock options. In relation to stock and stock options (i.e. equity compensation), Healy and Wahlen (1999) argue that because financial information is used in valuing a firm (and thus stocks) managers have the incentive to manipulate this financial information to affect the value of stock prices. This is also referred to, by Healy and Wahlen, as the capital market motivation. Related to this, Fuller and Jensen (2002) argue that due increased equity compensation managers are more focusing on increasing short-term stock prices and
therefore overvalue stocks. But why is it interesting for the managers to improve short-term stock prices? Well, by improving the short-term stock price they also improve the value of their equity compensation and therefore they increase their own wealth. So it is clear that managers who are compensated on the basis of equity have incentives (i.e. equity incentives) to manipulate financial information for their own interest. Note that the capital market motivation for manipulating financial information is the focus of this thesis.
In the next sections extant literature on whether CFOs are likely to use accounting discretion to improve their own interests is discussed. Furthermore, there will be a discussion, on the basis of previous literature, whether the discretion in SFAS 141 and 142 leads to unreliable goodwill amounts.
3.2.2 Previous Literature on Equity Incentives to Manipulate Financial Information
A leading study about the effect of equity incentives on earnings management, i.e.
manipulation financial information, is the research of Cheng and Warfield (2005). In their research they explore whether CEOs with high equity incentives are more likely to engage in earnings management. In other words, they explore the capital market motivation of Healy and Wahlen (1999).As indication for earnings management they focus in their study on the extent that reported earnings are meeting or just beating analysts’ forecasts. Their results show that the probability that CEOs report earnings that meet or just beat analysts’ forecasts increases when CEOs have higher equity incentives. On the basis of this study it is clear that managers with equity incentives are likely to manipulate financial information with the aim to increase their own wealth.
The study of Cheng and Warfield (2005) focuses on CEOs, however because CFOs are responsible for the financial reporting it is likely that CFOs have more influence on manipulating financial information. Before exploring empirical research on whether CFOs
manipulate financial information for their own interest, it is interesting to discuss the study of Dichev, Graham, Harvey and Rajgopal (2013) which gives insight on why CFOs engage in earnings management. The results of their study, based on surveys and interviews, show that the main reasons why CFOs use accounting discretion to report manipulated accounting amounts is to influence stock price, to hit earnings benchmarks and to influence executive compensation.
Now it is clear why CFOs are likely to manipulate financial information it is turn to discuss whether CFOs indeed manipulate financial information on the basis of extant
literature. One study that explores whether CFOs engage in earnings management is the study of Geiger and North (2006). They assume that if a CFO has really an impact on financial reporting than the appointment of a new CFO should lead to different reported results. In their study they focus on the change in discretionary accruals, which are defined as “the difference between actual levels reported and the level expected to be reported by the
company for the period” (Geiger & North, 2006, p. 785). In the extant literature discretionary accruals are used as indication of the manipulation of accounting amounts (see e.g. Healy, 1985). Geiger and North assume on one side that in the period before the appointment of a new CFO the current CFO has incentives to increase discretionary accruals to maximize his compensation just before being fired. While on the other side they assume that the new CFO has incentives to decrease accruals in order to show higher financial amounts in the future. Their results, indeed, show that in the period before the appointment of the new CFO the amount of discretionary accruals is much higher for firms that hire a new CFO than for firms that don’t. Furthermore, the results show that the amount of discretionary accruals reduces under the new CFO. This study shows thus two important things, namely on one hand that CFOs have a significant impact on financial reporting and on the other hand that CFOs manipulate financial information for their own interest.
In the extent of the study of Geiger and North (2006), which provides evidence that CFOs impact and manipulate financial reporting to maximize own wealth, the study of Jiang, Petroni and Wang (2010) is interesting. In their study they examine the impact of CFO’s equity incentives on earnings management. To be more specific, Jiang et al. (2010) examine the effect of CFO’s equity incentives on the amount of accruals (both total and discretionary) and the probability of beating analysts forecast; both are indications for manipulating
accounting amounts (see e.g. Healy, 1985; Cheng & Warfield, 2005). Overall, their results show that accruals (total and discretionary) and the likelihood of beating analysts forecast are increasing in CFO’s equity incentives. Also interesting is that the results show that CFO’s
equity incentives have more impact on both measures than CEO’s equity incentives, this suggest that CFOs have more influence on manipulating financial information than CEOs.
3.2.3 Discretion in the Accounting for Goodwill under SFAS 141 and 142
The aim of this section is to explore the accounting discretion in SFAS 141 and 142 which could be used by managers to provide biased and unreliable information with the goal to maximize their own interest. The focus will be both on managerial discretion when goodwill is initially recognized (i.e. purchase price allocation) as well as on discretion in the
subsequent measurement of goodwill (i.e. impairment testing).
As previously discussed, SFAS 141 requires that the acquiring entity allocates the purchase price of the acquired entity to the acquired assets and assumed liabilities at their fair value. The excess of the purchase price over the fair value of the net assets acquired is than recognized as goodwill. So it clear that during the purchase price allocation process the acquiring firm has to measure the acquired assets and assumed liabilities at their fair value. Fair value is defined as “the price that would be received to sell an asset or paid to transfer a liability in an orderly transaction between market participants at the measurement date” (FASB, 2006, p. 2). However, in order for the measurement of the fair values to be reliable it is important that liquid markets exits for the assets and liabilities (Kothari, Ramanna, & Skinner, 2010; Lhaopadchan, 2010). Shalev, Zhang and Zhang (2013) remark that this is likely to be the case for marketable securities, however for other assets (in particular intangible assets) and liabilities there are no observable prices. They note that the
consequence of no observable prices is that the fair value of these assets and liabilities has to be estimated and this of course involves judgment.
Managers could make use of this discretion in the purchase price allocation process to overstate goodwill by understating the fair values of the net assets acquired so that the excess of the purchase price over this fair value increases. Another (simpler) manner to overstate goodwill is to push up the purchase price. But why would managers overstate goodwill? One reason relates to the capital market motivation of Healy and Wahlen (1999). Managers could have incentives to overstate goodwill to affect the short-term value of stock prices. Another reason why managers would overstate goodwill relates to the subsequent
measurement of goodwill. Goodwill under SFAS 142 is namely subject to impairment testing instead of amortization, unlike other intangible assets (with a finite useful life). So by
managers reduce the amount of amortization and thus increase profits. However, overstating goodwill increases the possibility of a large impairment in the future (Zhang & Zhang, 2015). This brings us to exploring the discretion that managers have in impairment testing under SFAS 142.
After the recognition of goodwill on the basis of SFAS 141 the subsequent
measurement is addressed by SFAS 142. As discussed earlier, under SFAS 142 goodwill has to be tested for impairment at the reporting unit level on the basis of a two-step approach. In short, under this two-step approach, firstly the fair value of the reporting unit is compared with its carrying amount (including goodwill). Secondly, the implied fair value of goodwill is compared with its carrying amount, if the carrying amount exceeds the implied fair value the excess is recognized as a goodwill impairment loss.
Hayn and Hughes (2006) as well as Ramanna and Watts (2012) argue that the impairment testing of goodwill as addressed by SFAS 142 creates discretion in three manners. In the first place, the total acquired amount of goodwill has to be assigned to reporting units. However, defining a reporting unit requires subjective judgment and thus introduces discretion (Beatty & Weber, 2006; Jerman & Manzin, 2008). So the facts that managers have to subjectively define reporting units and then determine how to allocate goodwill to the units, creates the possibility that managers allocate as much as goodwill to ‘self-defined’ reporting units that have a high implied fair value of goodwill to avoid or reduce impairments (Zang, 2008). Secondly, the fair value of the reporting unit has to be determined. However, as discussed earlier a fair value measurement is only reliable when there are observable prices. In light of this Bens (2006) notes that it is not likely that these are available for the reporting units. The consequence of this is that management has to use its own estimates and therefore judgment to determine the fair value of the reporting units. Thirdly, in order to determine the implied fair value of goodwill the fair values of the assets and liabilities of the reporting unit have to be determined. As mentioned earlier, for most assets and liabilities there are no observable prices to be used as fair value, thus the fair values have to be determined on the basis of estimates and subjective judgment.
It is clear that the impairment testing of goodwill under SFAS 142 creates a lot of discretion. However, it is important to note here that, as discussed before, discretion itself is not a bad thing. It depends on the intentions of management; if management uses the
discretion in order to manipulate financial information then it is harmful for the shareholders. For example, managers could use the discretion under SFAS 142 in order to understate impairment with to aim to overstate goodwill because of capital market motivations.
Furthermore, managers could avoid or reduce impairments in order to show higher earnings, with the hope that there will be no impairment in the future. In the next section previous literature on the effect of discretion in both the purchase price allocation (i.e. SFAS 141) as in the impairment testing (i.e. SFAS 142) is explored in order to examine whether this discretion is likely to be used by managers to maximize their own interests.
3.2.4 Previous Literature on the Reliability of Goodwill
A study that explores the reliability of initial recognized goodwill under SFAS 141 is the study of Shalev, Zhang and Zhang (2013). Shalev et al. explore in their study the effect of CEO compensation on the purchase price allocation process. They examine whether CEOs with bonuses as a large part of their compensation are more likely to allocate a higher amount of the purchase price to goodwill and thus overstate goodwill amounts. Shalev et al. assume that CEOs have the incentive to allocate a higher proportion of the purchase price to goodwill (instead of to other intangible assets) due to the fact that goodwill is not subject to
amortization and that the impairment tests under SFAS 142 provide discretion which could be used to reduce or eventually avoid impairments in order to show higher earnings. The incentive to show higher earnings relates, according to Shalev et al., to the fact that earnings are used as performance measure for the bonuses. The results of their research confirm that CEOs with significant bonuses in their total compensation allocate more of the purchase price to goodwill. In other words, the research of Shalev et al. shows that managers are likely to overstate goodwill to maximize their own interests; in the case of Shalev et al. to maximize their bonuses.
Another study that focuses on CEO incentives to overstate goodwill by using the discretion in the purchase price allocation process is the study of Zhang and Zhang (2015). They argue that because older CEOs have a short horizon and less career concerns they are more focused on short-term performance in order to maximize their compensation. Therefore Zhang and Zhang expect that older CEOs are more likely to allocate a greater amount of the purchase price to goodwill instead to other identifiable intangible assets, with the same reasons as discussed above. The results of their research are consistent with their view that older CEOs allocate a higher proportion of the purchase price to goodwill in contrast to other intangible assets. In order to strengthen their results Zhang and Zhang also explore the purchase price allocation process in the pre-SFAS 141 and 142 period. Interestingly, the results of this test do not provide evidence that older CEOs allocate a higher amount to
goodwill. So, on the basis of the results of Zhang and Zhang it is also clear that managers overstate goodwill for their own interests and that the overstatement is likely the consequence of the discretion under SFAS 141 and 142.
A related study that explores the overstatement of goodwill at initial recognition is the study of Olante (2013); however Olante focuses on the overstatement of goodwill due to overpayments. To be more specific Olante explores the reliability of goodwill by examining whether impairments arise as due to overpayments rather than as consequence of a decline in the economic value of goodwill. In addition to this Olante also examines whether the
impairment testing under SFAS 142 improved the timely recognition of impairments in situations where goodwill is likely to be overstated. Her research is based on a model that predicts impairments on the basis of overpayment indicators and uses a sample of
acquisitions in the US during the period 1999-2007. The results of this model show in the first place that some of the used overpayment indicators predict the occurrence of an impairment loss. To be more specific, Olante (2013, p. 252) shows that 37.4% of the impairment losses were predicted by the significant overpayment indicators. This result shows that it is likely that managers overstate goodwill by overpaying acquisitions.
Secondly, the results show that an impairment loss is recognized, on average, after two or three years following the initial recognition of goodwill. Olante compares this result with the results of Hayn and Hughes (2006) their research on the timely recognition of
impairments. Hayn and Hughes show that impairment losses are recognized, on average, after four or five years during the period 1988-1998; i.e. in the pre-SFAS 142 period. On the basis of this comparison, Olante concludes that impairment recognition is timelier under the new standard SFAS 142. So this suggests that although the initial recognition of goodwill is not realiable as it is overstated due overpayments; the subsequent impairment tests improve the reliability of goodwill to some extent by providing timelier impairments.
Ramanna and Watts (2012) explore the discretion under SFAS 142 in more detail by examining whether managers use the discretion to provide inside, private information or to provide biased and unreliable information for their own interests. For their research they use a sample of firms which have for two years in a row a book-to-market ratio that exceeds one and they use that as criterion for the indication of goodwill impairment. The results of Rammana and Watts (2012, p. 764) show in the first place that 69% of the firms with an indication of goodwill impairment do not record any goodwill impairment. In the second place, they show that non-impairment is not affected by managers their inside, private information about positive future cash flows. However, their results show that some agency
motives explain the avoidance of impairments. To be more specific, they find significant coefficients for CEO compensation, CEO reputation and for debt covenants concerns. So it is likely that managers use the discretion under SFAS 142 to manipulate financial information for their own interests instead of providing private information useful for investors.
In contrast to Rammana and Watts (2012), who focus in essence on the timing of impairments to explore whether the discretion under SFAS 142 is used by managers for their own interests, Zang (2008) examines the discretion by focusing on the amount of
impairments. Specifically, Zang explores whether managers use the discretion at the initial impairment test of goodwill when SFAS 142 was adopted (in 2001) to affect the amount of the impairment. Zang expects that firms with high leverage, therefore more likely to have debt covenant concerns, are more likely to use the discretion to recognize lower initial
goodwill impairments. Furthermore, they expect that firms that have a change in management are more likely to recognize a higher initial amount of goodwill impairment, so that the new managers can report higher earnings in the future (see also Geiger & North, 2006). The results of Zang provide evidence for both predictions and therefore Zang concludes that impairment amounts are driven by managers’ incentives (similar to the timing of impairments according to Rammana and Watts (2012)).
3.3 Hypothesis Development
On the basis of the previous sections it is clear that goodwill is increasing in both importance as in proportion. Extant literature (e.g. McCarthy & Schneider, 1995; Jennings et al., 1996) shows that goodwill, even before the introduction of the more sophisticated standards SFAS 141 and 142, is value relevant and perceived as an asset by the market. To improve the reporting of goodwill the FASB introduced SFAS 141 and 142 in 2001. The adoption of SFAS 141 increased the importance and proportion of goodwill even more. Furthermore, SFAS 142 introduced impairment testing of goodwill and the extant literature shows that goodwill impairments are value relevant compared to goodwill amortization expenses (see Jennings et al., 1996 and Hirschey & Richardson, 2002). This all suggest even an increase in the value relevance of goodwill and this is also preliminary confirmed by Chen et al. (2004). Therefore I hypothesize that:
Hypothesis 1: Goodwill under SFAS 141 and 142 is positively associated with market values (i.e. goodwill is value relevant).
The accounting standards SFAS 141 and 142 are introduced by the FASB to improve the reporting of goodwill. However, as discussed previously, both standards provide
discretion in their application. Discretion in accounting can be used by managers either to provide inside, private information that is useful for investors or to provide biased and unreliable information for own interests. The agency theory and the positive accounting theory suggest that because managers have their own interests they are likely to use discretion in accounting to manipulate information to maximize their own wealth. Previous literature shows that CFOs, as representatives of the financial reporting of a firm, have the ability to impact financial reporting and indeed manipulate financial information to maximize their own wealth (see e.g. Geiger & North, 2006; Jiang et al., 2010). Furthermore, extant literature shows that the discretion under SFAS 141 and 142 can be used in order to manipulate
goodwill amounts for own interests. For example, discretion in the purchase price allocation process is used to allocate a higher proportion of the purchase price to goodwill (see Shalev et al., 2013; Zhang & Zhang, 2015). Furthermore, the discretion in the impairment testing gives managers the opportunity to influence the timing and amount of impairments (see Ramanna & Watts 2012; Zang, 2008).
As value relevant goodwill is perceived to be relevant and thus seen as an asset, it is likely that investors use reported goodwill in valuing a firm. Therefore CFOs with more equity incentives are likely to have capital market motivations to overstate goodwill in order to boost short-term stock prices and therefore increase their equity compensation. On the basis of this all I hypothesize that:
Hypothesis 2: The amount of reported goodwill will be higher under CFOs with high equity incentives compared to CFOs with low equity incentives.
4. Research Methodology
4.1 Sample and Data Selection
In order to test my hypotheses I conduct a quantitative research. To be more specific, this thesis is based on an archival research using publicly available databases. First of all I obtain financial statement and stock data for U.S. listed firms from the database CRSP/Compustat
Merged. Secondly, I obtain data on CFO compensation from the database ExecuComp. The database ExecuComp includes data on compensation of different types of executives; therefore I make use of the option annual title in order to obtain only CFO information. Following Balsam (2012) and Jiang et al. (2010) I search for CFO compensation data by looking for annual titles that include CFO, chief financial officer or chief finance officer.
The two datasets cover a five-year period from 2002 to 2007. Although the
accounting for goodwill under SFAS 141 and 142 has been effective since 30 June 2001, I exclude the adoption year 2001 from the sample period. The exclusion of the adoption year is because of the fact that the FASB (2001b) required that the initial impairment of goodwill as consequence of the adoption of SFAS 142 had to be reported in the income statement as a result of an accounting change, i.e. below-the-line. This requirement places managers in a one-off situation with specific incentives to deal with the impairment of goodwill (Hayn & Hughes, 2006; see also Beatty & Weber, 2006; Zang, 2008) and therefore the year 2001 is not included in the sample period. Furthermore, the sample covers a period only until 2007, because of the occurrence of the financial crisis in 2007-2008 which had a huge impact on stock markets. Therefore in order to avoid the influence of the financial crisis on stock data (i.e. stock prices and market capitalization) I only focus on the period from 2002 to 2007.
Table 1 summarizes the sample selection procedures taken to obtain a unique, merged dataset that covers financial statement and stock data as well as data on CFO compensation. The initial sample obtained from CRSP/Compustat Merged (excluding the financial industry) includes 22,176 firm-year observations. From this initial sample a number of 9,849
observations are deleted, because either the firms have no goodwill or goodwill data is missing. Subsequently 245 firm-year observations with negative book values of equity are removed, because a negative book value of equity indicates an uncertain future and therefore the inclusion of these firms probably leads to bias of the results (Dahmash, Durand &
Watson, 2009). Next, I remove a number of 1,722 observations with missing data on either net income or market value of equity. In order to get the unique, final sample I merge the CRSP/Compustat Merged sample with the sample derived from ExecuComp. The combining of the two datasets generates a sample of 4,280 firm-year observations. Next, I test the merged sample on duplications and as a result 96 observations are removed. The result of these sample selection procedures is a final sample of 4,184 firm-year observations for the period 2002-2007. Last but not least I winsorize the data on the test variables (which are discussed in the following section)at 1% and 99% of their distributions in order to deal with outliers.
4.2 Research Design
The purpose of this thesis is to examine the value relevance (H1) and the reliability (H2) of goodwill on the basis of an archival research. The examining of the value relevance and reliability of goodwill will be done on the basis of a value relevance test; to be more specific by the use of an incremental association test (see Holthausen & Watts, 2001). As previously discussed in section 3.1, value relevance test are used to determine whether there is an association between stock data and accounting amounts and are performed on the basis of a valuation model (Barth et al., 2001; Holthausen & Watts, 2001). An example of these models is the valuation model of Ohlson (1995). The purpose of the Ohlson model is to provide a
Table 1
Sample Selection
Observations deleted Number of observation Initial Sample CRSP/Compustat Merged (2002-2007) - 22,176 Delete observations with no goodwill amounts or missing data 9,849 12,327 Delete observations with negative book
value of equity Delete observations with missing data on
either net income or market value of equity 245 1,722 12,082 10,360 Merge the CRSP/Compustat
sample with the sample of ExecuComp Delete duplications 6,080 96 4,280 4,184
model that specifies the relation between on one side the market value of equity and on the other side the book value of equity and net income. The Ohlson model is specified as follow:
MVE = 0 + 1BVEt + 2NI + (1)
where:
- MVEt is the market value of equity at year t;
- BVEt is the book value of equity at year t;
- NIt is the net income at year t.
The Ohlson model will serve as the basis for my own regression models.
In order to test the value relevance of goodwill (H1) I need to modify the Ohlson (1995) model by including goodwill as an independent variable. So my first regression model is constructed as follow:
MVE = 0 + 1BVE + 2NI + 3GW + (2)
where:
- MVEit is the market value of equity for firm i at end of fiscal year t;
- BVEit is the book value of equity (excluding goodwill) for firm i at end of fiscal year t;
- NIit is the net income for firm i at end of fiscal year t;
- GWit is the goodwill for firm i at end of fiscal year t.
It is the coefficient on goodwill (i.e. 3) which is of interest for this value relevance test of
goodwill. If this coefficient 3 is positively and significant it provides evidencethat goodwill
is value relevant (i.e. goodwill is relevant and reliable to some extent). As probably noted I use non-deflated variables in my modified Ohlson model.The reason thatI choose for a non-deflated model is because Barth and Clinch (2009) show in their study that the unnon-deflated model and the share-deflated model are the best specifications of the Ohlson model to reduce scale effects and bias in the coefficients.
In order to examine the reliability of goodwill (H2) in further detail I will extend the previous regression model (i.e. equation 2). However, before presenting the model it is necessary to explain the steps taken to create this model and important to define the variables used. As noted in section 3.3 I examine the reliability of goodwill by exploring whether
CFOs with high equity incentives are more likely to report a higher amount of goodwill, in other words whether they overstate goodwill. So I need to define a variable that captures the equity incentives of a CFO. For this purpose I use the research method of Bergstresser and Philippon (2006). Bergstresser and Philippon construct in their study (in which they examine the effect of CEO equity incentives on earnings management) a measure that captures the power of equity incentives. In the first place, Bergstresser and Philippon define a measure, ONECPTit3, that captures ‘the dollar change in the value of a CEO’s stock and options
holding that would come from a one percentage point increase in the company stock price” (p. 519). I use this measure, ONECPTit, and I constructed it for the purpose of my thesis as
follow:
ONEPCT = 0.01 * Price * (Shares + Options ) (3)
where:
- Priceit is the share price of firm iat end fiscal year t;
- Sharesit is the number of shares held by the CFO of firm i in year t;
- Optionsit is the number of options4 held by the CFO of firm i in year t.
Subsequently, Bergstresser and Philippon use this measure ONECPTit to calculate another
measure, INCENTIVE_RATIOit5, which according to them “captures the share of a
hypothetical CEO’s total compensation that would come from a one percentage point increase in the value of the equity of his or her company” (p.520).
For the purpose of my thesis I construct this measure as follow:
EQ_INCENTIVES = ONEPCT / (ONEPCT + SALARYit + BONUSit) (4)
where:
- SALARYit is the salary of the CFO of firm i in year t;
- BONUSit is the bonus of the CFO of firm i in year t.
3 Bergstresser and Philippon (2006) use in their study the measure ONECPT
it for CEOs; however the measure
is also applicable for CFOs (see Jiang et al., 2010).
4 Following the studies of Bergstresser and Philippon (2006) and Jiang et al. (2010) I use the following types of
stock options in order to calculate the number of options held by a CFO: newly granted options, unexercised exercisable options and unexercised unexercisable options. Note that both Bergstresser and Philippon (2006) and Jiang et al. (2010) in turn follow the study of Core and Guay (2002).
5 Bergstresser and Philippon (2006) use in their study the measure INCENTIVE_RATIO
it for CEOs; however
It is this variable EQ_INCENTIVESit (i.e. the equity incentive ratio) which I use to capture
the power of CFOs their equity incentives.
In order to distinguish CFOs with high equity incentives and CFOs with low equity incentives I separate the data on EQ_INCENTIVESit in two parts, namely above the median
and below the median. The data above the median represents CFOs with high equity incentives, while the data below the median represent CFO with low equity incentives. In order to capture the effect of high equity incentives versus low equity incentives I create a dummy variable DEQ_INCENTIVES which equals 1 for firms with a CFO with high incentives
(and 0 for firms with a CFO with low equity incentives).
On the basis of the variables discussed and constructed above I specify my second (extended) regression model as follow:
MVE = 0 + 1BVEit + 2NI + 3GWi + 4BVEit*DEQ_INCENTIVES
+ 5NI *DEQ_INCENTIVES + 6GWi *DEQ_INCENTIVES + (5)
where:
- DEQ_INCENTIVES is a dummy variable which equals 1 if the CFO has an equity incentive ratio
above the sample median (i.e. has high equity incentives); it equals 0 if the CFO has an equity incentive ratio below sample median (i.e. has low equity incentives).
The coefficients of this model that are of interest for examining the reliability of goodwill are the coefficients 3and 6.The coefficient 3 capturesthevalue relevance of goodwill for firms
with a CFO who has low equity incentives; while 6 captures the difference between the
value relevance of goodwill for firms with a CFO who has low equity incentives and firms with a CFO who has high equity incentives. But how do these coefficients tell us more about the effect of equity incentives on the reliability of goodwill? Well, according to Holthausen and Watts (2001) the manipulation of financial information (trough noise and bias) can negatively affect the value relevance of this information. They further note that in an incremental association study (the type of study used in this thesis) this negative impact on the value relevance can be reflected in the coefficients. The manipulation of financial information reduces the reliability of the information, so the decrease in value relevance is likely to be due to less reliable information, instead of less relevant information.
So returning to the coefficients 3 and 6, it should be the case that the coefficient on
coefficient on goodwill for firms with CFOs with high equity incentives (i.e. 3 + 6). This is
because CFOs with high equity incentives, compared to CFOs with low equity incentives, are as hypothesized more likely to overstate goodwill trough manipulation and this leads to a lower coefficient on goodwill. In other words 6, which captures the difference between
firms with a CFO with low and high equity incentives, has to be negative and significant in order to show that the coefficient on goodwill for firms with high equity incentives is lower than for firms with low equity incentives. This result then provides evidence that CFOs with high equity incentives overstate goodwill and therefore the goodwill amounts of these firms are less reliable.
Last but not least, following the arguing of Kallapur and Kwan (2004) I also include the interactions between the dummy DEQ_INCENTIVES and BE and NI, respectively, in the model
as control variables to ensure that 3 and 6 are not affected by the association between
market values and these variables.
5. Results
5.1 Descriptive Statistics and Correlation Analysis
Prior to the testing of the hypotheses it is interesting to examine the obtained data. I examine the data by looking at the descriptive statistics of the main variables and the correlations of the test variables. The descriptive statistics and the correlations are presented in table 2 and table 3, respectively.
Table 2 shows that the mean of market value of equity is $9,074.97 million and the mean of book value of equity (excl. goodwill) is $1,817.92 million. As consequence of the huge difference between market value of equity and book value of equity the sample firms have, on average, a market-to-book ratio of approximately 5. According to Jenkins and Upton (2001) this huge market-to-book gap is, amongst other things, due to the lack of recognition of intangible assets. Furthermore, table 2 shows that the sample firms report, on average, an amount of $1,176.57 million in goodwill on their balance sheets. The inclusion of the goodwill in the book value of equity reduces the market-to-book ratio, on average, with approximately 40%; which is in line with the arguing of Jenkins and Upton (see also Ayers et al., 2000). The amount of reported goodwill furthermore represents approximately 16% of the total reported assets. This percentage is rather in line with the results of Hayn and Hughes
(2006, p. 41), because they show that during the period 1988-2001 goodwill as percentage of total assets increased from 10.7% to 16.8%. These descriptive statistics indicate the increased importance and proportion of goodwill; however it is interesting to note that the reported goodwill amounts of the sample firms have a wide range with a minimum of $1.14 million and a maximum of $17,397 million. Furthermore, the mean (median) of net income for the sample firms is $438.55 million ($90.62 million). The descriptive statistics of the equity incentive ratio show that the sample firms’ CFOs have, on average, an equity incentive ratio around 17%, while the median is about 13%. Moreover, the equity incentive ratios range from a minimum of 0.04% to a maximum of 65.69%; which indicates that the sample includes CFOs with high equity incentives, but also CFOs with practically no equity incentives. Note here that the equity incentive ratio median, 0.1322, is used to make a distinction between CFOs with high equity incentives (i.e. above the median) and with low equity incentives (i.e. below the median); see for this section 4.2.
In addition to the descriptive statistics of the main variables I also analyze the correlations between the variables used in the regression models (i.e. the test variables). However, before analyzing the correlations it is very important to note that correlations only Table 2
Descriptive Statistics of the Main Variables
Variable n Mean SD Median Minimum Maximum
MVE 4,184 9,074.97 20,933.23 2,117.69 110.62 142,256 BVE 4,184 1,817.92 4,331.43 440.11 -1,436 29,196 NI 4,184 438.55 1,158.46 90.62 -1,078.99 7,831 GW 4,184 1,176.57 2,713.67 240.23 1.41 17,397 GW/ASSETS 4,184 0.1618 0.1479 0.1218 0.0004 0.7553 EQ_INCENTIVES 4,184 0.1654 0.1301 0.1322 0.0012 0.6569
Note. The variables MVE, BVE, NI, GW and ASSETS are measured in millions of $. MVEis market value of equity; BVE is book value of equity (excluding goodwill); NI is net income; GW is goodwill; GW/ASSETS is goodwill divided by total assets; EQ_INCENTIVES is the equity incentive ratio.
indicate whether there is a linearrelationship between variables. So it is crucial to acknowledge that correlation has nothing to do with causality. Table 3 presents the correlations among the test variables.
Table 3
Correlations among the Test Variables (Pearson)
MVE BVE NI GW DEQ_INCENTIVES
MVE 1.0000
BVE 0.8623 1.0000
NI 0.9010 0.8357 1.0000
GW 0.6662 0.5594 0.6089 1.0000
DEQ_INCENTIVES 0.2437 0.1980 0.2300 0.1745 1.0000 Note. MVE is market value of equity; BVE is book value of equity (excluding goodwill); NI is net income; GW
is goodwill; DEQ_INCENTIVES is a dummy variable which equals 1 if the CFO has an equity incentive ratio above the
sample median (i.e. has high equity incentives); it equals 0 if the CFO has an equity incentive ratio below sample median (i.e. has low equity incentives).
In the first place it is interesting to note that all the presented correlations in table 3 are significant at the 1% level (i.e. p < 0.01). The correlations table shows that the variables book value of equity, net income and goodwill are positively correlated with market value of equity. The positive correlation of goodwill with market value of equity (66.62%) can be seen as a preliminary indication of the value relevance of goodwill. However, as stated earlier, the correlation only indicates that goodwill and market value of equity are linearly related to a certain degree. Furthermore, the dummy variable DEQ_INCENTIVES is positively correlated with
all other variables. Considering the previous saying about correlations, the positive correlation between DEQ_INCENTIVES andgoodwill (17.45%) can be seen as a preliminary
indication in favor of the second hypothesis. Last but not least, due to the significant correlations among all the independent variables (BVE, NI, GW and DEQ_INCENTIVES) I
checked for multicollinearity problems on the basis of the variance inflation factor (VIF). However, the test provides no indication for multicollinearity (all the VIFs are below 5).
