Revisions in future earnings forecasts : changes in perceptions following a goodwill impairment loss : a study of U.S. companies

(1)

MSc Accountancy & Control, track Accountancy Faculty of Economics and Business, University of Amsterdam

Revisions in Future Earnings Forecasts:

Changes in Perceptions Following a Goodwill

Impairment Loss

A study of U.S. companies

Name: Rémy Hendrikx Student number: 10596879

Supervisor: Dr. G. Georgakopoulos Date: 21 June 2015

(2)

Statement of Originality

This document is written by student Rémy Hendrikx 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.

(3)

Abstract

This paper examines the consideration of goodwill impairment losses by financial analysts in revising future earnings forecasts following the mandatory adoption of Statement of Financial Accounting Standards (SFAS) 142 ‘Goodwill and Other Intangible Assets’ in 2001. SFAS 142 was issued to improve the accounting treatment for goodwill since analysts and other financial statement users argued that goodwill amortization expenses did not provide useful information regarding analyzing investments. Hence, ex-ante, it is unclear whether the impairment-only approach is more relevant for decision-making. Using data from U.S. firms., the empirical results reveal that the impairment-only approach is not more value-relevant to financial analysts. Furthermore, no significant impact on this relation is found when managers have incentives to use big bath accounting. At last, this study indicates that the financial analysts’ revision of future earnings forecasts is not affected by goodwill impairment losses during the years of the financial crisis.

(6)

1

1 Introduction

Goodwill is considered to be a part of Intangible Assets under both US GAAP and IFRS. Within the Accounting Terms of Codification Topic 350, Intangibles – Goodwill and Other (Financial Accounting Standards Board, 2011), an intangible asset is defined as a non-monetary asset without physical substance. In 2001, the U.S. Financial Accounting Standards Board (FASB) issued a new standard in order to eliminate pooling of balance sheets in favor of purchase accounting (Financial Accounting Standards Board, 2001). The recognition criteria for the accounting model of intangible assets requires that there be probable future economic benefits from costs that can be reliably measured. Goodwill is therefore an asset that is not physical in nature and recognized only in a business combination (Financial Accounting Standards Board, 2001a). The main purpose of this study is to investigate the economic consequences of a goodwill impairment loss, due to the Statement of Financial Accounting Standards (SFAS) 142. At first, I will examine whether a loss of goodwill impairment is reflected into the forecast of earnings determined by financial analysts. Prior research has shown that impairment in the value of goodwill indicates that the expected benefits of an acquisition are overestimated on the balance sheet (Morck et al., 1990). If analysts fully focus on earnings, then goodwill write-offs might have an impact on their forecast determination. Prior research provides mixed evidence on whether future earnings forecasts are affected by goodwill impairment losses (Li et al., 2011; Jarva, 2014). This examination is related to Li et al. (2011) and similar to that research, I hypothesize that financial analysts revise the firms’ future earnings forecast downward following an impairment on goodwill. For determining this, I drew a sample of 175 U.S. firm year observations between 2002 and 2006. These five years were chosen because SFAS 142 was introduced in June 2001 and the financial crisis erupted in 2007 (Macdonald, 2012), which I consider to have a moderating effect on the relationship between goodwill impairment losses and the revision of future earnings forecasts. Consistent with the findings of Jarva (2014), I was unable to find significant evidence that goodwill impairment losses are negatively related to future earnings forecast revisions in the period 2002-2006.

Secondly, I investigate whether financial analysts react more strongly to a goodwill impairment loss when a manager has incentives to do big bath accounting than when he does not have these incentives. This is measured by a new CEO entering the firm since prior research has shown that new CEOs have incentives to engage in big bath accounting (Bornemann et al., 2015;

(7)

2 Davidson et al., 2007; Geiger & North, 2006; Hazarika et al., 2012). This effect is investigated in the same years as the first hypothesis, that is, between 2002 and 2006. Using a sample of 102 observations, I again did not find any significant influence of big bath accounting incentives on the relation between goodwill impairment and future earnings forecast revisions.

At last, I investigate whether the future earnings forecast revision is moderated in the years of the financial crisis. This is important since prior research has shown that many companies had to perform impairment tests more than once a year and hence, recognize impairment losses on goodwill (Devalle & Rizzato, 2012). Subsequently, Camodeca et al. (2012) has shown that the level of disclosure of goodwill impairment write-offs decreased during the financial crisis leading to lower quality disclosures. Due to these findings, I expect the financial crisis to have an impact on the relation between goodwill impairment write-offs and the revision of future earnings forecasts by financial analysts. Based on Macdonald (2012), I defined the financial crisis as the period that started in 2007. Since no specific research has been conducted on when the financial crisis has ended, I can only focus on the determination by Macdonald (2012) and hence, draw a sample of the years 2007-2012. There are various views on a crisis, however literature shows that a crisis is a disturbance in economic activities (Friedman & Schwartz, 1963; Minsky, 1981; Reinhart & Rogoff, 2013; Schwartz, 1986). Also, Mishkin (1992) mentions that a crisis leads to more severe adverse selection and moral hazard problems. Due to this, one would expect that impairment losses are more severe and that this will influence the future earnings forecasts of financial analysts following an impairment on goodwill. For determining this, I took the sample of firms that did an impairment between 2002 and 2006 and examined whether they also did an impairment on goodwill between 2007 and 2012. Of the 175 impairment observations between 2002 and 2006, 72 impairment observations were found between 2007 and 2012. Similar to the first hypothesis, I was unable to find significant results between goodwill impairment losses and future earnings forecast revisions.

This study contributes to the existing literature in multiple ways. The main contribution of this paper is that I examine whether financial analysts consider goodwill impairment testing of value relevance for revising their future earnings forecast. Li et al. (2011) found significant evidence for this hypothesis, however they focused on both unexpected and expected goodwill impairment losses. This raises the question whether the results would still hold by solely focusing on expected impairment losses, provided in annual reports of the firms. Additionally, I

(8)

3 investigate whether this relation is moderated when firms have incentives to engage in big bath accounting. The influence of big bath accounting has already been investigated by AbuGhazaleh et al. (2012) but they focused on stock prices and not on future earnings forecasts. Due to their findings that impairments are reflected in stock prices, I examine whether this is also reflected in the future earnings forecasts of financial analysts. At last, I examine whether the relation between goodwill impairment losses and future earnings forecasts is stronger in the years of the financial crisis. Mishkin (1992) argues that a financial crisis leads to increased interest rates and stock market declines. Furthermore, Camodeca et al. (2012) showed that goodwill impairment disclosures are lower during the financial crisis. This is not in line with the findings of Devalle & Rizzato (2012) who state that goodwill impairment write-offs are higher during the crisis. Hence, one would expect more impairment disclosures. This raises the question whether the increase in impairment losses during the financial crisis has an impact on the relation between goodwill impairment losses and future earnings forecasts. The contribution of this paper will be further explained in section 2.6 Summary and contribution.

The remainder of this paper is organized as follows. The next section describes the concepts, the applicable theories and a literature review of prior research about the topic. Section 3 describes the research method including the hypotheses, empirical models, data collection and sample determination. Section 4 elaborates on the results of the empirical analyses. At last, section 5 concludes with a summary, conclusion and limitations of this investigation.

2 Concepts, theories and literature review

This chapter will provide an overview of the concepts, theories and the available literature on the accounting of goodwill. At first, the concepts will be discussed. These concepts include the definition of goodwill, the new regulation on the accounting for goodwill and the determination of the period of the financial crisis. These concepts are important to increase understanding of this investigation. Secondly, the applicable theories are discussed which include the agency theory and the efficient market hypothesis theory. Thirdly, prior research will be discussed related to this topic. At last, a summary of this chapter will be provided including the contribution of the research.

(9)

4 2.1 Concepts

2.1.1 Goodwill

Goodwill is an unidentifiable intangible asset. Within US GAAP, there exist two kinds of these assets. The first one is purchased goodwill. This arises on the acquisition of an existing business concern. Lander & Reinstein (2003) argue that entities should only record goodwill when another business is purchased, since recognized goodwill is the difference between the acquisition price of the buyer and the fair value of assets less liabilities of the target. This acquisition price is determined by factors such as the individual conception of the buyer, its financial position, and the circumstances of the acquisition. An enterprise is willing to pay goodwill because synergy-effects arise due to the purchase of another enterprise. The acquirer then expects to accrue potential economic benefits of the acquisition which are not identifiable on their own. However, it is also possible that only a part of the business is purchased. In this case, only the acquired part is recognized on the balance sheet (Wines & Ferguson, 1993).

The second kind of goodwill is internally generated goodwill. This appears due to the operational activities of the enterprise itself. The difference with purchased goodwill is that, in essence, this may not be activated on the balance sheet under US GAAP. Lander & Reinstein (2003) also accentuate internally generated goodwill. They confirm that this amount is not recognized since there does not exist an objective valuation method for this. Subsequently, internally generated goodwill can easily be used as an earnings management instrument due to the subjectivity of the recognition and measurement (International Accounting Standards Board, 2007).

The emphasis of this study will be on goodwill write-offs. Furthermore, it is important for this research to mention that the accounting for goodwill has been adjusted. The previous regulation APB 17 was replaced by SFAS 142. This has been done because the new regulation provides more timely and better information to investors (Lander & Reinstein, 2003). The following section will discuss both accounting regimes.

2.1.2 From APB 17 towards SFAS 142

Analysts, company’s managements, and other users of financial information mentioned that intangible assets are fundamental economic resources for many entities. Furthermore, their value

(10)

5 are an increasing proportion of the assets acquired in many transactions. For this reason, better information about these assets was needed in order to improve decision making. APB 17 argues that the goodwill amount is amortized over its useful life of a maximum of 40 years. However, financial statement users indicated that, for their decision making, they did not take goodwill amortization into account (Financial Accounting Standards Board, 2001b). Jennings et al. (2001) investigated the relevance of amortizations by looking at the usefulness of earnings after an amortization of goodwill. They state that if goodwill amortization was a useful (useless) expedient for determining declines in goodwill amounts, then the recent change might make earnings information less (more) useful to investors. To investigate this issue, the authors drew a large sample of publicly traded companies over the 1993-1998 period. The results indicate that earnings before the amortization of the goodwill amount explain significantly more of the deviation in stock prices than earnings after goodwill amortization. Overall, excluding goodwill amortization from the income statement under the new impairment regime will not reduce the usefulness of earnings. Rather, the authors mention that this new standard may eliminate noise in earnings as measured under the previous standard (Jennings et al., 2001).

Because goodwill is no longer amortized under SFAS 142, the reported amounts will not decrease at the same time and in the same manner as under the previous approach. Due to a fundamental change in accounting, there may be more volatility in reported income than under previous standards because impairment losses are likely to occur irregularly and in varying amounts. SFAS 142 also changes the way in which managers evaluate whether goodwill needs to be impaired. Firstly, SFAS 142 requires fair values rather than undiscounted cash flow as a threshold. Secondly, the required fair value cash flows under SFAS 142 relate to reporting units rather than the entire groups of assets. The adjustments in the evaluation of managers whether goodwill is impaired could cause firms to take write-offs when adopting SFAS 142.

The trigger and amount of any write-off due to SFAS 142 adoption may be moderated by reporting discretion. SFAS 142 provided the possibility for accounting discretion by requesting managers to make two exceptional accounting choices:

(a) Defining the reporting units and;

(11)

6 When determining the allocated amount to each unit, managers should assign goodwill to reporting units expected to benefit from synergies of the acquisition even if the acquired assets and liabilities are not appointed to those units. SFAS 142 also states that the chosen methodology to determine the amount of goodwill allocated to a reporting unit shall be reasonable, supportable and utilized in a consistent manner (Beatty & Weber, 2006; Financial Accounting Standards Board, 2001a).

SFAS 142 also issues adjustments for the presentation of goodwill. The aggregate amount of goodwill has to be presented as a separate line item in the income statement. This should be stated before the total income from operations unless a goodwill impairment is related to a discontinued operation. A goodwill impairment loss arising from a discontinued operation must be included in the results of discontinued operations (Financial Accounting Standards Board, 2001a).

Subsequently, SFAS states that testing for impairment on goodwill of a reporting unit shall be on an annual basis. This test is performed every year at the same time. The testing for impairment of goodwill is performed in two steps:

(a) The fair value of the reporting unit is compared with its carrying amount. In case that the fair value is higher than the carrying amount, goodwill is not impaired.

(b) The implied fair value of the reporting-unit is compared with its carrying amount. If the carrying amount is greater than the implied fair value of goodwill, an impairment loss is recognized in an amount equal to that excess.

However, it is also possible that goodwill of a reporting unit is tested for impairment between the annual tests. This happens when an event occurs or circumstances change that would more likely than not decrease the fair value of a reporting unit below its carrying value (Financial Accounting Standards Board, 2001a).

2.1.3 The financial crisis of 2007

The financial crisis is expected to impact the relation between an impairment loss of goodwill and the future earnings forecasts determined by financial analysts. It is important to determine when and how a financial crisis arises, because of the interpretation of the results. This is difficult to ascertain because there are a lot of different views on a financial crisis. Mishkin

(12)

7 (1992) mentions that these different views are divided in two directions. The first one is the monetary view, which is supported by Friedman and Schwartz (1963). They have provided a relation between a financial crisis and banking panics. They state that banking panics caused large declines in the money multiplier, and therefore in the supply of money. Also, they mention that these declines would not have occurred in case of a better functioning central bank

(Friedman & Schwartz, 1963).

The second view is a much more general one. This view is outlined by Kindleberger (1978) and Minsky (1981). They provide a much broader definition of what creates a real

financial crisis, compared to monetarists. In their articles, both authors state that a financial crisis either involves severe declines in prices of assets, bankruptcies of major financial and non-financial companies, or deflations. Also, every failure consists out of initial displacements or economic shocks, and human error. All these failures will lead to a decrease in economic efficiency and effectivity (Kindleberger, 1978; Minsky, 1981). However, Kindleberger (1978) -Minsky’s (1981) view does not provide an applicable overview of the characteristics of a financial crisis, since it is considered to be too broad to do so. On the other hand, the monetarist view of Friedman and Schwartz (1963) is extremely narrow because it only focuses on bank panics and the effect on the supply of money. These contradictory views are offset in the

research of Mishkin (1992). He defines a financial crisis as: ‘a disruption to financial markets in which adverse selection and moral hazard problems become much worse, so that financial markets are unable to efficiently channel funds to those who have the most productive

investment opportunities’. For defining this, he provides an asymmetric framework to understand a financial crisis. Information asymmetry is explained by the agency theory of Jensen and

Meckling (1976). They mention that this problem arises when the desires or goals of one party differ from the other party and both parties have different information at their disposal (Jensen & Meckling, 1976). For this study, I will focus on the financial crisis which started in 2007

according to Macdonald (2012).

The genesis of this financial crisis can be divided into two related processes. The first process was a boom in the housing market in the U.S. This boom was caused by a series of government interventions which were assigned to provide aid to the less wealthy people. The consequence was that this led to an increase in the demand for houses, and thus to an increase in the housing prices. The second process was the growth of securitization. Securitization can be

(13)

8 best described as the practice of pooling different kinds of debt, such as mortgages, and sell these at a higher price to make a profit. The increase in this practice led to a bigger impact of

government interventions to acquire more ownership (Macdonald, 2012). Schwartz (1986) states that government intervention in a ‘pseudo-financial crisis’ can be harmful since it leads to a decline in economic efficiency and an increase in money growth, which stimulates inflation. Next to that, securitization led to an explosive growth of mortgages. These two processes had the following effects: at first, a lot of people were not able to pay their mortgages anymore. Next to that, a large number of investors saw that the ‘little-risk securities’ actually contained a very unpredictable value (Macdonald, 2012).

2.2 Applicable theories

2.2.1 Agency Theory

Watts (as cited in Jarva, 2014) mentions that agency theory predicts that managers will use discretion offered by SFAS 142 regulation in cases where they have incentives to do so. Agency theory is directed at the broad agency relationship, in which one party (the principal) delegates work to another (the agent), who performs that work (Jensen & Meckling, 1976; Ross, 1973). The problem that occurs within this relationship is when cooperating parties have different goals

and divisions of labor.

Furthermore, this theory is concerned with resolving two problems that can occur in agency relationships. The first problem that arises is when the desires or goals of the principal and agent are in conflict and it is difficult or expensive for the principal to verify what the agent is actually doing. The principal cannot verify that the agent has behaved appropriately. The second problem that arises is the problem of risk sharing when the principal and agent have different attitudes toward risk (Jensen & Meckling, 1976). Eisenhardt (1989) states: “The problem here is that the principal and the agent may prefer different actions because of different risk preferences”. In case of the goodwill impairment regime, it is likely that managers have incentives to delay an impairment of goodwill since this indicates a decrease in future economic benefits or overpayment of the acquired intangible asset. Prior studies (Amihud & Lev, 1981; Morck et al., 1990; Olante, 2013) have provided evidence that overpayment of the acquired target is related to immediate goodwill write-offs. These studies furthermore argue that overpayment may result from agency conflicts in mergers and tender offers. This is because

(14)

9 managers may act in their own self-interest at the expense of shareholders in order to remain embedded or to decrease the risk associated with their managerial human capital (Amihud & Lev, 1981; Morck et al., 1990).

Ramanna & Watts (2012) state that agency theory predicts that managers will, on average, use the unverifiable discretion in SFAS 142 consistent with private incentives. In their research, they investigate managers’ implementation of the SFAS 142 goodwill impairment test in a sample of firms with market indications of goodwill impairment. They have examined whether goodwill impairment is associated with proxies for managers’ private information on positive future cash flows and with agency-based incentives, such as management’s interest in increasing their personal compensation and in shielding their reputation from the implications of a goodwill write-off. They find significant evidence that both compensation and reputation are negative predictors of goodwill impairment (Ramanna & Watts, 2012).

2.2.2 Efficient Market Hypothesis (EMH)

The allocation of ownership of the provided stock within the economy is one of the primary goals of the capital market. In order to achieve this, prices should provide accurate signals for the allocation of resources. This is important in order to provide investors with the opportunity to make production-investment decisions. However, investors will choose among the securities that fully reflect the performed activities of the firm, based on the adjustments in the pricing of these securities. This means that security prices should ‘fully reflect’ all available information towards the market. A market in which prices always ‘fully reflect’ available information is called ‘efficient’ (Malkiel & Fama, 1970). Financial analysts consider all available information in the market in order to provide a future earnings forecast. Since earnings per share is one of the most commonly used variables for determining this, it is of high importance that all information is fully reflected in the market (Brandon & Jarrett, 1979).

However, the definitional statement that an efficient market “fully reflect” available information is so broad that it has no statistically testable hypothesis. As Malkiel & Fama (1970) mention, most of the available work is based only on the assumption that the conditions of market equilibrium can be stated in terms of expected returns. This assumption is the basis of the expected return or “fair game” efficient market models. For determining their efficient market model, the authors use three relevant information subsets, concerning the adjustment of security

(15)

10 prices. Firstly, (1) Weak form tests are used in which the information set is based on historical prices. Secondly, they use (2) Semi-strong form tests in which prices are assumed to fully reflect all obviously, publicly held available information. At last, (3) Strong-form tests are assumed to fully reflect all available information. Since financial analysts can provide the best earnings forecasts in the strong-form tests, this is the most important form. Malkiel & Fama (1970) found evidence in favor of the efficient market model. They state that this model is a very useful approximation of reality (Malkiel & Fama, 1970). For this reason, this research builds on this theory since it provides a framework to investigate the availability of information to financial analysts. This is important since financial analysts adapt their future earnings forecast based on the available information in the market.

2.3 Literature review

This literature review elaborates on all the research that has been conducted related to the accounting for goodwill. At first, prior research will be discussed regarding the value-relevance of the amortization regime and the impairment regime. Subsequently, section 2.3.2. will go into detail on the research between goodwill accounting and the reaction of market participants.

2.3.1 Value relevance of APB 17 and SFAS 142

An early definition of goodwill was documented by Bourne in 1888, who noted it as ‘the benefit and advantage accruing to an existing business from the regard that its customers entertain towards it, and from the likelihood of their continued patronage and support’ (Bourne, 1888). Due to the introduction of SFAS 142 by the FASB, the accounting of goodwill changed fundamentally. Instead of amortizing the amount of goodwill over a maximum of 40 years, SFAS 142 requires annual impairment testing of goodwill. One of the arguments of the FASB for adjusting this method is the conviction that the method of amortization may lead to a situation of arbitrary accounting (Financial Accounting Standards Board, 2011).

However, this change in the accounting for goodwill after the initial recognition of the investment has been discussed among researchers and regulators for an extensive period of time. The question that raises is whether goodwill impairment tests better reflect the value than amortizing the investment over its expected useful life.

(16)

11 Calvert et al. (2014) argue that goodwill impairment testing is a complex area of financial reporting that requires careful judgment. They have conducted interviews with stakeholders from various backgrounds and geographies concerning the usefulness and complexity of goodwill impairment testing. The interviewees mention that goodwill impairment testing is relevant in determining how well a subsidiary has performed, however its relevance to the market is confirming rather than predicting. Additionally, they state that the level of impairment disclosures is exaggerated, however financial analysts would be in favor of more disclosures. Furthermore, analysts mention that they are annoyed by the inconsistency in impairment-related disclosures, which makes it difficult to compare companies and provide forecasts. One interviewee even states to be in favor of the straight-line amortization of goodwill because this would support the decision usefulness much better. The varied responses raises questions about the quality of the current regime (Calvert et al., 2014).

In addition, Massoud & Raiborn (2003) discuss the impact of SFAS 142 on financial statements and earnings quality. They provide their negativity by stating the opinion that financial analysts have ignored goodwill amortization in the previous regime and have been using unexpected earnings measures for a long time, implying that SFAS 142 is not likely to affect the decision models used by informed individuals. To amplify their negativity, they also mention that the impairment testing and related write-offs are non-cash items and hence, should not be highly influential in stock prices over the long run (Massoud & Raiborn, 2003). Ellis (2001), instead, did not identify considerable changes in the financial statement disclosure, and concluded that the impairment accounting method for goodwill does not provide investors with a

representative view of business combinations.

Chalmers et al. (2011) investigated whether the impairment regime better represents the underlying economic value of goodwill than systematic amortization. For doing this, they made a comparison between the association of goodwill accounting charges against income and the economic investment opportunities of firms in both the amortization regime and the impairment regime, based on IFRS in Australia. They have found results that the comparison is stronger in the IFRS regime than in the AGAAP regime (Chalmers et al., 2011). This implies that impairment write-offs better reflect the underlying economic characteristics of goodwill than systematic amortization. However, this research is based on the impairment regulation of IFRS, which differs from U.S. GAAP to some extent.

(17)

12 Subsequently, Godfrey & Koh (2009) drew a sample of listed firms in the U.S. They build on Bradbury et al. (2003) who found that in New Zealand, where the impairment regime does not apply, firms with high investment opportunities amortized goodwill less rapidly than firms with less investment opportunities. Hence, Godfrey & Koh (2009) investigate whether these findings apply to the new SFAS 142 regime since impairment testing reflects a firm’s investment opportunities. Overall, their findings support the argument that the introduction of SFAS 142 provides information relevant to the users of financial statements and therefore, are consistent with Bradbury et al. (2003).

As mentioned before, a lot of researchers provide their skepticism about the current accounting method. They state that the impairment regime offers managers discretion in performing a goodwill write-off. Ramanna (2008) verifies this. In his research, he builds on Holthausen and Watts (2001), which argue that when fair values are unverifiable, the likelihood of opportunistic disclosure increases. Ramanna (2008) investigates this by using SFAS 142, since this regulation also uses unverifiable fair value estimates. He finds that the discretion potential increases for firms supporting SFAS 142. This discretion can be used opportunistically according to the agency theory. Concluding, these results comply with unverifiable fair values in SFAS 142 (Ramanna, 2008).

Additionally, AbuGhazaleh et al. (2012) also conducted a study related to the discretion afforded by SFAS 142. They found empirical evidence that managers are exercising this discretion in the reporting of goodwill impairment losses following the adoption of IFRS 3 – Business Combinations in the UK. Their results also reveal that goodwill impairments are strongly associated with effective governance mechanisms, suggesting that managers have incentives to be using the discretion allowed by IFRS 3 efficiently to transmit their private information and expectations about the underlying performance of the firm rather than acting opportunistically in the write-off year (Abughazaleh et al., 2012).

Also, fair value accounting is often criticized due to the discretion since management bias arises which may result in incorrect fair value estimates of both earnings and equity (Beatty & Weber, 2006; Watts, 2003a; Watts, 2003b). Beatty & Weber (2006) examined this by looking at the factors that determine the write-off of goodwill compared to the amount that is literally written-off. They have focused on the trade-off between recording certain current goodwill impairment charges below the line and uncertain future impairment charges included in income

(18)

13 from continuing operations. The authors used a censored regression to examine the percentage of goodwill written off when SFAS 142 is adopted. The results of this analysis indicate that, after controlling for the effects of firms’ economic conditions and changes in those economic conditions, firms are less likely to take a write-off when they have less slack in their net worth covenant and the covenant is affected by accounting changes (Beatty & Weber, 2006). This is confirmed by Riedl (2004) who finds evidence of lower reporting quality for asset write-offs in the post-SFAS 121 period. He also shows that write-offs after the introduction of SFAS 121 are associated with accelerated or delayed expense recognition, suggesting more discretion for managers in financial reporting (Riedl, 2004).

On the contrary, Lee (2011) performed research on the effectiveness of SFAS 142 by looking at the ability to predict future cash flows related to a goodwill write-off. He used firms that have different incentives to exercise the accounting discretion following SFAS 142 by looking at discretionary accruals and the firms meeting (or beating) the earnings forecasts. The results of his tests do not support the hypothesis that discretion is used opportunistically and hence, suggest that the effect of opportunistic reporting is neutralized by the informative value of the discretion (Lee, 2011).

As the above literature shows, there are mixed views on the new regime of accounting for goodwill. It is important to discuss how the firms’ stockholders view the goodwill write-off regime because the current study focuses on financial analysts. However, since this has not been investigated thoroughly, the next paragraph will elaborate on the relation between goodwill write-offs and market participants in general.

2.3.2 Relation between goodwill impairment and market participants

AbuGhazaleh et al. (2012) conducted a research in which they found evidence for the accomplishment of the IASB for the increase in relevance, reliability and comparability due to the goodwill impairment. In their research, the authors use a sample of 528 firm-year observations from the UK top 500 listed firms in the years 2005 and 2006 and found significant results of a negative association between impairment losses and market value. This suggests that these impairments are incorporated in their firm valuation assessments (Abughazaleh et al., 2012). However, as the authors suggest, a lack in their investigation might be that they drew a sample of only two years.

(19)

14 Li et al. (2011) did a similar study in the U.S. over more years. They did an extensive research over the years 1996-2006, thus covering both SFAS 121 and SFAS 142 regulatory regimes. Using a total sample of 1,584 over the entire decade, the authors compared the reaction of market participants by testing whether the public announcement of a goodwill write-off provided new information to market participants before, during and after the implementation of SFAS 142. Similar to AbuGhazaleh et al. (2012), Li et al. (2011) found that investors and financial analysts adjust their earnings expectations downward following the goodwill impairment loss announcement. Significantly, this impact is lower in the post-SFAS 142 period compared to the pre-SFAS 142 period and transition period. An explanation for this are the smaller post-period losses and less repeated occurrence during the post-SFAS 142 period (Li et al., 2011).

Another study of the relation between goodwill impairment and the market reaction is done by Hirschey & Richardson (2003). They find that public announcements of goodwill lead to a negative reaction of 2 to 3 percent of the stock price. They also find a decrease of 40 percent in the one-year before the announcement period and 11 percent in the one-year after the announcement period. These results suggest that investors consider and underreact to impairment write-offs even before the announcement date. A related research was conducted by Li et al. (2005). In 2005, they investigated the goodwill impairment sample over a one-year period after the implementation of SFAS 142. For doing this, they looked at the three-day stock return around the write-off announcement period (day -1 till +1) and also found a significant negative reaction by the market surrounding the impairment announcement (Li et al., 2005).

Liberatore & Mazzi (2010) developed a hypothesis contrary to prior research (Hirschey & Richardson, 2003; Li et al., 2011). The authors expected no significant reaction from financial markets to a goodwill write-off in the impairment regime. They present their skepticism by stating that the impairment regime presents an economic estimate without financial signification. For measuring this hypothesis, the authors compared public goodwill write-off announcements in relation to the volatility in the stock market prices. Contrary to their expectations, their results indicate a relation between the write-off announcement and the behavior of financial markets (Liberatore & Mazzi, 2010).

Also, Segal (2003) saw that goodwill impairment is positively associated with overpayment of the asset, but negatively associated with stock returns and financial performance.

(20)

15 However, he did not find any evidence that the reaction of the market differs following the introduction of SFAS 142, implying that the introduction of SFAS 142 did not accentuate stock prices. Jarva (2014) used a sample of 280 material SFAS 142 goodwill write-offs from the period from 2002 to 2005 to conduct a related research. He builds on Bradshaw et al. (2001), who have shown that analysts do not consider the declines in future earnings associated with high accruals in providing their earnings forecast. Jarva (2014) examined whether market participants fully incorporate the information of a goodwill write-off into stock prices. Also, he examined whether the analysts’ future earnings forecast is adjusted when goodwill is written-off. His results show that the future earnings forecasts determined by financial analysts is not lower for write-off firms. Hence, he fails to find evidence that both investors and analysts incorporate write-off information in their stock price forecast (Jarva, 2014).

In summary, the provided evidence shows mixed results regarding the reaction of market participants and the accounting for goodwill. For that reason, it is important to investigate whether SFAS 142 is of value relevance for financial analysts. This will be discussed further in the hypotheses development.

2.4 Summary and contribution

The FASB changed the accounting for goodwill in 2001. They agreed goodwill accounting to be tested for impairment on an annual basis instead of amortizing its value over its useful life. This change meant that companies need to perform these annual tests for themselves by ascertaining if the recoverable amount is higher than the carrying amount. Goodwill is impaired only when the carrying amount is higher than the recoverable amount.

After the abolishment of APB 17 and the implementation of SFAS 142, there has been a lot of investigation in comparing both methods and the quality of earnings. The FASB argues that the amortization method leads to arbitrary accounting which implies a decrease in quality information for financial analysts, entailing that the impairment regime will provide more useful information for financial analysts in providing future earnings forecasts (Financial Accounting Standards Board, 2011). However, this opinion is not fully supported by academic evidence, since there is no irrefutable evidence that the new statement indeed improves the accounting quality of goodwill.

Since it was the goal of the FASB to increase the quality of accounting information for financial statement users by introduction SFAS 142 (Financial Accounting Standards Board,

(21)

16 2001b), this study investigates whether financial analysts revise their future earnings forecasts downwards following the new regulation on the accounting of goodwill. This is investigated by comparing the publicly available goodwill impairment losses with the volatility of the future earnings forecasts. Building on the research of Li et al. (2011) who provide significant evidence for this proposition, this research distinguishes itself by examining two moderator variables which are expected to influence this relationship. The main proposition will look at the effect of goodwill impairment losses after the implementation of SFAS 142. Based on discussed prior literature, I believe the relationship to be stronger in the post SFAS 142 period. The first moderator considers the effect of big bath accounting incentives. This is because prior research has shown that big bath accounting is negatively associated with goodwill impairment losses (AbuGhazaleh et al., 2012). Secondly, the years of the financial crisis are expected to moderate the association between goodwill write-off and earnings forecasts. For that reason, this study investigates whether financial analysts amplify their future earnings forecast revision in the years of the financial crisis. To confirm these propositions, a dataset of U.S. listed firms is used over the period between 2002-2012 which includes the period after the SFAS 142 implementation as well as the period of the financial crisis.

This paper contributes to the literature in several ways. At first, it contributes to the timing of goodwill impairments and the reaction of market participants on these impairments, focusing on financial analysts. This study builds on the findings of Li et al. (2011), who also examine the reaction of market participants to an announcement of a goodwill impairment loss. Although they focused on both expected and unexpected goodwill impairment losses. Secondly, the influence of big bath accounting incentives will be examined which has not been done before in prior literature. This influence will be measured by a new CEO entering the firm since companies that have experienced a recent CEO change are associated with higher reported amounts of goodwill impairment losses (Lapointe-Antunes et al., 2008; Masters-Stout et al., 2008; Zang, 2008). Since the impairment-only approach of SFAS 142 offers discretion to manage earnings, there is a likelihood that managers who have incentives to engage in big bath accounting take advantage of this. Thirdly, the effect of the relation in the years of the financial crisis will be considered, since goodwill impairment testing will be done more frequently in these years, leading to an increase in write-offs (Devalle & Rizzato, 2012). If impairments are more severe during the financial crisis, this is expected to have a significant influence on the

(22)

17 future earnings forecasts. Due to its differentiation from prior research, this study is one of the first to combine analysts’ future earnings forecasts with goodwill impairments.

3 Research method

This chapter will provide the method that has been used in order to analyze the research question. Section 3.1 elaborates on the hypotheses development, which is followed by the moderation effects on the first hypothesis in section 3.2. More specifically, this research focusses on two moderating variables which are discussed in section 3.2.1 and 3.2.2. After that, the three research models are discussed in sections 3.3.1, 3.3.2, and 3.3.3. At last, this chapter will conclude with the data and sample determination in section 3.4.

3.1 Hypothesis development

The main statement that is being investigated in this study is whether financial analysts revise their future earnings forecasts following a loss on the impairment of goodwill. As discussed in the literature review, authors have shown that goodwill impairment losses are considered as value relevant (Godfrey & Koh, 2009; Bradbury et al., 2003). However, there are mixed views on whether a goodwill write-off announcement will change the perceptions due to these announcements (Li et al., 2011; Jarva, 2014).

Prior studies have examined the relation between stock price and goodwill write-offs (Liberatore & Mazzi, 2010; AbuGhazaleh et al., 2012; Hirschey & Richardson, 2003; Segal, 2003), however not much research has been done about the relation between goodwill impairment and future earnings forecasts. As mentioned before, Jarva (2014) focuses on the accuracy of financial analysts’ forecasts. He shows that this accuracy is not worse for firms that do an impairment on goodwill. More specifically, he fails to finds evidence that analysts fixate on earnings with regard to goodwill write-offs under the SFAS 142 regime. On the other hand, Li et al. (2011) examined the relationships between goodwill impairment and the reaction of the stock price as well as the reaction of financial analysts. They compared the impairment regime with the amortization regime considering goodwill. The authors found evidence that financial analysts deteriorate their expectations based on the performance of a firm by means of a decline in the earnings forecast. Building on this evidence, the following hypothesis is stated as:

(23)

18

H1: Goodwill impairment losses are negatively related to financial analysts’ future earnings forecast

3.2 Moderation effects

Since Li et al. (2011) and Jarva (2014) have already investigated the relationship between goodwill write-offs and future earnings forecasts, this study will further examine the effect of several characteristics which are expected to moderate the relationship between goodwill impairment losses and financial analysts’ future earnings forecasts. At first, the effect of big bath accounting incentives is investigated. Research has shown that managers have incentives to engage in big bath accounting when a new CEO enters the firm (Bornemann et al., 2015; Masters-Stout, 2008; Davidson et al., 2007). For that reason, a change in CEO is used as a proxy to measure incentives for big bath accounting. Secondly, the years of the financial crisis are expected to influence the relation between goodwill impairment losses and future earnings forecasts since impairment testing is done more frequently during these years, leading to an increase in write-offs (Devalle & Rizzato, 2012). The sections 3.2.1 and 3.2.2 will elaborate on these moderation effects.

3.2.1 Accounting conservatism

Big bath accounting is associated with accounting conservatism. Overboom & Vergoossen (1997)describe big bath accounting as: ‘companies that suffer a bad year are more conservative towards their costs and losses by taking them into account in a bad year to report better results in the subsequent year’. Feltham & Ohlson (1995) state that accounting is conservative (unbiased) if the value of a firm is expected to be higher than the firm’s book value of equity in the longer run. They identify two different types of accounting conservatism:

1. Conservatism which arises due to accounting rules and;

2. Conservatism which is associated with investments in positive net present value projects. AbuGhazaleh et al. (2012) build on the first type of accounting conservatism by examining whether managers are opportunistically using the discretion offered in IFRS 3 when testing for

(24)

19 goodwill impairment. This research contains a sample of 528 UK-listed firms for the years 2005 and 2006. After controlling for economic factors, their main results indicate that managers are exercising discretion in the reporting of goodwill impairment losses following the adoption of IFRS 3. Furthermore, their results suggest that managers are more likely to exercise discretion in their accounting to convey their private information and expectation about the underlying performance of the firm rather than acting opportunistically. Collectively, these results indicate that managers are more likely responding to changes in economic circumstances and real declines in the value of the firm in recording goodwill impairments. Effective governance mechanisms, including big bath accounting, intent to restrict the ability of managers to report goodwill impairments that differ from predicted economic losses, resulting in the recognition of more timely impairments that better reflect the firm’s underlying economics (Abughazaleh et al., 2012; Verriest & Gaeremynck, 2009).

Goodwill impairments can be taken immediately when a firm decides to do so. Managers have incentives to accelerate goodwill write-offs in periods in which the write-off earnings before the initial write-off are already below expected earnings. Hence, managers may undertake a ‘big bath’ when this occurs in order to boost future earnings and provide a signal that ‘bad times’ are behind them and better times will follow. This means that they will continue with a ‘clean sheet’ in the subsequent period (Zucca & Campbell, 1992). This is confirmed by Kirschenheiter & Melumad (2012) who provide a model in which big bath accounting is part of a balancing reporting strategy. They argue that if the news is bad, managers have incentives to under-report earnings by taking all losses, hence preferring to take a ‘big bath’ in the current period in order to decrease costs for the following period. On the other hand, if the news is good, managers prefer to delay the recognition of a goodwill write-off. In both cases, this maximizes the value of the firm (Kirschenheiter & Melumad, 2002).

The evidence of the discussed literature suggests that the recognition of goodwill impairment is accelerated or delayed based on the incentives of managers to maximize firm value. For that reason, the prior literature provides evidence that managers are more likely to use the discretion offered in the new goodwill impairment regime (Beatty & Weber, 2006; Camodeca et al., 2012; Ramanna & Watts, 2012). In case that big bath accounting is used, hence accelerating goodwill impairments, the revision of the future earnings forecast of financial analysts is expected to magnify. Consequently, the following hypothesis will test whether the

(25)

20 relation between goodwill impairment and future earnings forecast is magnified by big bath accounting:

H2: The negative relationship between goodwill impairment losses and future earnings forecast is stronger when managers have incentives to do ‘big bath’ accounting.

3.2.2 Financial crisis

The financial crisis has led to an increase in the number of companies reporting write-offs and has led investors to focus more attention on the disclosure of this goodwill impairment (Devalle & Rizzato, 2012). This opinion is shared by Laux & Leuz (2010) who found that companies should recognize bigger impairment losses in times of economic downturn. This is because SFAS 142 mentions that goodwill accounting should be based on fair-values. If an acquisition decreases in value during the financial crisis because of operating results, the firm has to write-off significantly higher amounts than in case of a stable economy. More specifically, it is possible that the estimated future profits are disappointing and that the goodwill amount was too high. This impairment loss on goodwill could be high and could have a serious impact on the earnings and book value of assets (Alciatore et al., 1998).

As discussed earlier, Beatty & Weber (2006) have shown that discretion offered by SFAS 142 leads to a decrease in accounting quality. Camodeca et al. (2013) agree with Beatty & Weber (2006). They investigated and compared the level of disclosure of goodwill impairment tests during the major listed companies situated in the UK by reviewing 85 annual reports. These are ranked by market capitalization and total assets which are listed on the stock exchange. Their evidence shows a lack of goodwill disclosure on the key assumptions of the estimation model required by IAS 36, especially in the years of the financial crisis (2009-2011). Furthermore, their evidence has shown that goodwill impairment losses were very volatile from 2007 to 2011, reporting the highest loss in 2007 (Camodeca et al., 2013). Since firms are less likely to disclose goodwill impairment amounts, they are delaying the recognition of bad news, which is also discussed by Vichitsarawong et al. (2010). In their research, they investigate conservative accounting and timeliness of earnings of firms in Hong Kong, Singapore and Thailand during the period of the Asian financial crisis. They find significant evidence that conservatism and timeliness of earnings are low during the crisis period. Ball et al. (2003) found that loss

(26)

21 recognition is less timely in East Asian countries. This is due to institutional incentives, such as the influence of family or insider networks and political power. Furthermore, a financial crisis leads to lower prices of assets, increase in bankruptcies and deflations. (Kindleberger, 1978; Minsky, 1981). Mishkin (1992) elaborates on this by arguing that a financial crisis also leads to increases in interest rates, stock market declines, increases in uncertainty and bank panics. This can lead to an increase in adverse selection and information asymmetry, which results in more volatility of the economic environment, and thus in inefficiency of the financial market. If the expectations of future income are lower, then the probability of buying shares also decreases. Additionally, if the interest rate increases, the discounted present value of future income decreases, making it less attractive to buy shares (Mishkin, 1992). Due to these results, I expect a decrease in value of firms, hence an increase in goodwill impairment testing and more impairment losses being reported during the financial crisis. Consequently, I expect financial analysts to accentuate their future earnings forecast revision during these years. Hence, the following hypothesis will be examined:

H3: The negative relationship between goodwill impairments and financial analysts’ future earnings forecasts is stronger during the financial crisis.

3.3 Methodology

This section will provide an explanation on all three models that have been determined for the analysis of the hypotheses. It furthermore argues why the chosen variables are relevant and how they are determined.

3.3.1 The effect of goodwill impairment loss on future earnings forecasts

To explore the nature of information transmitted by the impairment loss which is expected to impact the analysts’ earnings forecasts, this study will use a model which is derived from Li et al. (2011). The revision of the future earnings forecast is considered to be the dependent variable, measured by the expected earnings per share of the firm. Goodwill impairment write-off will be used as an independent variable to determine the effect on future earnings forecasts. Li et al. (2011) focus in their research on the unexpected part of goodwill impairment. However, this investigation does only take expected impairment loss into account due to time constraints and

(27)

22 access limitations to public press releases. For that reason, the model used in this investigation deviates from Li et al.’s (2011). Also, this investigation takes different control variables into account. The determined model is as follows, examining the period 2002-2006:

In which the variables are:

REV Revision of the future earnings forecast of firm i as measured by the expected earnings per share of financial analysts, one quarter subsequent on the impairment loss of firm i;

IMPSHA Impairment loss of firm i at period t scaled by total shares outstanding at the end of that year;

ROA The average return on assets of firm i at period t calculated as income before extraordinary items divided by beginning total assets in the write-off year of firm i.

FSIZE The size of firm i at period t calculated as the log of market value of common equity1, determined by Beatty & Weber (2006).

LEV Equals long-term debt divided by long-term debt plus market value of common equity at the beginning of the announcement year for firm i at period t (Li et al., 2011).

The control variables that have been used in the model are derived from Beatty & Weber (2006) as economic determinants of impairment loss which are used by Li et al. (2011). As discussed in the literature review, Beatty & Weber (2006) investigate the likeliness of managers to take an impairment loss. I took out the unexpected earnings variables since I only focus on expected goodwill impairment losses and used return on assets, firm size and leverage.

3.3.2 The effect of incentives for big bath accounting on the relationship between goodwill impairment loss and future earnings forecast revisions.

Bornemann et al. (2015) build on existing literature which state that CEOs in non-financial firms tend to take an ‘earnings bath’. This is done by reducing the first year’s profit of new employment through discretionary expenses and save income for their subsequent accounting periods. Godfrey et al. (2003) confirm this by arguing that new managers have incentives to both

1_{The market value of common equity is calculated as the common shares outstanding multiplied by the closing}

(28)

23 manage earnings and manipulate the impressions by taking a big bath. The authors mention that a new CEO wishing to take a ‘big bath’ in the years of change and hence, managing earnings upward in the following year, is consistent with the desire to maximize value of his or her human capital (Godfrey et al., 2003). Davidson et al. (2007) did a related research in which they hypothesized that CEOs will be motivated to manage earnings prior to a turnover decision. The authors argue that this arises due to the horizon problem for CEOs nearing retirement age and for CEOs whose profit-based bonus is a major part of their total compensation. Furthermore, the authors find that firms in which CEOs are nearing retirement age have large discretionary accruals one year before the takeover (Davidson et al., 2007).

Based on this prior research, big bath accounting is measured using a CEO change within one year of the impairment period. This is done by the creation of a dummy for this variable. The empirical model for this hypothesis is as follows, examining the period 2002-2006:

In which the variable is:

CEOCHANGE Equals ‘1’ if firm i has experienced a change in executive one year before the

impairment loss was recognized and ‘0’ otherwise.

3.3.3 The effect of the financial crisis on the relationship between goodwill impairment and future earnings forecast revisions.

At last, to measure whether the financial crisis was of any influence on the relationship between goodwill impairment and future earnings forecast revisions, the same model is used as for the first relation, however in a different time setting. Hence, the model will be exactly the same, although I now examine the period between 2007 and 2012:

3.4 Data and sample

The sample of this study consists of U.S. firms. The required data of the sample exists only of financial information of these firms. This financial information was gathered from the WRDS

(29)

24 database. Within WRDS, I used I/B/E/S to find all firms in the U.S. that made a revision in their future earnings forecasts, as measured by expected earnings per share (EPS)2 in the period 2002-2006. This period was chosen since SFAS 142 was introduced in 2001 and the financial crisis started in 2007 according to Macdonald (2012). Subsequently, I used Compustat to retrieve data about goodwill balances and goodwill impairment for the same period. This resulted in 788 firms that did an impairment on their goodwill balance and additionally, an EPS change was reported on these firms in the quarter subsequent to the impairment loss. Table 1 provides an overview of the sample selection.

Table 1: Sample determination

Firm-year observations

Expected Earnings Per Share (EPS) adjustments between 2002-2012 (annual) 63,400 (-) Observations related to financial industries (SIC: 6000-6700) 24,274 (-) Observations without goodwill on the balance sheet 20,264 (-) Observations on which no goodwill information is available 5,513 (-) Observations that did not do an impairment on goodwill 2 (-) Observations on which no goodwill impairment loss is available 12,559

Final sample WRDS3 788

(-) Observations with no total assets information available 178 (-) Observations with no closing price available 40

(-) Observations without long-term debt 301

Outliers (See Table 2 for specification on outliers) 22

Final sample for analysis 247

Observations related to the influence of impairment loss on future earnings forecast 175 Observations related to big bath accounting 102 Observations related to the influence of impairment loss on future earnings forecast during the financial crisis

72

2

This proxy is derived from Li et al. (2011) to investigate the forecast revision of financial analysts. EPS forecasts are widely recognized as inputs for planning, investing and decision models. Furthermore, EPS provides relevant information on the expectations of a firm’s future prospects and help anticipate to a firm’s changing internal and external environment (Brandon & Jarrett, 1979). Also, it is considered to be the expected earnings per share of a firm since in a given year, impaired goodwill is already an expense.

3_{The information below Final sample WRDS was needed to determine the control variables. Besides that, total}

(30)

25 Within the database of Compustat, I deleted observations that belong to the financial sectors. These firms include banks, life insurances, nonlife insurance and real estate investments. As is projected in the table above, the first draw consisted of 63,400 observations of firms on which an adjustment of EPS was recorded. This was then compared to the observations of goodwill on the balance sheet. 20,264 firms did not contain any goodwill balance and for 5,513, this information was not available in Compustat. After that, the remaining sample was divided into firms that did an impairment on their goodwill balance in the quarter prior to the EPS adjustment was recorded. Of that sample, only 2 firms did not do an impairment on goodwill and for 12,559 observations, there was no information available regarding an impairment loss on goodwill. For that reason, this resulted in a final sample of 247 observations.

Table 2 specifies the subdivision of the outliers to get to the final sample which is used for the empirical analysis. Observations are expected to be outliers once they deviate more than three standard deviations from the mean of the variable. This correction resulted in a loss of another 22 observations, as can be seen in the table below. The final sample consists of 175 firm observations which is used to investigate the relationship between future earnings forecast revisions and goodwill impairment loss after the implementation of SFAS 142. Furthermore, 102 observations are available to determine the influence of big bath accounting. Prior literature has shown that big bath accounting occurs within the first year that a new CEO enters the firm, since they have incentives to blame the ‘bad outcome’ on their predecessors, lower the performance benchmark, and save income for subsequent accounting periods (Bornemann et al., 2015; Davidson et al., 2007; Geiger & North, 2006; Hazarika et al., 2012). For that reason, I will measure big bath accounting by investigating firms in which a new CEO has entered within one year prior to the impairment loss versus firms which remained their CEO. This division leads to 28 observations in which a new CEO entered the firm one year prior to the impairment loss and 74 observations in which no new CEO entered the firm. This equals 102 observations. At last, the same 175 firms of the sample of WRDS were selected to investigate impairment losses in the years of the financial crisis. As determined previously, the financial crisis started in 2007. Since there is no precise indication when the financial crisis has ended in the U.S., I investigated the impairment losses only until 2012. This resulted in 72 observations that also did an impairment between 2007 and 2012.

(31)

26

Table 2: Descriptive statistics to determine the boundaries for outliers per variable.

Variable Mean Standard

Deviation Lower bound Upper bound Exclusion

Expected Earnings Per Share (EPS)

0.543442 2.234459 -12.67 7.14 7

Impairment scaled by total shares outstanding

0.1039452 0.084583 0.0001023 0.4532752 5

Firm Size4 9.4384449 1.000034 7.271075 11.67782 0

Return on assets (ROA) -0.1245798 0.1482349 -1.0335864 0.2459454 5

Leverage5 0.0812933 0.1087763 0 0.6939299 5

22

4 Empirical results

Using the sample of 247 firm year observations as determined in section 3.4, this section will describe the results of the analyses performed by using this sample. I made use of STATA for performing all analyses. This section starts with the descriptive statistics in section 4.1 and continues with the correlations and multivariate results for all three hypotheses separately in section 4.2.

4.1 Descriptive statistics

The descriptive statistics for all the continuous variables in the period 2002-2006 are presented in Table 3. Since NEWCEO is a categorical dummy, this is not shown in this overview. In this table, the value relevance of goodwill impairment loss is shown for the sample after the implementation of SFAS 142 and before the financial crisis. The sample of this table includes firms that contain goodwill balances and subsequently did an impairment on goodwill, existing of 175 firm year observations between the period 2002 and 2006. Table 4 shows the descriptive statistics of the sample firms that also did an impairment during the financial crisis. Since not all firms also did an impairment or had impairment information available in these years, this resulted in less firm year observations.

Next to the 175 firm year observations that reported a goodwill impairment loss before

4_{Firm size is determined by the log of market value of common equity.}

Revisions in future earnings forecasts : changes in perceptions following a goodwill impairment loss : a study of U.S. companies