Goodwill impairment used for earnings management

(1)

Goodwill Impairment Used For Earnings Management

Student name: Navdeep Mander Student number: 6063829 Date of final version: 17-08-2015 Word count: 18,284

MSc Accountancy & Control, variant Accountancy Amsterdam Business School

Faculty of Economics and Business, University of Amsterdam Supervisor: dhr. drs. J.J.F. van Raak

(2)

2

Statement of Originality

This document is written by student Navdeep Mander 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)

3

Abstract

The purpose of this thesis is to examine whether goodwill impairments are used as a tool for earnings management when earnings are unexpectedly high or low. This thesis will particularly examine U.S. companies after the issuance of standard SFAS 142 and the time period examined is from 2005-2014. Based on a sample of 3059 firm-year observations I have found a positive relation between goodwill impairment and earnings management. I will specifically examine whether managers use big bath accounting or income smoothing through the use of goodwill impairment. This study contributes to existing literature from Van de Poel et al. (2008) and Francis et al. (1996) as there is evidence that earnings are managed when earnings are either unexpectedly low in the recession period from 2007-2009. I also contribute to the existing literature by looking at whether a new CEO with a maximum tenure of three year in a firm with unexpectedly low earnings uses bath accounting by impairing goodwill. Based on my findings, I document that (1) there is no positive relation found between new CEOs taking a bath when earnings are unexpectedly low, (2) overall firms with unexpectedly low earnings use bath accounting by impairing goodwill, (3) firms with unexpectedly high earnings smooth income by impairing goodwill, (4) in the recession there is a positive relation between unexpectedly low earnings and the impairment amount, (5) there is a negative relation between unexpectedly high earnings and impairment amount in the period from 2007-2009.

(4)

4

1 Introduction

There has been controversy on how to account for goodwill for decades (Bloom, 2009). The main approach used in the past has been the amortization of goodwill over a certain period. Due to the growing significance of intangible assets and goodwill this is justified. Internationally the two dominating accounting frameworks used are International Financial Reporting Standards (IFRS) and US Generally Accepted Accounting Principles (US GAAP) (Daske, 2006). However, this paper will mainly focus on a sample selection of U.S. companies which use accounting standards conform to the U.S. GAAP for the accounting of goodwill.

It is a controversial problem whether the accounting of goodwill impairment leads to more earnings management. After many years of discussions and consideration, the FASB decided that better information was needed regarding intangibles (Massoud and Raiborn, 2003). The major basis underlying SFAS 142 was that goodwill does not necessarily decrease in an orderly way. In addition the statement identified that externally purchased goodwill may indeed have an indefinite useful life and therefore should not be amortized over the arbitrary maximum period of 40 years. The FASB declares that the approach of goodwill impairment and the elimination of goodwill amortization has the ability to better assess the drop in value of the goodwill (Zang, 2008). In addition, the impairment of goodwill gives a better representation of the economic value of goodwill.

The factors used in the impairment test are dependent on many assumptions made by the manager. This is because it is their responsibility to calculate the impairment of goodwill. SFAS 142 requires managers to approximate the fair value of goodwill in order to determine the write off of goodwill (Ramanna and Watts, 2012). The goodwill is measured as the excess of price paid for a business acquisition over the fair value of the business (Jerman and Manzin, 2008).

The Statement of Financial Accounting Standards 142 addresses the accounting of goodwill and other intangible assets. This is a fundamental standard issued by the Financial Accounting Standards Board (FASB) which eliminated the amortization of goodwill and replaced it by the goodwill impairment test. The amortisation method was eliminated because it did not give a faithful representation of the true economic value of goodwill. IAS 22 which required amortization was criticized as the balance sheet amounts did not reflect the true economic value of the goodwill (Lhaopadchan, 2010). Under this standard goodwill was recognized as an asset and had to be amortized over its useful life on a straight line basis.

Authors, such as Beatty and Weber (2006) have suggested that the impairment of goodwill is subject to opportunistic behaviour by managers. This agency-principal problem can lead to a false portrayal of the company’s underlying economic value in the reports. For the impairment of assets, managers have discretion over the impairment criteria. The impairment process relies on

(6)

6 professional judgement and not on strict rules-based criteria. It is interesting to note that new opportunities arise for managers to engage in earnings management by allowing managers more flexibility in determining the amount of goodwill and the impairment charge (Massoud and Raiborn, 2003). This flexibility could lead to management manipulation.

There have been several reasons for managers to manipulate earnings such as preventing debt covenant violation and incentives of new managers to manipulate impairment charges. Managers may not impair goodwill because it could affect their bonuses for example. Standard setters have been trying to eliminate or at least reduce this flexibility for managers (Carmona and Trombetta, 2008). In addition, the impairment of goodwill relies on the professional judgement of managers and there are no strict rules based criteria for it (Carmona and Trombetta, 2008). According to some researchers such as Carlin and Finch (2010) and Zang (2008), the impairment of goodwill has led to increased discretion for managers which in turn has led to opportunistic behaviour. Carlin and Finch (2010) found that managers manipulate discount rates lower than the expected discount rates.

Previous literature shows that there are mixed findings on the impairment of goodwill. There are many factors that can have an impact on the decision to write off goodwill. Zang (2008) found that highly leveraged firms tend to report a lower impairment charge on goodwill. He found that after the adoption of SFAS 142, stock return is negatively associated with an unexpected initial impairment loss (IIL). There is support that highly leveraged firms manage earnings strategically by impairing goodwill. Furthermore, new managers have the incentive to lower earnings by taking a bath in the beginning of their office years. Another way in which managers take advantage of their discretion in accounting is through picking the discount rate. Carlin and Finch’s (2010) results suggest that there is bias in discount rates by applying lower rates than expected. They argue that this is due to the opportunistic behaviour of managers to avoid impairment charges.

Even though there are disadvantages of goodwill impairment, there have been benefits as well. Some researchers have found that impairment of goodwill leads to a better reflection of the true economic value. Bens (2007) found that impairment is more informative of the real economic value of publicly listed companies. Impairment was reflected timely in the market reaction through the stock price. The mixed results of whether the impairment of goodwill is manipulated by

managers for their own benefits has led me to the following research question:

To what extent is goodwill impairment conform SFAS 142 used as a tool for earnings management in the U.S.?

Motivated by the fact that the accounting of goodwill impairment involves more discretion and judgement, this study will examine whether managers manipulate goodwill impairment charges

(7)

7 after the issuance of the SFAS 142 standard. This standard provides extensive discretion for

managers as it uses a fair value based approach. Since market prices are often not available to use as a reference point the managers have the flexibility to decide how they are going to measure the fair value of goodwill. The impairment of goodwill is an alternative way to examine the behaviour of earnings management. Masters-Stout et al. (2008) have examined whether new CEOs use bath accounting through the use of goodwill impairment in the first two years when the net income is negative. Their findings demonstrate that new CEOs tend to impair more goodwill then their senior counterparts. There has not been much empirical research of whether new CEOs record accelerated goodwill impairment charges. Therefore, I thought that it would be interesting to look at the effect that new CEOs with a maximum tenure of three years, have on the impairment amount recorded during the first three years. Moreover, I specifically wanted to examine CEOs with a maximum tenure of three years so that only CEOs hired from the period 2005-2014 would be taken into consideration. Another reason to study this association was that I thought that many CEOs would have a tenure of around three years due to the recession period.

Discretion regarding the impairments of goodwill leads to earnings management and the two techniques that will be studied specifically are big bath accounting and income smoothing. There will particularly be looked at how manager’s decisions regarding the impairment of goodwill are

motivated by personal factors such as bonuses (Healy and Wahlen, 1999).

Although there is much literature on how the impairment of goodwill in accordance of SFAS 142 is used to manage earnings, I thought it would be interesting to study the effect the recession has on the impairment of goodwill. In addition, I thought it was interesting to study if earnings are managed in the recession period through the manipulation of goodwill amounts. Furthermore, this will be tested when the earnings of a firm are unexpectedly high or unexpectedly low. Until now I have not found studies that specifically research this association.

The purpose of this research is to examine whether goodwill impairment is used to manipulate earnings in the United States from the time period 2005-2014. Based on a sample of 3059 firm-year observations that are obtained from Execucomp and Compustat I have found a positive relation between goodwill impairment and earnings management. However, I was unable to obtain significant results with a positive association regarding new CEOs and impairment amount. This study is important for standard setters, investors and oversight bodies and other stakeholders because it will provide insight how the fair value used in determining the value of goodwill from acquisition is affected either positively or negatively. This will allow them to evaluate the financial statements’ quality after the implementation of SFAS 142. One of the main objectives of the FASB is to increase comparability between financial statements and increased decision

(8)

8 always contribute to this. Investors, debt lenders and society have to be aware of the implications that could possibly affect the value of goodwill in a negative way. In addition, the discretion provided by the SFAS 142 may lead to an untrue reflection of the underlying economic value of the firm.

(9)

9

2 Goodwill Impairment and Earning Management

In this chapter goodwill impairment and earnings management will be discussed. This will be followed with a development of the hypotheses.

2.1 Goodwill Impairment

As mentioned before the SFAS 142 replaced the amortization of goodwill with the annual impairment of goodwill in 2002. Goodwill has to be evaluated for yearly impairment regarding the fair value of the acquired business estimates. According to SFAS No. 121 an impairment test has to be conducted if circumstances changes to such an extent that the fair value of goodwill decreases and if it is certain that it will not regain that same value.

In this chapter American Principal Board 17 (APB 17) opinion and SFAS 142 will be discussed. Furthermore, the impairment of goodwill will be tested according to SFAS 142. In the first section a definition of goodwill will be given. Next the distinction between internally generated and externally acquired goodwill will be made. In the third section the APB 17 opinion, SFAS 141 and SFAS 142 will be addressed. In the fourth section the impairment method will be discussed and finally a summary and conclusion of this chapter will be provided.

2.1.1 Definition of goodwill

Before the impairment method is discussed it is crucial to define goodwill. Goodwill is the value above the value of the equity which can be seen on the financial statement. In purchased goodwill, goodwill is the difference between the price paid for the assets and the market value of the assets (Lander and Reinstein, 2003). When businesses are acquired or merged goodwill can come into play. The difference between the price that the acquirer pays for assets and liabilities and the fair value of the assets minus the liabilities is the goodwill. Goodwill acquired is simply the hidden economic future value paid to the other party. The amount of goodwill is subjectively determined by the acquirer. The acquirer determines the amount of goodwill based on assumptions of the future economic value.

2.1.2 Internally generated goodwill and externally acquired goodwill

Goodwill that has been created by the firm itself through various activities is called internally generated goodwill (Bloom, 2009).These activities increase the value of the company by building a customer base, expanding business and promoting brands. Standard setters have been supportive to not recognize internally generated goodwill on the balance sheet whereas acquired goodwill is capitalised. Furthermore, this goodwill is not accounted on the balance sheet due to the problems that arise for its treatment in double entry bookkeeping (Jerman & Manzin, 2013). Bloom (2009) examined 400 firms in Australia which had internally generated goodwill, representing nearly 50% of

(10)

10 their market capitalization. This indicates that internally generated goodwill can be of great value for a company.

When historical cost accounting is used because it recognizes its value at one point in time after the market transaction has taken place. It is important to understand this difference.

Particularly, the impairment of acquired goodwill will be examined in this thesis.

2.1.3 APB 17 opinion

Background on the APB 17, intangible assets was issued in 1970, for guidance regarding the accounting of goodwill (Bens et al., 2007). Under this standard goodwill was linked to the entire company. Therefore the testing of impairment of goodwill was done on the entire company. Another drawback of this standard was that it was not described when and how the impairment should be measured. Therefore this standard provided too much discretion for managers to decide on goodwill write offs. Due to the problem that this standard was non-consistent it yielded information of questionable value. Under this standard the intangible assets that were acquired were assumed to have a definite life and were amortized over a maximum period of 40 years.

The FASB issued SFAS 142 Goodwill and other intangible assets, which supersedes the APB opinion No. 17, Intangible Assets. Financial statement users such as analysts and management of firms noted that intangible assets were getting more significant economic resources for firms. In the acquisition of assets they are becoming an increasing part of the assets. Financial statement users were not regarding the amortization expense of goodwill as useful information for taking investment decisions (FASB, 2015).

2.1.4 Differences APB 17 and SFAS 142

Under the SFAS 142 some intangible assets and goodwill will no longer be amortized. Therefore the reported amount of goodwill and intangible assets will not decrease at the same time as under the APB 17 opinion. The new standard causes more volatility in the income as the impairment amounts and will occur at different times and in differing amounts. Under the transaction based approach for the accounting of goodwill the acquired business was treated as a standalone entity instead of integrated with the acquiring entity. This caused issues because the part of the premium paid for the acquired entity was not accounted properly. The SFAS 142 instead accounts for goodwill on an aggregate basis by accounting goodwill on the units of the combines unit.

APB 17 assumed that all assets had a finite life and the amounts that were assigned to them should be amortized in order to decide the net income. However SFAS 142 does not assume that and instead goodwill and other intangible assets that have an indefinite life are not amortized but are tested for impairment. There is no maximum period for amortizing finite assets.

(11)

11 process which will be discussed in the next section. In addition disclosure of information regarding goodwill is required subsequent the acquisition. For example, disclosures has to be made about the changes in the carrying amount of goodwill from period to period.

The changes which were further developed in SFAS 142 will improve the financial reporting according to the FASB. In addition the underlying economic value of goodwill will be reflected better. If the financial reporting is improved then the investors will be able to make better decisions linked to the investments in those assets and their performance as a result. As a result the shareholders will be able to determine the changes in assets over time and better predict the future profitability of the company (FASB, 2015).

2.1.5 Impairment method

Under U.S. GAAP, an entity may perform a two-step goodwill impairment test if it qualitatively determines that the fair value of the reporting unit is more likely than and not less than the carrying amount. Alternatively, they can chose not to perform a qualitative test. The asset group which is determined for goodwill is stated as a reporting unit which in addition is defined as an operating segment for which separate financial information is available (Massoud and Raiborn, 2013). In addition, SFAS 142 necessitates that goodwill impairment has to be tested on an annual basis even if there are no special circumstances.

The reporting unit’s total fair value is estimated by management or their agents (Masters-Stout et al., 2008). After this the fair value is compared with the unit’s book value. If the fair value of the unit is lower than the book value, impairment may be present. To estimate fair value, quoted market prices will give reliable information. In a case where no quoted market prices are available, information such as cash flow estimates, present value techniques and multiple of earnings can be used. In addition a more extensive test has to be completed if there is indication that the value of goodwill has decreased. With the help of this extended impairment test, it will be decided if the fair value of the reporting unit has lowered due to a decrease in goodwill or a decrease in the other assets. In addition the amount of the fair value of the reporting will be allocated to the assets and liabilities. The amount that is above the fair value is called goodwill. Furthermore, if the recorded amount is higher than this amount, the impairment has to be recorded. In order to implement this process, discretions and estimations are required by the managers.

2.1.6 SFAS 141 and SFAS 142 goodwill impairment

In 2001 the FASB issued the SFAS 141 Business combinations and SFAS 142 goodwill and other intangible assets. The significant change that the SFAS 141 brought was the elimination of the pooling of interest accounting for business combinations and this requires the purchase accounting method for all accounting transactions (Jennings, LeClere and Thompson, 2001).

(12)

12 The SFAS 142 prohibited the amortization of goodwill that was purchased. Once the

amortization method was eliminated, the firms were required to use the impairment method of goodwill. Firms are required to test goodwill for impairment at least once a year using a two- step test. However additional testing may be needed. Firstly, one has to determine if an impairment has occurred. This is done by comparing the fair value of the reporting unit with its book value. Secondly, if it is determined that the value of goodwill is less than its current book value, an impairment charge has to be recorded in income and goodwill has to be written off.

2.1.7 Goodwill impairment used for earnings management

Unfortunately, the criteria used for the impairment of goodwill leaves a significant amount of discretion for judgement bias and manipulation, at the time of acquisition of a business as well as in the upcoming future periods (Massoud and Raiborn, 2003). This may lower the quality of the

earnings numbers. There are many ways in which the accounting for goodwill may be manipulated by managers.

Firstly, when operations are at a downturn management may opt to record large goodwill write offs. Managers can reason that this can bring them future income statement benefits because taking an impairment during such a period would not make a significant difference after all.

Additionally, Massoud and Raiborn (2003, p. 30) call goodwill impairment as “asset rightsizing” or “future income cosmetic enhancement”.

Secondly, managers may take goodwill impairments in times when earnings are

proportionately above the expectations of analysts. Massoud and Raiborn (2003, p. 31) state that firms that “make the numbers” have been the darlings of Wall Street.

Lastly, managers may be highly creative in calculating fair market values of the reporting firms and computing a basic impairment test for legal reasons. Should manager for example be held responsible by the shareholders, if excessive amount of goodwill is created as a result of an

acquisition, but it was a poor acquisition? Managers might record an amount above the fair value of the reporting unit so that the goodwill will stay unimpaired.

Massoud and Raiborn (2003, p. 31) suggests that companies have the possibilities to choose points in time to identify goodwill impairment in a way that selectively fits their “operating results”. As a result for the extreme opportunities for manipulation by managers, empirical evidence for the manipulation of goodwill impairment charges by companies will be examined.

2.1.8 Empirical evidence goodwill impairment

Authors such as Beatty and Weber (2006), as well as Massoud and Raiborn (2003) argue that management is left with much discretion, which allows them to opt for earnings management. They have discretion in making assumptions and estimations in identifying the reporting units and how

(13)

13 assets liabilities and goodwill is assigned to the reporting units.

Furthermore Zang (2008) found that managers don’t impair goodwill in order to restrain from violating debt covenants. They also found that manager turnover leads to great amount of impairment of goodwill in the beginning so that they can avoid impairments in the future.

The recoverable amount in the impairment method also involves subjectivity by the manager. Wines, Graeme and Dagwell (2007) argue that the impairment of goodwill method allows managers to act opportunistically. In particular they highlighted that the valuation and identification of cash generating units needs many assumptions in order to estimate fair value, recoverable amount and value in use. This is due to the result that there is limited market information available to compare information. Furthermore, this effect is intensified when a new management team comes. They argue that the managers overvalue the recoverable amount so that they take a “big bath” during the transition period to prevent losses in the future as much as possible.

Henning and Shaw (2004) investigate the criticism that U.S. GAAP gives too much discretion regarding the timing and amount of goodwill written off. Their results do support the criticism that the timing of impairments are delayed. In addition, they found that write offs of goodwill were delayed until the new rule was issued and had become effective.

Another relevant research regarding the impairment of goodwill was conducted by Van de Poel, Maijoor and Vanstraelen (2008). They had conducted research in the European countries from 2005-2006. They found that goodwill impairment are related to the reporting incentives that

managers have. In addition they found significant results that firms typically impair goodwill in order to take a bath or smooth earnings. Impairment charges are recognized when the earnings are either unexpectedly low indicating that big bath accounting is used. On the other hand, when earnings are unexpectedly high and impairment loss is recognized as well indicating that earnings are smoothed. Zucca and Campbell (1992) found that the majority of the firms in their sample wrote down their assets. They conducted a database research from the period 1978-1983. They had chosen this period to study discretionary write-downs. They found that a great percentage of the firms wrote down assets in a period where the earnings were already below the normal earnings indicating big baths were taken by firms. In addition they found that 25% of the firms off set the impairments with other gains or extremely high earnings indicating income smoothing. They concluded that write offs are used by managers to manage earnings.

Chen, Kohlbeck and Warfield (2008) examined the timeliness of impairment recognition in the period after the issuance of SFAS 142.The new standard gave the opportunity to allow firms to report two types of impairment in the adoption year. One was the “adoption” impairment which would be recorded as the cumulative effect of the change in standard. This recording of impairment would be recorded below-the-line extraordinary loss and could be related to previous periods as well.

(14)

14 After this the company could decide whether the goodwill was impaired during the year and this would be recorded against operating income. Their findings suggest that the impairment charge that would be recorded against operating income would be delayed by a year.

2.1.9 Goodwill impairment before and after the financial crisis

Impairment testing of goodwill has become a hot topic after the taking over of firms in the boom phase and the financial crisis diffusing into the general economy (Camodeca and Almici, 2012). The information content of goodwill impairment was tested for European banks before and after the financial crisis from 2006-2011. There was a significant amount of goodwill recorded by the firm before any impairment was recorded, especially before 2008. The goodwill was written off in 2008 mainly and was trending down in 2009 and 2010. Furthermore, the steady drop of real GDP in 2008 was related to an increase in goodwill impairment. Additionally, the level of disclosure on the variables used in the discounted cash flow model (DCF model) increased in the periods 2008-2011. This model is used to estimate the recoverable amount, such as the discount rate and growth rate. They further argue that their analysis show that some of the important issues of the model in IAS-36 seem to be confirmed with regards to the banking sector by the discretion used by managers in the DCF calculation model.

Another example which recorded a significant amount of goodwill loss is TomTom (Privat, 2009). TomTom Group recorded a goodwill impairment loss of 1,048 billion euro. They wrote off more than one third of the amount they had paid to acquire Tele Atlas. The results of Tele Atlas have been disappointing in 2008. They had a drop in revenues of 6% compared to 2007 and a loss of 9 million euro compared to 1 million euro in 2007. These bad results explain the large impairment loss recognized by TomTom.

2.2 Earnings management

In this section of the paragraph forms of earnings management will be discussed. First earnings management will be defined.

2.2.1 Definition of earnings management

In order to gain a more general picture of how earnings management is defined, the following definitions from prior academic literature are given.

Schipper considers earnings management from an informational view as differentiated from an economic view and explains earnings management as follows:

“A purposeful intervention in the external financial reporting process, with the intent of

obtaining some private gain (as opposed to, say, merely facilitating the neutral operation of the process)… (Schipper, 1989, p. 93)”

(15)

15 In this definition he further explains that earnings management can be present in all parts of the external disclosure procedure. A slight extension of this explanation would be real earnings management, which is done to alter reported earnings. Real earnings management is accomplished with timing investment or financing decisions.

Another definition comes from Healy and Wahlen (1999)

“Earnings management occurs when managers use judgement 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. (Healy and Wahlen, 1999, p. 369)”

This definition implies that managers use earnings management in order to mislead stakeholders or either to influence contracts which depend on reported accounting numbers. Although these definitions are widely accepted, they are hard to operationalize because it is based on managerial intent, which cannot be observed (Dechow and Skinner, 2000). In addition clear definitions of earnings management are difficult to discern from pronouncements, one example of an extreme form of earnings management is financial fraud. Financial fraud is defined as:

“… the intentional, deliberate, misstatement or omission of material facts of accounting data,

which is misleading and, when considered with all the information made available, would cause the reader to change or alter his or her judgement or decision (National Association of

Certified Fraud examiners, 1993)”

The above provided definitions indicate that managers manage earnings depending on the alteration of the accounting data so that they can manipulate financial statements. Having provided these definitions it is clear that earnings management is operationalized to get gains. Healy and Wahlen (1999) state that there is evidence in general, that companies manage earnings to window-dress financial statement before public securities offerings, to increase corporate managers’ compensation and job security, prevent debt covenant violation and reduce regulatory costs or increase regulatory benefits.

There are many ways in which earnings can be managed because of the discretion that managers have in making judgements. Managers have to make estimations regarding future economic events such as expected lives and salvage values of long-term assets, obligations for pension benefits and asset impairments. In addition they also have to choose to make or defer expenditures such as research and development (R&D). Furthermore, judgement is exercised by manager in estimating working capital as well (such as receivables, inventory levels, the timing of inventory shipments and purchases).

(16)

16 Another point to note in the definition of Healy and Wahlen (1999), is that the objective of earnings management is to misguide shareholders and other stakeholders about the underlying economic value of the firm. Managers are able to mislead stakeholders because they have inside information about the firm that the stakeholders do not have so that earnings management is not visible for them (Stein, 1989). This can lead to stakeholders expecting a certain amount of earnings management.

Dechow and Skinner (2000) clearly differentiate between managerial choices that are

fraudulent and those that are contentious, but acceptable in which managers can use their discretion. The main point that they want to clarify is to distinguish between fraudulent accounting and earnings management using the following groups:

Accounting choices Within GAAP

Conservative accounting -Overly aggressive recognition of provision or reserves

-Overvaluation of acquired in process R&D in purchase acquisitions -Overstatement of restructuring charges and asset write off Neutral earnings Earnings that result from a neutral operation of the process “Aggressive accounting” -Understatement of the provision for bad debts

-Drawing down provisions or reserves in an overly aggressive manner

VIOLATES GAAP

“Fraudulent accounting” -Recording sales before they are realizable -Recording fictitious sales

Backdating sales invoices

-Overstating inventory by recording fictitious inventory

Although in theory the distinction between earnings management and fraudulent accounting may seem clear, in practice it can be difficult to observe earnings management and fraudulent accounting. The three forms of accounting that are within the GAAP are conservative accounting, neutral earnings and aggressive accounting. These three forms can be considered as earnings management if they are used to “obscure” or “mask” the true underlying economic value of the entity which depends on managers intentions (Dechow and Skinner, 2000). However, fraudulent accounting is out of the GAAP principles and these methods are treated as fraud.

2.2.2 Accrual accounting and Earnings Management

Earnings management is not a recent phenomenon; discussions have been going on for decades. Earnings management is caused from accounting conditions such as the practice of accrual accounting.

(17)

17 earnings management and the effects it had on the distribution of resources (Healy and Wahlen, 1999). He expressed his concern about managers abusing “big bath” restructuring charges, “cookie jar” reserves and recognizing revenues prematurely. In addition, write offs of purchased in-process R&D are also endangering the trustworthiness of financial statements. After these concerns were expressed actions were taken by the SEC by forming an earnings management task force to take actions against firms who manage earnings. Furthermore, the SEC requires more firms to restate their earnings and will enforce them to oblige to disclosure requirements.

Accruals consist of the discretionary part and nondiscretionary part (Dechow, 1993). As it has been mentioned previously, earnings management is a strategy used by management to manipulate the earnings of a firm to achieve a certain target (Investopedia, 2015). The analysis of earnings management often involves the use of discretionary accrual by managers (Dechow, Sloan and Sweeney, 1995, p. 194).

The objective of accrual accounting is to evaluate the firm’s economic performance during the period. This is done through the use of basic accounting principles such as revenue recognition and matching (Dechow and Skinner, 2000). The accrual process causes earnings to be smoother than the underlying cash process. In addition it also tends to provide more accurate information about the future economic performance compared to cash flows. However, when there occurs a point that manager’s decisions regarding accruals become too much income smoothing, there needs to be paid attention (Dechow and Skinner, 2000, p. 238).

Investopedia gives the following definition of accrual accounting:

“An accounting method that measures the performance and position of a company

regardless of when cash transaction occur. The general idea is that economic events are recognized by matching revenues to expenses at the time in which the transaction occurs rather than when the payment is made. This method allows the current cash inflows/outflows to be combined with future expected cash inflow/outflows to give a more accurate picture of a company’s current financial condition. (Investopedia, 2015)”

There are models that tend to separate total accruals in discretionary accruals from nondiscretionary accrual. Accruals are the difference between financial results and cash flows. The nondiscretionary accrual part is not influenced by managers. Accruals can be influenced by managers to a certain extent. The nondiscretionary accruals are the accruals that are affected by managers. On the other hand discretionary accruals are affected by managers. Although managers have influence on accruals, they are regulated by regulatory bodies as well as auditors to have oversight over their actions.

(18)

18 managers can have incentives to manipulate data which is used in the contracts. Managers use discretionary accruals to perform earnings management, though this is not always so (Schipper, 1988). The inability to undo earnings management is due to the fact that there is information asymmetry. Managers have inside information that the stakeholders do not have. Therefore a condition that has to be present for earnings management to exist is information asymmetry. One assumption that allows earnings management is the form of blocked communication which can’t be changed by agreements in the contracts. Blocked communication means that all private information is not communicated.

In order to understand the condition how earnings management occurs, it is important to understand the principal- agency problem. In a company the managers are the agents and the stakeholders are the agents. The agency relationship can be defines as follow:

“… a contract under which one or more persons engage another person to perform some

service on their behalf which involves delegating some decision making authority to the agent. (Jensen and Meckling 1976, p. 308)”

Furthermore, if both the agent and the principal want to maximize their utility there is a bigger chance that the agent will not always act in the interest of the principal. However, to align the actions of the agent more within the interest of the principal, he can establish incentives for the agent and monitor the agent by incurring costs. In addition, there will be situations that it will be beneficial for the agent to incur costs to guarantee that he will not cause harm to the principal. The agent could also expend resources to ensure that the principal will be compensated if the agent does not take harmful actions.

However, it is difficult to ensure that the agent will make optimal decisions from the principal’s viewpoint at zero cost. In the agency relationship there will be positive monitoring and bonding costs which will be incurred by the principal and the agent. Furthermore, there will generally be variation in the decisions of the agent and those decisions that would actually maximize the welfare of the principal. This cost is called the residual loss which is caused due to the reduction of the welfare experienced by the principal as a result of this divergence.

2.2.3 Motivations for earnings management

In previous literature various motivations for earnings management have been discussed. Despite, the fact that earnings management does exist in general; there have been significant difficulties for researchers to credibly document it (Healy and Wahlen, 1999). Before the effects of earnings

management are taken into consideration, the earnings have to be estimated. One important step in this process is to determine the conditions where earnings management incentives for manager are

(19)

19 the strongest.

Healy (1996) considers two motivations that managers have to manage earnings in his paper. These are performance measurement and opportunism. The first motive, performance measurement, assumes that managers use accruals to make accurate signals of firm value, improve the value of accounting as a language for informing the investors. For instance, managers can write off some assets if they think that their long-term performance has declined.

On the other hand, the opportunistic hypothesis speculates that discretion is used by managers to obscure poor performance and magnify good performance. This is done to keep their job or maintain their compensation perks or either to move the wealth from investors to other stakeholders such as debtholders etc. An example of opportunistic behaviour by a manager is that the he or she may not recognize asset impairments on a timely basis out of fear to lose their job or to default on debt covenants. However, Healy (1996) explains that managers have other motivations for earnings management as well. For instance, they may have motivation to lower taxes or to lower regulatory costs. In another research by Healy and Wahlen (1999) it is claimed that researchers have studied various incentives to manage earnings. Two different incentives that managers have to manage earnings which are the most relevant in that paper, are on the one hand capital market expectations and valuation, and on the other hand contracts written in terms of accounting numbers.

2.2.4 Capital market expectations and valuation

Incentives to manipulate earnings can be established by the fact that accounting information is used extensively to value the stock price by stockholders, specifically investors and by financial analysts. Dechow and Skinner (2000) state that according to some researches managers have incentives to meet simple benchmarks. These simple earnings benchmarks include: (1) avoiding losses, (2) reporting increases in seasonally adjusted quarterly earnings and (3) meeting analysts’ expectations for quarterly earnings (Dechow and Skinner, 2000, p. 242).

Market prices are sensitive to benchmark and these earnings patterns have become more persistent over the years (Skinner and Sloan, 2002). The capital market incentives that managers have to manage earnings have been intensified through the widespread use of accounting

information by investors and financial analysts to enable the valuation of stocks (Healy and Wahlen, 1999). Managers want to meet certain benchmarks set internally by managers and externally set by analysts. If they don’t meet the benchmarks it may have consequences for the market prices.

Earnings are manipulated upwards to influence certain investors. Firms manipulate earnings through cutting on R&D costs to prevent a drop in earnings particularly within firms that have a high

percentage of ownership by institutions with momentum trading strategies and a high portfolio (Healy and Wahlen, 1999).

(20)

20 There is evidence provided on capital market incentives in a couple of studies. For instance, Burgstahler and Eames (2006) find that managers manage earnings upwards in order to meet analysts’ forecasts. Particularly they manage earnings upward to prevent the reporting of earnings below the forecasts of analysts. In another research it is found that firms who report continuous growth in the yearly reports are priced at a premium compared to other firms (Barth, Elliot and Finn, 1999). As the length of the string increases, the premium rises as well. Likewise as the string

disappears, the premium is reduced. Furthermore, in the research by Skinner and Sloan (2000) adverse earnings revelations have a disproportionate effect on stock prices especially for growth stocks. In other words when there are even slight disappointments in earnings reported by growth stocks compared to analysts’ forecasts, their stock price can decline tremendously.

2.2.5 Contracts written in terms of accounting numbers

To control for the contracts between the firm and its stakeholders data is used to have a regulation between both parties (Healy and Wahlen, 1999). In order to align the incentives of managers with external stakeholders there are compensation contracts used. In addition, there are lending contracts written so that stockholders are not benefitted at the expense the creditors due to the manager’s actions. Earnings management caused due to contracting motivations is important as well because financial information is communicated to stock investors, debt investors and investors’ representatives on board of directors.

There are studies that have studied whether earnings are managed by firms who are close to violating their debt covenants and are in dividend constraint. For instance, Deangelo, H., Deangelo, E. and Skinner (1994) study whether the 76 NYSE firms who had continuous losses and dividend

declines (which were forced by binding covenants) altered their accruals, accounting methods or accounting estimates. They find little evidence of earnings management for the firms close to their dividend covenants. These firms placed more importance on the reduction of dividend payment and restructured their contractual agreements by managing their cash flows.

The bonus hypothesis of the contracting theory states that firms who have bonus plans for managers will lead managers to choose estimations and accounting methods that will maximize their short term and long term bonuses based on the earnings of the firms. In addition, managers may have incentives to increase earnings when the earnings are between the lower bound and higher bound of the earnings level to receive bonuses. However, firms do not always have the motivations to manipulate earnings by increasing earnings. Alternatively, if managers are below the lower bound of receiving a bonus or above the upper bound of the bonus they may have incentives to decrease earnings. Strong and Meyer (1987) reveal that new CEOs in their first year have incentives to lower the earnings so that the benchmark against which the future earnings are assessed regarding their

(21)

21 bonuses will be lower. Additionally, by reducing the current earnings they will increase the bonuses that they will receive in the future and it have more job stability.

Although, firms who pay dividends can cut their dividends when they have dividend

constraints, it may be more difficult for firms to meet other covenants. In addition, firms may have to meet debt-equity ratios for their covenants and meet their interest coverage ratio. For instance, Press and Weintrop, (1990), DeFond and Jiambalvo (1994) and Sweeney (1994) examine firms who violated their debt contracts and they had mixed results.

Press and Weintrop (1990) study public and private debt agreements for 83 American Stock exchange firms. They chose firms who were similar in size and risk level. They find that significant variation in leverage levels exist between those companies that have accounting based constraints compared to those that don’t have that. Additionally, the find that firms’ there is a significant association between the accounting choice with leverage and leverage constraints. In general, firms with high leverage and close to the violation of debt covenants, usually try to avoid the violation of debt by shifting the income from future periods to the current period (Warfield, Wild and Wild, 1995).

Sweeney (1994) finds that managers act by income-increasing accounting changes who are moving closer to default. In addition, he find that the default costs that the lenders have to pay and the flexibility that managers have regarding the accounting are significant determinants of the way managers respond to the violation of debt contracts. He finds that managers of firms approaching default respond with income-increasing accounting changes and that the default costs imposed by lenders and the accounting flexibility available to managers are important determinants of managers’ accounting responses. The income increasing accounting changes that covenant violators make are typically done after the covenant violation. This finding shows that these firms did not necessarily make accounting changes to prevent violating the covenant (Healy and Wahlen, 1999) However, it could indicate that the income increasing accounting changes were possibly made to prevent future debt covenant violation.

Furthermore, DeFond and Jiambalvo (1994) examine abnormal accruals of 94 companies that had violated their debt contracts in the annual reports. They find that in the year prior to debt covenant violation that violation firms have significantly positive working capital and abnormal total accruals. The fact that the firms used in the sample accelerated their earnings one year before the covenant violation is thought to indicate according to the authors as evidence for earnings

management.

2.2.6 Big bath accounting and Income smoothing

The main types of earnings management that will be discussed are “big bath accounting” and “income smoothing” since this will be relevant in the research. Big bath accounting follows a simple

(22)

22 reasoning where earnings are manipulated to look worse in the recent period (Jordan and Clark, 2011). When profits are depressed, large non repetitive losses are taken in order to prevent the burdening on future earnings. As a result of this action, the future earnings are either higher or are less variable over the years. The reasoning behind this form of earnings management is that when the earnings are already depressed then clearing out small mess would do limited harm to the manager or the company’s reputation. Furthermore, Gode and Mohanram (2003) argues that big bath accounting is used by firms that did not reach their earnings target. In addition, he says that the year is lost anyways so if it incurs more cost in the bad year because the additional costs are minimal as the targets have not been achieved.

Kirschenheiter and Melumad (2002) found that managers have incentives to engage in income decreasing behaviour. In addition, they find that reporting larger earnings surprises lowers the earnings precision, reducing the impact of reporting higher earnings and a demand for smoother earnings. They show that for “bad news” the earnings are lowered as much as possible by taking a “big bath” so that they can report higher earnings in the future. This could be done by accelerating the impairment charges. On the other hand, if the news is good the earnings are manipulated by smoothing the earnings. Here the amount of earnings that are smoothed depend on the cash flows.

Fudenberg and Tirole (1995) examine where contracting motivations drive earnings smoothing. Their main findings are that income smoothing is used in bad times by boosting the income reported and the income reported during good time are manipulated by lowering the earnings. Trueman and Titman (1988) discuss that income is smoothed in order to demonstrate to potential debtholders that earnings have a lower volatility; therefore there is limited risk present. The advantage that comes along with this is that since debt costs will be lower, and hence there will be increased expected cash flows to investors.

2.3 Hypothesis Development

Wanting higher bonuses and preventing debt covenant violation are some of the motivations of managers to manipulate earnings (Healy and Wahlen, 1999). The link between reporting incentives and goodwill impairment can be examined in this regard. Managers have strong incentives to write off goodwill (Francis, Hanna and Vincent, 1996). In addition, Guler (2006) found that executives reporting incentives affect the decision of write off. He found a negative association between bonuses based compensation plan and recognizing goodwill impairment.

New managers may have incentives to record large impairment charges. Masters- Stout, Costigan and Lovata (2008) examine publicly listed companies where the link between CEO tenure and goodwill impairment charges is studied. They hypothesize that in the early tenure of the CEOs impairment will be recognized so that previous management’s acquisition decisions can be blamed

(23)

23 on. By recognizing the impairment expense in the early years, the future earnings will look better than it actually is. A new management may consciously overstate impairment charge because it may attribute the impairment charge due to the poor decisions made by the previous management so that they can reduce the possibility of future impairment charges (Zang, p. 43). Wells (2002) studied Australian companies from 1984-1994 regarding the earnings management connected to CEO changes. His main findings were that external CEOs manage earnings downward. However, higher write offs by new CEOs may be justified in some situations (Pourciau, 1993). It is possible that firms who are financially troubled want to improve their performance by hiring new CEOs. However, new CEOs may have incentives to manipulate earning by writing the goodwill off in order to clear the decks. These new CEOs have incentives to impair so that they can set a lower benchmark against which the performance of the firm is measured. For the following hypothesis the proxy variable that will be used is CEO change. It is predicted that a new CEO with a maximum tenure of three years will have a positive relation with the impairment amount. Usually the CEOs have a tenure longer than three years especially if the company is performing well. Since, I have taken data from the recession years as well. I thought it was interesting to see if CEOs impair goodwill in their first three years as well. Therefore the first hypothesis is established as follow:

H1: A recent CEO change with a maximum tenure of three years will affect the impairment amount positively

Secondly, managers have incentives to transfer excess earnings to future periods when the economic climate is poor and hence the earnings are low. If the earnings are below the normal earnings level of the industry median than it can be said that the firm is not performing well. This is done so that there will not be a need to impair goodwill in the future. As discussed before

management can take a “big bath” and therefore improve future earnings to a certain extent. For example, when earnings are excessively low the management bonus will not be reached anyways. Therefore, management has incentives to accelerate the impairment of goodwill because the ceiling of management’s bonus has not been reached. The bonus of management has a lower ceiling of earnings target which has to be achieved in order to get a bonus. By recognizing an impairment loss in this period, management has a higher chance of receiving the bonus in the next year. To test the second hypothesis the variable which will be measured is BATH. By using this variable it can be tested if the income before taxes is below the median of the specific industry. This is consistent with

Lapointe-Antunes et al. (2008) who find that firms report goodwill impairment charges in order to limit the deviation from the industrial return on assets. If a firm accomplishes lower return on assets than the industrial average than these earnings are considered to be unexpectedly low. Van de Poel

(24)

24 et al. (2008) find that firms have low unexpected earnings will have a positive effect on the

impairment amount. Based on this explanation the second hypothesis is:

H2: Unexpectedly low earnings have a positive impact on the impairment amount recorded by managers

Thirdly, based on the article of Kirschenheiter and Melumad (2002) as discussed previously, managers have incentives to smooth earnings when earnings are unexpectedly high. They stated that if earnings are unexpectedly high it can negatively affect the precision of the earnings and therefore negatively affect the value of the firm. Zucca and Campbell (1992) state that managers may want to manipulate earnings in order to comply with market demands to show a steady earnings growth rate. Based on this the following hypothesis is established. In a situation where the earnings amount is unexpectedly high so that the bonus stays stagnant and does not increase with the earnings amount anymore. In this situation it is not beneficial for the manager anymore to increase the earnings. In addition accelerating the impairment charge would be more profitable for the manager since they would get more chance to increase the earnings in the future and increase their bonuses as well. I expect that an unexpectedly high earnings will have a positive effect on the impairment amount. The proxy is established to see if the earnings significantly deviate from the median ROA of the industry. This leads me to the following hypothesis:

H3: Unexpectedly high earnings have a positive impact on the impairment amount recorded by managers

Lastly, the effect that the recession has on the impairment of goodwill has not been examined yet, as a tool for earnings management. During the financial crisis managers have more opportunity to manipulate earnings as the there is more uncertainty due to the credit crunch. It was also interesting to look whether there was a significant effect of the recession on the impairment of goodwill

compared to the non-recession period. The recession period is defined from the period 2007-2009. According to (Sinnet, 2013) goodwill impairment has stabilized since the peak of 2008.

The Financial Executives Research Foundation (FREF) examined 5000 companies in their goodwill impairment study of 2012 (Sinnet, 2013). The study indicates that although the goodwill impairment amounts have decreased significantly from 2009-2011 with impairments around 26 to 30 billion dollar. This is a significant decrease in goodwill impairment compared to the goodwill

impairment charge which had a peak of 188 billion dollar. Some companies which had the highest amount of goodwill impairments were AT&T, Dean Foods and Bank of America.

Furthermore, it is claimed that the impairment of goodwill/total goodwill was 1% from 2009-2011 while in the period of recession the rates were often 10 times this amount. In addition, the

(25)

25 ratio of goodwill impairment to total assets was on average far larger in the year 2008 with a

percentage of 1.1% while in 2011 it was 0.2%. The amount of goodwill reported on the balance sheets differs between industries. For instance, the consumer staples and telecom services have a high ratio of goodwill to total assets while utilities and energy have a low rank ratio of goodwill to total assets. This has led me to test if the recession truly had a significant effect on the decision to impair goodwill. It certainly is logical that the recession would have an effect on the impairment of goodwill. Especially if one reads articles on companies that had to write off all the goodwill from the balance sheet during the recession.

In addition, I would like to examine the proxies bath and smooth in the recession. I expect a positive relation between bath and impairment amount in the recession. However, I expect a negative relation between smooth and impairment amounts during the recession. This is due to the fact that overall many companies had operating earnings which were affected negatively to a great extent by the recession. Therefore, I don’t expect that unexpectedly high earnings will have a positive influence on the impairment amount. In general, when earnings are high or unexpectedly high I think that managers rather have incentives to increase earnings during the recession. Additionally, managers would want the companies to make progress and increase earnings so that their personal motives of getting a bonus or a higher bonus could be fulfilled as well. In the credit crunch the managers would be less inclined to record an impairment charge to smooth earnings. I expect a positive relation between goodwill impairment and the variable bath. However, I expect a negative association between the variable smooth and impairment amount. This leads me to the following hypotheses:

H4: Unexpectedly low earnings have a positive effect on the impairment amount recorded by managers during the recession

H5 Unexpectedly high earnings have a negative effect on the impairment amount recorded by managers during the recession

(26)

26

3 Methodology

In this chapter the methodology will be developed based on the literature discussed in the previous chapter. First the sample and data will be discussed. This will be followed by a model designed by me, to test the four earlier formulated hypotheses.

3.1 Sample Selection

The sample consists of all firms covered by the Compustat (North America) and Execucomp database from 2004 to 2014. Compustat (North America) is a database of U.S. and Canadian fundamentals and market information on active and inactive publicly held companies. Execucomp tracks executive compensation in S&P1000 firms. From the Compustat database, variables on goodwill impairment are selected. Execucomp will be used to provide me with info regarding CEO tenure of these companies.

The hypothesis will be tested for companies in the United States. These firms will be tested to give a general result regarding firms in a similar economy. In addition, the reporting standard are similar in the US because the US GAAP is used. Canadian firms are therefore excluded for this research. The time period that will be studied is from 2005-2014. This time range has been chosen in order avoid the first couple of years when the standard was issued. In the transition years companies were not obliged to record the impairment charges in the operating earnings. This period is chosen since it seems interesting to examine the effect that the recession had on earnings management and goodwill impairment. This has been done because the transitional impairment recorded in the transition year was not recorded in the operational earnings. Moreover, firms that have no impairment amount are excluded, since they are not relevant for this research. This leads to an exclusion of 61,983 firm-year observations. Financial institutions with SIC codes between 6000 and 6999 are also exclude due to the difference in accounting regulations compared to the other industries. This results to the exclusion of 493 firm-years observations. This yields a final sample of 3,059 firm-year observations derived from 1,597 firms.

3.2 Development of the model

While taking the previous models as a foundation I estimated the following empirical model:

Impairment amount =  +CHANGE CEO + BATH + SMOOTH+ RECESSION +

SIZE+GOODWILL +  ROA +  LEVERAGE+ CHANGESALES +ε

(27)

27 TABLE 1

Variable definitions

Dependent variables

Impairment amount Impairments of goodwill pre-tax yeart deflated by assetst-1

Independent variables

Change CEO Change CEO equals 1 If CEO tenure is less than three years, otherwise record a 0. According to theory a positive relation is to be found between change CEO and the impairment decision.

+

Bath To proxy bath accounting it has to be calculated (Operating earnings year t) – (operating earnings year t-1))/total assets year t-1. This is a truncated variable and it will have the value of the unexpected negative earnings if the earnings are below the median of negative and non-zero values and if the unexpected earnings are above the median of the nonzero negative earnings it will have a value of 0. It is expected that a positive relation will be found between the impairment amount and bath.

+

Smooth To proxy smooth accounting this formula has to be calculated: ((Operating earnings year t - impairment) – (operating earnings year t-1 –impairment))/total assets year t-1.This is also a truncated variable and it will have the value of unexpected positive earnings if the earnings are above the median of positive and non-zero earnings and if the unexpected earnings are below the median of the nonzero positive earnings it will have a value of 0.It is expected that a positive relation will be found between smooth and the impairment amount

+

Recession If the years are either 2007, 2008 or 2009 it indicates it’s a recession year and it is equal to 1, if not it is 0. It is expected that a positive relation will be found between recession and impairment amount. However, there is a negative relation expected between bath and impairment amount during the recession. On the other hand, there is a negative relation expected between the variable smooth and impairment amount during the recession.

-/+

Control variables

Size Natural log of total assets. It is expected that there will be a positive relation between size and impairment amount

+

Goodwill Goodwill before impairment/total assets. The goodwill amount needs to be measured relative to the firm’s size instead of getting an absolute value of goodwill. Firms that have a higher amount of goodwill compared to their assets will record a higher impairment charge. Therefore it is expected that there will be a positive relation found between goodwill and the

(28)

28 impairment amount.

ROA Net income/ lagged total assets. It is expected that there will be a positive relation between ROA and impairment amount

+

Leverage This is the ratio of total debt / total assets. Therefore you calculate the ratio as followed LEVERAGE= total debt/ total assets. It is expected that there will be a negative relation between leverage and the impairment amount

-

Change sales (sales year t-sales year t-1)/ assets b year t-1. It is expected that there will be a negative relation between change sales and the impairment amount

-

3.2.1 Dependent Variable

The dependent variable is the ‘Impairment amount’. Lapointe-Antunes et al. (2008) use the amount of transitional goodwill impairment as the dependent variable. This was calculated by them as TGIL (transitional goodwill impairment loss) = reported transitional goodwill impairment loss deflated by lagged total assets. However, in this thesis the year that transitional goodwill impairment could be measured are not used. This was done to exclude the effect of the transitional year. Therefore, in order to calculate this variable the impairment amount will be divided by the lagged assets.

3.2.2 Independent variables

There are several independent variables in the model. Firstly, there is the variable ‘Change CEO’. This variable is used in order to test the first hypothesis. I will examine CEOs with a total tenure of three years to see if an impairment charge in year t and year t-1 is recorded by the CEO. The dummy variable is equal to 1 there is a change in CEO and impairment is recorded and 0 if otherwise.

Furthermore, a number of studies have found that CEOs manipulate earnings downward in the hope for improved earnings in the future and to set a lower benchmark against which their performance is measured (Masters-Stout et al., 2008; Pourciau; 1993). Furthermore, authors such as Strong and Meyer (1987) reveal that new CEOs in their first year have incentives to lower the earnings to set a lower benchmark against which their performance is measured. Masters-Stout et al. (2008) found that new CEOs tend to take most impairment charge in the first two year compared to the rest of their years in office. New managers have the incentive to take a bath. They used a sample of the 500 biggest firms in the world from the time period 2004-2006. They find evidence that new CEOs are more likely to record a goodwill impairment in the first two years in office than their predecessors. Besides this they also found an association that the lower the net income, the higher the impairment charge recorded by CEOs. Lapointe-Antunes et al. (2008) found that firms understate or overstate their impairment charge in order to limit the deviation from the industrial’s return on assets.

(29)

29 impact on the decision of managers to write off. Similarly, Van de Poel et al. (2008) found that the dummy variables for bath and smooth were significant at the 5 % level. Therefore the 2 variables that will be used are bath and smooth.

Secondly, another independent variable that is going to be tested in this thesis is ‘Bath’. Master-Stout et al. (2008) argue that if a company is in a loss position, then the company might be willing to take a big bath. These earnings can be measured as the pre-impaired operating earnings before taxes in year t minus the operating earnings before taxes in year t-1 dividing this by the total assets in year t-1. Earnings are considered to be unexpectedly low if the nonzero negative values are below the median of the nonzero negative values. This is the calculation as used as by Van de Poel et al. (2008) and Francis et al. (1996). These authors found that when firms earnings are

unexpectedly low managers record impairment losses. The income reported during bad times are manipulated by lowering the earnings even further (Trueman and Titman, 1988). They found that there is a high association between the firm’s impairment decision and their reporting incentives. This variable will have a positive relation with the impairment amount.

The variable ‘Smooth’ is equal to the changes in earnings from year t to t-1 before impairments and before taxes divided by the total assets. Earnings are considered to be

unexpectedly high if the non-zero positive values are above the median of nonzero positive values. This is the method as used by Van de Poel et al. (2008) and Francis et al. Additionally, Fudenberg and Tirole (1995) find that income smoothing is used in bad times by boosting the income reported and the income reported during good time are manipulated by lowering the earnings. Furthermore, Trueman and Titman also find that if income is high it is manipulated by lowering the earnings. This variable is expected to have a positive relation with the impairment amount.

Lastly, there is the variable ‘Recession’ which analyses the impact of bath and smooth on the impairment amount during the recession. If the years are either 2007, 2008 or 2009 it’s a recession year and it is equal to 1, otherwise a 0. It is expected that smooth will have a negative impact on the impairment amount. On the other hand, it is expected that bath will have a positive effect on the impairment amount. Firms tend to lower their earnings even further when earning are unexpectedly low (Francis et al, 1996). During the recession firm had to record large losses, so in a case where earnings are already low it would not have an impact on the bonus of a manager if it is far below the lower bound of the bonus (Watts and Zimmerman, 1990).

3.2.3 Control Variables

The control variables are held constant to test the relative impact of independent variables. There are five control variables in this model.

Goodwill impairment used for earnings management