Are Goodwill Impairments Value Relevant? : A comparative study between two European countries

(1)

Are Goodwill

Impairments Value

Relevant?

(2)

Master Thesis in Business Administration

Title: Are Goodwill Impairments Value Relevant? Authors: Gustav Posth, Marcus Stoltz

Tutor: Andreas Jansson Date: 2021-05-24

Key terms: goodwill impairments, value relevance, learning effect, institutional settings

________________________________________________________________________

Acknowledgements

First and foremost, we would like to express our sincerest gratitude to our supervisor Andreas Jansson. We would like to thank him for his immense knowledge, unwavering enthusiasm, incredible personality and impeccable sense of humour. With his guidance and presence in this final semester of our studies we could not have been better off. A better supervisor and mentor would be impossible to find.

We would also like to thank our friends and colleagues for helpful comments on how to improve our thesis during the time it was written.

Jönköping International Business School May 2021

_____________________________ _____________________________

(3)

Abstract

The purpose of this study was to examine the value relevance of goodwill impairments in regard to market value. A quantitative methodological approach was used to determine if the relation between goodwill impairments was influenced by different institutional settings, which was the first research question. The second research question was to see if there was a learning effect associated with the use of goodwill impairments that has appeared in the post-implementation period of the mandatory IFRS standards, that were enforced in the European Union in 2005. This was done by statistically comparing data from Swedish firms with data from British firms. The findings show that there are indeed institutional differences, but the evidence also suggests a learning effect to some extent. These findings add to the literature that there are important institutional differences and learning effects associated with the implementation of accounting standards, as well as offering some insight to standard setters on the value relevance of goodwill impairments.

(4)

ABBREVIATIONS

APB

Accounting Principles Board

ASB

Accounting Standards Board

ASC

Accounting Standards Committee

EMH

Efficient Market Hypothesis

GAAP

Generally Accepted Accounting Principles

IASB

International Accounting Standards Board

IAS

International Accounting Standards

IASC

International Accounting Standards Committee

IFRS

International Financial Reporting Standards

(7)

1. Introduction

This part of the study will explain the history of goodwill impairments and amortisations. An explanation is also provided on how legislation has evolved over the years. This is followed by a problematisation about the research topic and why further research is necessary in regards to value relevance of goodwill impairments.

1.1 Regulation - past to present

The beginning of the 2000s was subject to regulatory changes in the handling of goodwill, this change meant a shift from amortisation of the goodwill to impairing it. The issue as to which way is the best for the accounting treatment of goodwill remains a controversial and interesting issue (e.g. Seetharaman, Balachandran & Saravanan, 2004; Wines, Dagwell & Windsor, 2007; Wen & Moehrle, 2016).

The accounting method that most countries applied for goodwill has historically been amortisation. Goodwill used to be amortised over a certain number of years, with some variations between different countries. For example, goodwill used to be amortised over a 40-year period in the United States (Wen & Moehrle, 2016), whereas in Europe, it was typically amortised over a shorter amount of time (Van der Zanden & Nobes, 2002), such as five years in Sweden (Hamberg & Beisland, 2014). In the UK, it could even be written off right away against the equity reserves (Seetharaman et al., 2004).

(8)

therefore impact firms’ earnings for many years to come. As a consequence of these high amortisation expenses, firms utilised the pooling method, a method where the acquiree and the acquirer would combine their recorded book values, thus eliminating any goodwill (Wen & Moehrle, 2016). However, it was difficult to qualify for the pooling method and during a time with inflated stock prices and many mergers and acquisitions going on, the FASB deemed that something had to be done. They therefore banned the use of the pooling method and lowered the maximum amortisation period to 20 years (FASB, 1999).

The United Kingdom has a slightly different history than America with the accounting of goodwill. In 1980, the Accounting Standards Committee (ASC), published a paper suggesting a shift from the generally used practice of writing off goodwill in the year of acquisition to instead amortising it over its useful life (Seetharaman et al., 2004). This led to the issuing of

Exposure Draft number 30 Accounting for Goodwill, ED30, in 1982 (ASC, 1982) which

allowed companies to either write off the goodwill in the year of acquisition or amortise it over a recommended period of 20 years or less. Two years later in 1984, the Statement of

Standard Accounting Practice No. 22, SSAP 22, was issued (ASC, 1984). This standard was

in many ways highly similar to ED30. The main difference however was that SSAP 22 recommended the method of writing off goodwill against the equity reserves. The firms at the time also preferred the use of this method, to write off goodwill directly against the equity reserves, thus reducing the amortisation cost which would then prevent it from impacting the yearly results (Seetharaman et al., 2004). In this way, SSAP 22 characterised the UK GAAP handling of goodwill in this period of time compared to for example US GAAP, where writing off goodwill right away was not permitted. In 1990, ASC was replaced by the Accounting Standards Board (ASB), and goodwill was once again further evaluated (Qasim, Haddad, AbuGhazaleh, 2013). They proposed a change from writing it down to instead amortising the goodwill over a period of 20 to 40 years, which was not appreciated by the firms and not much came out of the proposal (Seetharaman et al., 2004). The next big change to the handling of goodwill to be issued was with Financial Reporting Standard 10, Goodwill

and Intangible Assets, FRS 10, in 1998. It shifted the focus away from write-downs and

instead amortisation was emphasised which required most firms to change their policies to that of amortisation. This standard also introduced impairment of goodwill, if deemed necessary (ASB, 1998).

(9)

In Sweden, the history of dealing with the accounting of goodwill has been quite mixed over the years and the practice has not always been adherent to the law, where in the early 1990s, several firms amortised goodwill over 20 years (Thorell, 1993), despite the law stating a maximum time of 10 years (SFS 1976:125, §17). After this, goodwill was supposed to be amortised over a period of 5 years according to the Swedish accounting standard RR 1:96 (Hamberg & Beisland, 2014), unless a longer lifetime of the goodwill could be estimated (SFS 1995:1554, chapter 4 §2-4). In 2000, it was changed once again when the standard RR

1:00 was issued, where goodwill was now to be amortised over a 20-year period (RR, 2000).

1.1.2 Present regulations

FASB’s removal of the pooling method and lowering of the maximum amortisation period to 20 years from the previous 40 years was not appreciated by the American firms, and after only two years FASB issued new standards after political pressure to do so (Wen & Moehrle, 2016). These standards are the ones currently in use, i.e., Statement of Accounting Standard

141 (SFAS 141) and Statement of Accounting Standard 142 (SFAS 142). SFAS 141 only

allows for the acquisition method to be used when acquiring a company. SFAS 142, disallows the amortisation of goodwill. Instead of amortising the goodwill, an annual impairment test of the goodwill is to be conducted (FASB, 2001a; FASB, 2001b).

In 2005, as part of the European Union, the United Kingdom and Sweden both had to apply the International Financial Reporting Standards (IFRS), issued by the International

Accounting Standards Board (IASB) (Hamberg & Beisland, 2014). The relevant IFRS

standards for goodwill and goodwill impairments are: IFRS 3 Business Combinations and IAS

(10)

1.2 Problematisation

These aforementioned regulatory changes concerning goodwill may very well have led to a change in the value relevance of goodwill impairments. Previously, with the amortisation of goodwill, the balances of goodwill were kept low, whereas the value now depends on the fair value of the goodwill, which is decided by the managers, estimated on the acquired business units (Hamberg & Beisland, 2014). In this paper, the value relevance of goodwill impairment is defined in short as: how relevant the goodwill impairments are in explaining the market value of the firm.

With the mandatory adoption of IFRS within the European Union there is the question if IFRS actually improves the value relevance of the accounting information. There have been several studies which conclude that IFRS adoption has different impact depending on which country is examined (Chalmers, Clinch & Godfrey, 2011; Turki, Wali & Boujelbene, 2017). What this indicates is that standards can have different effects depending on the country in which it is implemented, since the market may react differently to information depending on the institutional environment (Barth, 2008). Therefore, the standard setters must have

knowledge of that specific country’s institutional and legal environment. Studies have also shown that a country’s legal tradition can affect its effect which makes this an important institutional setting (Kouki, 2018; Lambertides & Mazouz, 2013).

The value relevance of goodwill has been researched in various countries around the world (e.g., Dahmash, Durand & Watson, 2009; Oliveira, Rodrigues & Craig, 2010; Eloff & de Villiers, 2015; Choi & Nam, 2020). By looking at Europe, it can be determined that impairment of goodwill is of lesser value relevance in Sweden after the accounting regime shift than before (Hamberg & Beisland, 2014). After the regime shift, a negative association was found between the goodwill impairments of a firm and the firm’s market value in the UK, which suggests strong value relevance of goodwill impairment (AbuGhazaleh, Al-Hares & Haddad, 2012). The regime shift had no effect on the overall value relevance of the

intangible assets in Portugal. However, the fact that goodwill no longer was amortised but tested for impairment had a positive effect on the value relevance of goodwill (Oliveira et al., 2010). Value relevance of goodwill has also been observed in other parts of the world, for example, in South Africa, IFRS 3 showed higher value relevance of goodwill when compared

(11)

to the previously used IAS 22 (Eloff & de Villiers, 2015). Australian GAAP exercised goodwill conservatively and while it was biased, it was also value relevant (Dahmash et al., 2009). The use of IFRS in Australia, however, suggests higher value relevance of goodwill than before (Chalmers, Clinch & Godfrey, 2008). In South Korea, it was found that managers do not only delay impairments but there are some who instead accelerate it, which was positively associated with the share price (Choi & Nam, 2020).

The United Kingdom and Sweden had to implement the International Financial Reporting Standards in 2005 as both were part of the European Union (Hamberg & Beisland, 2014). The question of whether the value relevance has been affected by the switch from

amortisation plus impairment to just impairment testing has been studied before. Hamberg and Beisland (2014) studied how the value relevance of the disclosed financial information was affected by the switch from the previous legislation to IFRS 3. More precisely, the study examined how the increase in managerial discretion affected the relationship between the stock market and the disclosed accounting information in Sweden. The authors came to the conclusion that there was less value relevance after the adoption of IFRS than before the adoption. This implies that there was less or no relation between the stock prices and goodwill impairments following the adoption of the new standard. The authors of the study acknowledge that it is reasonable to assume that if the increased managerial discretion is expected to show an indication of future cash flows the opposite effect would take place. However, it might be the case that the decision to use impairment of goodwill is based on the private incentives of the managers instead of an indication of future cash flows, which would explain why there is less of a relationship after the adoption of IFRS (Ramanna & Watts, 2012). It is therefore important to understand the consequences of a regime shift as it has broader effects on the market as a whole.

(12)

The two studies that outline the basis for this paper (Hamberg & Beisland, 2014; AbuGhazaleh et al., 2012), one conducted in a Swedish setting and one the other in the United Kingdom, came to different conclusions of how value relevance is affected by the implementation of IFRS. This could point towards that the same standard might have different outcomes in different countries (Chalmers et al., 2011; Barth, 2008). The studies tried to answer how value relevance was affected in the respective countries. However, they did not examine the difference in institutional settings between different countries as they only investigated one country each. Therefore, there is a need to examine why the results of these two studies differ and if there are any institutional settings which may explain this difference.

The United Kingdom has throughout the years been known for more aggressive accounting. For example, firms within the UK have been shown to report profits of up to 25% higher than American firms of the same size (Weetman & Gray, 1990). On the contrary, Sweden has been known for a more conservative approach (Hellman, 2011). The difference in accounting approaches constitutes one way in which these two countries differ. Therefore, an

examination of how these different institutional structures might alter the result of the IFRS implementation may be instrumental in finding out how these country-specific variables alter the results of the implementation. Another reason for a comparison between Sweden and the United Kingdom is the fact that the United Kingdom adheres to the common law tradition while Sweden belongs to the civil law. Therefore, a comparison between these two countries would allow us to further examine how different legal traditions may affect institutional settings.

It has been stated that there are different aspects to an institution. First there are the formal structures such as laws and regulations but there is also an informal structure such as norms and social conventions (North, 1990). Judge, Li and Pinsker (2010) argues that increased globalisation of the capital markets has brought the accounting legislation towards

harmonisation. This is because the capital market is not limited to the domestic market but also to foreign investors, there is therefore a need to harmonise the country’s legislation. With the IFRS being implemented across the world, the national institutions play a major role in how these standards are applied in different countries. Research has also shown that national institutional factors still play a major part when predicting accounting choices in different

(13)

countries even after the IFRS adoption (Fearnley & Gray, 2015; Jeanjean & Stolowy, 2008). The results also seem to indicate that the mandatory IFRS adoption by member states of the European Union did not result in greater comparability, which has been one of the primary objectives of regulation harmonisation (Liao, Sellhorn & Skaife, 2012). It is important to examine how these national institutional factors affect value relevance since different countries have different levels of enforcement (Pope & McLeay, 2011). The problem with this variance in enforcement of the standards is according to Pope and McLeay (2011) that high quality firms operating in low enforcement countries will have a hard time signaling their value to the market through their financial reporting. Therefore, understanding how the institutional factors actually impact value relevance is crucial in order to understand how IFRS adoption will affect value relevance of the goodwill impairments.

One important factor to consider when evaluating the value relevance of goodwill

impairments is the efficient market hypothesis (EMH). Since there have been other studies on the value relevance of goodwill impairment of implementing IFRS in Sweden (Hamberg & Beisland (2014) and the United Kingdom (AbuGhazaleh et al., 2012), a new study might get the same results if similar variables are being used. The EMH would predict that a replicated study would get the same results, as the market is efficient and accurately values the financial information (Fama, 1970). However, the EMH has been put under scrutiny during the last decades as some authors argue that the markets are not efficient at all times (Malkiel, 1989; Daniel, Hirshleifer & Teoh, 2002; Borges, 2010). Instead, there are scholars who argue that there are psychological factors affecting the market valuation (Hirshleifer, 2001; Chandra, 2008; Daniel et al., 2002). If the EMH does not hold true, that might lead to misallocation of resources due to the market not acting in a rational manner (Daniel et al., 2002), which in its turn might result in a sub-optimal reaction to the new regulation regarding goodwill

(14)

regulations might have on value relevance over time, as there is already some evidence that supports it. Consequently, we want to examine if there is a learning effect in regard to goodwill impairments. Since previous studies have made some observations during previous years it is therefore interesting to further examine if there is a learning effect or if the results will be the same. If this study reaches highly similar results to the previous studies, we can probably make the assumption that the market is efficient, and no learning effect is in play. This study could also offer some insight into what practical, theoretical and societal aspects might exist between goodwill impairments and value relevance. This insight could prove valuable not only for the standard setters in how goodwill impairments are seen from the market’s viewpoint but also for investors to understand the market as well as for future research within the value relevance field.

As has been stated, a study that compares the value relevance of goodwill impairments between Sweden and the United Kingdom has not yet been conducted. Although, several studies have examined the value relevance of goodwill impairments in a specific country setting (e.g. Hamberg & Beisland, 2014; AbuGhazaleh et al, 2012; Xu, Anandarajan & Curatola, 2011; Duangploy, Shelton & Omer, 2005) none of these have examined the

differences between two institutional settings. Since Sweden is known for more conservative accounting (Hellman, 2011), while the UK is known for more aggressive accounting

(Weetman & Gray, 1990) and because Sweden is a civil law country while the United Kingdom is a common law country (Lee, 1915), the two countries are different in many regards and a comparison would be instrumental in answering how institutional settings may affect the value relevance of goodwill impairments.

1.2.1 Purpose

Since the IFRS standards are used in many countries it is important to know how the standards affect each individual country. Sweden and the United Kingdom have been presented as two opposites in many regards. They will therefore serve as two interesting countries to compare and this study will aim to assist in providing a context of how the standard works in practice and in different settings. Another aspect which this study will seek to cover is the potential learning effect of the market. Several studies have shown that in various aspects there is indeed a learning effect within the stock market (Hwang, et al, 2018; Filip et al, 2021). This opens up the possibility for the value relevance of goodwill

(15)

impairments to change as more years pass from the IFRS adoption in 2005. The practical relevance of this is primarily toward standard setters. Knowing how the standards are used in practice helps to develop new and/or improved standards in the future.

1.2.2 Research Questions

Since we have reason to expect institutional settings and learning effects to affect the value relevance of accounting information of goodwill impairments today, in the post

implementation period of the IFRS adoption, this study therefore arrives at the following research questions:

1) How do institutional settings affect the value relevance of goodwill impairments? 2) Is there a learning effect in place, which may affect the value relevance of goodwill

impairment over time?

1.3 Definitions (key terms)

To be able to familiarise oneself with the context of this paper, it is important to first have some knowledge about the different concepts explored within the paper and what they comprise. Central terms in this study include: goodwill; amortisation; impairment; value relevance; and learning effect.

Goodwill arises, and can only arise, when one firm acquires another firm. The specific

criteria as to when it is considered an acquisition is regulated in the standard IFRS 3 Business

Combinations. The goodwill is calculated by aggregating the purchasing price of the firm

(consideration), the amount of any non-controlling interest in the acquiree (if there is any) and the previously held equity interest in the acquiree (if their business combination has been

(16)

Impairment in this context refers to impairments of assets and this is regulated in the standard IAS 36 Impairment of Assets. As its core principle, IAS 36 states that an asset must not be

valued in the financial statements at more than the highest amount that can be recovered through the use of the asset or by selling it. If the amount an asset is valued at, the so-called “carrying amount”, exceeds the recoverable amount, then the asset is to be described as impaired. The carrying amount of the asset is then to be reduced to its recoverable amount and this difference is recognised as an impairment loss. In accordance with IAS 36, the recoverable amount has to be assessed each year for a certain set of assets, one of which is goodwill (IASB, 2017)

Value relevance research examines how useful accounting information is to investors. After

all, one of the main ideas of financial reporting is to aid investors in their assessment of a company’s value (Beisland, 2009). The value relevance of goodwill impairment is therefore the part that the impairments play in the eyes of investors when they estimate a company’s value. Our definition of the value relevance of goodwill impairment is linked to stock price and return, which determines if the impairments are value relevant or not.

Learning effect refers to certain actors learning from their previous behaviour as time passes

and they become more experienced. De Houwer, Barnes-Holmes & Moors (2013, s. 631) defines learning as the following “We define learning as ontogenetic adaptation - that is, as changes in the behavior of an organism that result from regularities in the environment of the organism”.In this study, the learning effect will specifically mean if there has been a change in the value relevance of goodwill impairments over time. With focus on the regulatory change to impairments with the mandatory IFRS adoption in 2005, the authors of this paper seek to examine if there has been a learning effect since the initial studies carried out shortly after the shift to IFRS (Hamberg & Beisland 2014, AbuGhazaleh et al., 2012). In other words, one of the goals of this study is to determine whether or not managers and investors have altered their behaviour in regard to goodwill impairments.

(17)

2. Relevant literature and theories

This chapter will explain previous literature and various theories which relate to goodwill impairments. This is done in order to examine the main theories and literature which might be analysed in order to understand the concept of goodwill impairments and its value relevance.

2.1 Goodwill value relevance

Previous literature on the value relevance of goodwill can largely be divided into three categories. Firstly, there are general studies done in just one country, often the home-country of the author(s) (e.g., Dahmash et al., 2009; Choi & Nam, 2020; Bugeja & Gallery, 2006; Baboukardos & Rimmel, 2014; Wyatt, 2008). Secondly, there is the research dedicated to how the impairment only regime has affected the value relevance of goodwill in countries (e.g., Ji & Lu, 2014; Wines et al., 2007; Chalmers et al., 2008; Aharony, Barniv & Falk, 2010; Horton & Serafeim, 2010; Eloff & de Villiers, 2015). Lastly, there are papers that focus specifically on the value relevance of the impairments themselves as well as the impairment losses (e.g., Xu, Anandarajan & Curatola, 2011; Duangploy, Shelton & Omer, 2005). Additionally, there is also numerous literature that is a combination of the categories above (e.g., AbuGhazaleh et al., 2012; Hamberg & Beisland, 2014; Oliveira et al., 2010; Cordazzo & Rossi, 2020), and this combination is often the most interesting, as it provides a more complete overview of the research.

(18)

this study sought to provide answers for the international debate on relevance and reliability concerning intangible assets. They also said that the goodwill tended to be reported

conservatively, while other identifiable intangible assets were reported aggressively. There are also those who have investigated how the relevance of goodwill is presented in countries outside of the Anglo-Saxon accounting regime. The shift in 2005 to the impairment only method for goodwill relies heavily on fair value accounting, which meant a fundamental change for some countries. Baboukardos and Rimmel (2014) examine how Greece, a country where the use of fair value accounting previously has been limited, handled the mandatory IFRS adoption. Using data from listed companies on the Athens Stock Exchange, they intended to determine if goodwill after the IFRS adoption carried value relevance. Their empirical findings found that goodwill was in fact value relevant to investors and that

goodwill has a positive impact on a company’s market valuation. However, they also identify a risk with the current regime. When substantial amounts of goodwill are being accumulated on balance sheets in times where prices are rising, companies are forced to recognise high impairment losses should a downturn happen, which could prove detrimental to the firm’ finances. Another non- Anglo-Saxon country is South Korea, one of the industrialised countries in the far east. Choi and Nam (2020) researched the value relevance of goodwill in the Korean market after its shift to the IFRS standards in 2011. They found goodwill to be value relevant and that the value is closely linked to when it is impaired. Firms exercising accelerated impairments exhibit a positive relation between share price and the impairment, as well as outperforming firms who exercise normally timed impairment or no impairment at all. One flaw they identify in their own research is that they do not compare the current value relevance by goodwill to that before the IFRS adoption, when amortisation was used instead of impairment. The literature mentioned so far, suggests that goodwill is value relevant. However, Bugeja and Gallery (2006) were interested in determining if the value relevance of goodwill is linked to how recently the goodwill was acquired. Their empirical findings showed that only goodwill acquired in the observation year as well as the two previous years had a positive impact on a firms’ value. The authors therefore suggest that older goodwill is not value relevant, and they hint at a future issue due to this, which is that, under the

impairment regime, where the goodwill amount grows large on balance sheets, most of it will in fact not provide useful information for investors.

The second category is about how the impairment-only legislation affected the value

(19)

was value relevant before the regime shift (Dahmash et al., 2009). However, with the

adoption of IFRS, Chalmers et al. (2008) show that goodwill is even more value relevant than previously. Ultimately conveying more information to investors, which is interesting since the two countries we will compare are Sweden, which is not an Anglo-Saxon country, and the United Kingdom which most certainly is one. Further research by Ji and Lu (2014),

contradicts Chalmers et al. (2008), saying that goodwill is not more value relevant after the IFRS adoption. Instead, Ji and Lu (2014) say that the value relevance of goodwill and other intangibles for Australian firms is closely related to how reliable the information is, both before and after the IFRS adoption. Wines et al. (2007), explored the regime shift from a more practical point of view and highlighted the complexity that this new accounting treatment of goodwill brings. It is now largely based on estimating fair value which requires auditors to rely on their professional judgment. In South Africa, Eloff and de Villiers (2015) compared the value relevance of goodwill before and after their shift to IFRS 3 from

previously applying IAS 22. The comparison was carried out by comparing the goodwill’s effect on the market value of shares, before and after the adoption of IFRS. They found a stronger association after IFRS 3 was introduced, i.e., the goodwill was now more valuable than before. In Europe, Aharony et al. (2010) investigated the value relevance of goodwill in several EU countries before and after the IFRS adoption. They found that goodwill was more value relevant after the IFRS adoption, and that countries which had local GAAP with high deviations from that of IFRS saw their value relevance increase the most. This is interesting both for a potential learning effect over time but also that there are institutional differences in play. In the UK, Horton and Serafeim (2010) analysed the differences between the old UK GAAP and the new IFRS standards. They found that the market responded worse to firms disclosing lower earnings under the new IFRS regime and that IFRS provides more value relevant information.

(20)

Duangploy et al. (2005) delves further into the topic of value relevance associated with impairment loss. Their empirical findings showed that while the amortisation was often overlooked by investors, the value relevance of goodwill impairment is not. The authors give possible explanations on why this is the case, such as an impairment loss being more

impactful on the results than amortisations or the potential threat of future write-downs which could cut the firm’s books as well as worsening its debt-to-equity ratio.

While the following literature contains a combination of the aforementioned categories, they constitute key literature in the field of goodwill value relevance in Europe. For instance, Oliveira et al. (2010) assesses the value relevance of goodwill and other intangibles in firms listed on the Portuguese Stock Exchange from the years 1998 through 2008. They also take into consideration what effect the adoption of the IFRS standards in 2005 could have on the value relevance of goodwill. Their findings show that goodwill is value relevant and highly associated with stock price. They also showed that while the shift to IFRS from Portuguese GAAP barely had an increase in the overall value relevance of intangibles, the increase in the value relevance of goodwill was evident. A similar, more recent study was carried out in Italy by Cordazzo and Rossi (2020). Their data consisted of firms from the Italian Stock Exchange, Borsa Italiana, during the years of 2000 to 2015. Their empirical results showed just like Oliveira et al. (2010) that goodwill was value relevant before the IFRS adoption, though the shift to the impairment only regime further increased the value relevance of goodwill. The two most central pieces of literature to this paper and arguably the most distinguished papers on value relevance of goodwill from their respective countries, are AbuGhazaleh et al. (2012) as well as Hamberg and Beisland (2014). AbuGhazaleh et al. (2012) investigated goodwill value relevance in the UK by focusing on the impairment losses from the top 500 UK listed firms in the years 2005-2006. Their empirical results showed that there was a negative association between goodwill impairment losses and market value, suggesting that

impairments provide value relevant information. The hypothesis they investigate is if these impairments are likely to reflect the managers’ predictions about future cash flows,

something that their results may indicate is in fact true. Hamberg and Beisland (2014)

examines the value relevance of goodwill in Sweden based on data from publicly listed firms in Sweden. Similarly to Oliveira et al. (2010), but also Cordazzo and Rossi (2020), they include several years where both the local GAAP was applied as well as the new standards, i.e., IFRS, using the years 2001-2010 in their study. Their empirical findings show that goodwill impairments in Sweden have lost relevance after the shift from Swedish GAAP to

(21)

IFRS. Not only were the impairments found to have lost value relevance, they were also found to not generally be associated with stock prices and returns under the new regime, contradicting AbuGhazaleh et al. (2012), who found a negative association between the two.

To summarise the literature on value relevance of goodwill impairments, goodwill has shown to be value relevant in many countries (e.g., Dahmash et al., 2009; Baboukardos & Rimmel, 2014; Choi & Nam, 2020). The consensus is that the current impairment-only regime of treating goodwill seems to be more value relevant than that of the amortisation method which was widely used previously (e.g., Chalmers et al., 2008; Oliveira et al., 2010; Eloff & de Villiers, 2015). Possible explanations for why this could be the case might include

managerial decisions to impair, hinting at a certain future performance (AbuGhazaleh et al., 2012) or that impairment losses are more impactful on results than amortisation (Duangploy et al., 2005). In Europe, AbuGhazaleh et al. (2012) found that goodwill is more value relevant in the UK after the IFRS adoption and Hamberg and Beisland (2014) found that in Sweden, goodwill impairments had lost value relevance after the shift to IFRS. The data used in both AbuGhazeleh et al. (2012) and Hamberg and Beisland (2014) is from two decades ago, and so is most other literature on this topic (e.g., Oliveira et al., 2010; Cordazzo & Rossi, 2020). Therefore, by using newer data, knowledge on how goodwill impairments are value relevant today can be obtained, how institutional differences affect it, as Hamberg and Beisland (2014) calls for. Concluding this section: there are differences that exist between countries that are interesting and relevant to answer the research questions of this paper concerning the institutional settings. The aspect of time and learning effects are also intriguing, as there seems to be a general tendency for goodwill to have become more value relevant over time, and this paper will take it one step further and investigate if the impairments themselves have become more value relevant over time.

(22)

2.2 Theoretical aspects of goodwill impairments

In this section an explanation of different theories which relates to the topic of goodwill will be examined, which the authors have further explored in the perspective of impairments. This has been done in order to give the user a deeper theoretical understanding of the concept.

2.2.1 Super-profit theory

Other than the previously mentioned EMH, there are several theories that have been linked to goodwill. The first theory that was adherent to goodwill was the super-profit theory,

developed in 1921, by P.D. Leake (Ratiu & Tudor, 2013). Leake defined goodwill as a right that grows out of all accumulated past effort in seeking profit, increase of value or other advantages, a definition that is not that different from the one commonly used today, a hundred years later. The theory of super-profit assumes that the acquirer of the goodwill will, as the name suggests, have considerably higher profits the following year, which would explain why goodwill was something worth pursuing. While not mentioned in their studies, this super profit-theory is to this day indirectly present in research in the form of both Hamberg and Beisland (2014) as well as AbuGhazaleh et al. (2012), Such as when they mention goodwill impairment as a way for firms to hint at future cash flows, indicating a relation between future performance and goodwill impairment. This bears similar meaning to the one developed by Leake in 1921, i.e., that a year with very high profits would follow with goodwill. This is a testament to the idea that while concepts gradually change over time, theories developed a hundred years ago cannot be disregarded.

2.2.2 Momentum theory

The next major theory of goodwill developed was the momentum theory by Nelson (1953). Nelson mentions goodwill as something quite difficult to obtain, referring to loyal customers, trade names, patents etc., and therefore firms are willing to pay hefty sums for it. The

momentum theory is built on the “push” associated with not having to start from nothing, and the tedious and challenging first stages of forming a company are skipped. Therefore, rather than the super-profits mentioned by Leake around 30 years earlier, the financial gain

proposed by Nelson with the momentum theory is instead about the momentum itself, i.e., the money saved by not having to start a new firm from scratch. This theory of momentum can

(23)

be applied today in a different yet similar matter. The equivalent of this “push” of goodwill can in current times be explained by the impairments of goodwill in two ways. Generally, goodwill impairments are seen in a negative manner as the negative impact or “push” on the results by the impairment loss, signalling a downward shift in the share price (AbuGhazaleh et al., 2012). However, goodwill impairments can also be seen as a positive force or “push” of momentum, as Choi and Nam (2020) indicate. They found that managers who exercised accelerated impairments led to a positive association between the impairments and the share price. Thus, this momentum can today swing in both directions, making goodwill research in different countries highly relevant from a theoretical perspective.

2.2.3 Contracting cost theory

Another theory that is relevant to goodwill impairment is contracting cost theory. This theory is built upon agency theory and argues that if the management compensation is closely tied to market performance, then managers will prefer the method that makes the market perform the better evaluation (Gore, Taib & Taylor., 2000). The evidence seems to point towards the idea that managers will, if given the choice, choose the goodwill accounting method that results in a higher market evaluation, thus further supporting this theory. Furthermore, research has found that as the market discovers a future goodwill impairment, there is a negative reaction (Li, Shroff, Venkataraman & Zhang, 2011). Therefore, any managerial incentives that are closely tied to the stock market performance are likely to affect managerial decisions regarding goodwill impairments. This problem is due to adverse selection, which can occur when there is information asymmetry in the market (Wilson, 1989).

(24)

2.3 Learning effects

The EMH was first coined by Eugene Fama (1970) which states that the market is efficient in using all available information available to evaluate firms. The theory suggests that stock prices always fully reflect the available information. There are different levels of efficient markets (Degutis & Novickyté, 2014). First there is the strong market efficiency where the market is valuing each firm according to its intrinsic value. Then there is the semi-strong market efficiency where valuation is done through available information as well as historical prices. At last, there is the weakly efficient market where it is only the historical prices which are taken into consideration. However, authors such as Malkiel (1989) argues that there are occasions when the market is wrong and argues that the market can make collective decisions that turn out to be false. Regardless, the author argues that in most cases, the market is in fact efficient. In recent years, a study has found mixed results when testing if the EMH actually holds true (Borges, 2010). It has also been pointed out that the EMH does not explain the occurrence of asset bubbles, volatility in stock prices or seasonality in returns (Degutis & Novickyté, 2014).

In recent years a new theory has emerged, namely the behavioural finance theory (Ritter, 2003). In contrast to the EMH, the behavioural finance theory does not believe that all investors are purely rational and that the market is efficient at all times. Instead, it takes into account that the market might be influenced by psychological factors as well as information asymmetries, which might cause markets to become inefficient (Wouters, 2006). The behavioural finance theory states that the psychological aspect might cause investors to overvalue certain information, which would therefore explain the occurence of bubbles (Ritter, 2003). According to Daniel et al. (2002), psychological biases in investors are a serious issue as it may cause misallocation of resources. The authors also found evidence of investors making errors on a systematic level due to the psychological bias and suggest that new regulations should take these findings into account in order to fix the problem.

Hirshleifer (2001) states that further research into this particular subject might provide further explanation of why the market behaves in a way that is not always rational. As the EMH builds upon the assumption that the market is always rational and acts on all available information (Fama, 1970), it would assume that it is impossible for the market to react differently today towards goodwill impairments than a couple of years ago. However, if the behavioural financial theory is correct in that the market is not always rational (Daniel et al.,

(25)

2002), then it is possible that psychological factors might cause the market to react differently today than at the time when the previous studies were performed. This in its turn might cause resources to be misallocated at times (Daniel et al., 2002). This might eventually lead to investors altering their behaviour over time which might increase market efficiency causing a learning effect. Chandra (2008) conducted a study which found that there are a number of factors which cause the capital markets to act in an irrational manner. Some of these might be psychological factors on the individual level such as fear, anchoring and greed, but there are also some behavioural factors on a collective level, such as cognitive dissonance and

heuristics.

The last subject called heuristics could be used in order to explain how the markets can learn over time. Tversky and Kahneman (1974) identified heuristics as biases that can lead to systematic errors over time in situations of uncertainty. Studies have also shown that there are certain behaviours that influence investors to make certain decisions (Bashir, Javed, Usman, Meer & Naseem, 2013; Johnson & Tellis, 2005). What this could imply, is that there are some assumptions that the market assumes to hold true, as have been found to be present in other studies (Filip et al, 2021). If these assumptions and biases were to change, a different result might follow. Another reason why the market could possibly react differently to

goodwill impairments over time is that evidence shows that the reason for the market reacting to goodwill impairments might be due to: a perceived high asymmetry of information

between investors and management; detailed information of expectations of future cash flows; and managers not providing truthful and reliable information to investors (Schatt, Doukakis, Bessieux-Ollier & Walliser, 2016). Further, Knauer and Wöhrmann (2016) found evidence of the market reacting more negatively towards goodwill impairments when there was no verifiable explanation given. They also found that where there is a low level of

(26)

also in line with what the super-profit theory suggests, i.e., goodwill impairments precede a year of high profits (Ratiu & Tudor, 2013). What this implies is that there are two sets of behaviours that may affect the market’s reaction towards goodwill impairments: 1) the behaviour of the managers and 2) the behaviour of the investors. If either of those were to alter their behaviour the relevance of goodwill impairments may change over time.

If the market is indeed efficient then the share price will fully reflect the financial information available and the market’s evaluation of certain events should not change over time but remain consistent (Beaver, 2002). However, due to the latest criticisms of the EMH (Ritter, 2003; Wouters, 2006; Degutis & Novickyté, 2014; Malkiel, 1989) and the findings that markets may not always be efficient (Borges, 2010), as well as there being different degrees of efficiency within a market (Degutis & Novickyté, 2014), it may not always the case that the market acts rationally based on all available information, which can lead to the

assumption that some resources are being misallocated at times (Daniel et al., 2002).

This may lead one to wonder if the market is efficient when it comes to the reaction of goodwill impairments and if there could be a learning effect in play. A general example of a learning effect is provided by Dye (1988) in the field of earnings management. He argues that when new legislation becomes ruling law, managers will eventually try to learn what

behaviour they can engage in regarding earnings management, in order to elude the law. Therefore, the behaviour of managers will change over time. Recent research regarding learning effects also exists in the form of Filip et al. (2021), presenting a learning effect over time in fair value accounting for banks applying IFRS 13. Filip et al. (2021) suggest that this learning effect applies to both the preparers and the capital market itself. They found that the value relevance of fair value levels increased over time following the adoption of IFRS. The authors speculate that this might be due to investors and preparers of financial information learning over time. Another example from the literature is how Hwang et al. (2018) found that there are identifiable learning effects over time with IFRS, especially regarding the fields of earnings management and value relevance accounting. They suggest that the learning effects that appear as time passes have to do with institutional factors.

With these observed learning effects, the authors of this study aim to investigate if there is a similar learning effect with goodwill impairments with regards to value relevance that has emerged since the mandatory adoption of the IFRS standards in 2005. If the investors'

(27)

heuristics biases as described by Tversky and Kahneman (1974) were to change over time due to investors making adjustments by improving their predictions. This would be in line with Malkiel (1989), who argues that the market as a collective can make the wrong

decisions. If the EMH holds true, it can be assumed that the market is efficient as suggested by Fama (1970). However, Degutis & Novickyté (2014) argues that there are different levels of market efficiency, which could open up the possibility for a learning effect. As have been demonstrated, the market can indeed learn over time as some studies have shown (FIlip et al, 2021; Hwang et al, 2018). As such, the first hypothesis is as follows:

(28)

2.4 Institutional settings

In 2005, the European Union decided that all member state companies have to adopt the IFRS framework for firms whose equity or debt are traded on a regulated market within the union (European Parliament, 2002). Institutional settings are considered to be an important variable when estimating the effect of new accounting standards according to Potter (2005). The author tries to figure out why some countries have been more successful in implementing the IFRS standards than others. What the study concluded was that there are mainly three

different kinds of institutional pressures that might cause the institutional practices to change: mimetic, coercive, and normative pressure. Mimetic pressure occurs when a country imitates a more successful country, coercive pressure occurs due to pressure from other organisations or governments, and normative pressure refers to behaviours that are being promoted in that specific institutional setting. The study found that all three of these affect the implementation of the IFRS standards. Nobes (1998) argues that in financial reporting, there are primarily two major types of institutional settings that are affecting differences between countries. Those are: strength of equity markets, and degree of cultural dominance. What can be derived from this, is that there are a wide number of institutional settings that can possibly affect value relevance of goodwill impairments.

Singleton-Green (2015) argues that the mandatory IFRS adoption within the EU has driven major positive changes, for example, greater transparency and comparability between

countries, along with other effects. However, the paper also concludes that the effect that the IFRS adoption has had on various countries differs wildly. It depends in part on how the national institutions work which causes the effect of IFRS adoption to differ between the countries. Singleton-Green claims that the effect of regulatory reforms may in part be judged by the level of institutional reforms. One reason why IFRS adoption is even possible in a wide variety of countries is because it is based on a principle-based system and not a rules-based system (Carmona & Trombetta, 2008). A rule-rules-based system requires strict adherence to specific rules and regulations while a principle-based system simply provides the professional with certain principles that should be followed when reporting.

Institutional settings are relevant when measuring the effect of goodwill impairments on an international level since previous research has shown that IFRS adoption can vary

(29)

found that the value relevance of the accounting numbers decreased after the implementation of IFRS within German firms. The authors argued that this occurred mainly due to

differences in institutional settings such as market and enforcement systems. However, a study performed on Australian firms saw the opposite result (Chalmers et al., 2011), meaning that value relevance of earnings actually increased following IFRS adoption.

As previously mentioned, Aharony et al. (2010) found that goodwill was generally more value relevant in Europe after the IFRS adoption and that countries which previously had legislation with bigger differences from IFRS increased the value relevance the most. This is interesting for H1 with regards to time, but even more interesting in the sense of institutional settings, since the research by Aharony et al. (2010) explains by other words that there are institutional differences in play. Therefore, it is interesting to see if there are institutional differences when it comes to the impairment of goodwill by comparing two countries.

Nobes (1998) argues that a country’s financial system is one of the primary drivers of its financial reporting systems, and unless two countries are culturally homogenous to each other, their financial reporting system and practices will differ. However, there have also been signs that when countries are mandated to implement the IFRS standards, the

institutional differences between countries tend to disappear (Andre, Filip & Paugam, 2012). It has also been shown that prior to IFRS implementation, European firms tend to be more conservative when it comes to goodwill accounting (Hamberg & Beisland, 2014). Since the UK is known for a more aggressive form of accounting (Weetman & Gray, 1990) and Sweden is known for a more conservative approach (Hellman, 2011), it is reasonable to assume that institutional differences will more likely be observable in comparing these two countries. Even though the difference in institutional settings might have eroded following

(30)

upon legislation but primarily on case law (Pejovic, 2001). The civil law tradition is

structured around general rules and principles. The job of the courts is to interpret the law and where there are certain gaps, the courts should then apply the general rules and principles in order to reach a decision. The common law, on the other hand, which originates from 11th century England and has since spread to the United States, Canada, and Australia, operates primarily on case precedent. The verdicts of the higher courts are intended to be followed by the lower courts. There are also so-called “statues” that are interpreted by the courts. La Porta, Lopez-de-Silanes, Shleifer and Vishny (1998) found that the investor protections are stronger in common law countries than in civil law countries. The study found that the protections for investors in Scandinavian civil law countries are positioned in the middle out of the various countries examined. The reason why this might be relevant for institutional settings is that studies have shown that countries’ legal tradition is a factor in earnings management (Kouki, 2018), which in turn might affect how relevant investors find goodwill impairments to be. Other studies have also found that the equity cost of capital following IFRS adoption will require a higher risk premium in order to invest, and that countries with a high level of investor protection will experience reductions in the cost of capital following IFRS adoption (Lambertides & Mazouz, 2013). This shows that the different legal traditions can also affect the effects of the IFRS implementation. It is therefore highly relevant to further assess the specific differences between the United Kingdom (common law) and Sweden (civil law). Understanding the differences between these two countries will

contribute further to understanding the deeper differences between countries in various legal traditions. Thus, to explore this issue in the context of goodwill impairments, especially as we have reason to believe the goodwill impairments to be more value relevant in the UK than in Sweden and to further clarify the differences between the legal traditions of these two countries and institutions. Our second hypothesis is therefore the following:

Hypothesis 2 (H2): Goodwill impairments are more value relevant in the United Kingdom

(31)

3. Methodology

In this section a detailed description of the study’s research method will be presented. The authors of the study further explain why and what methodology is used. The process of data collection is further explained, and a description of the sample is provided. The various variables and how these have been calculated are also described.

3.1 Methodology Introduction

This study has been conducted while using a positivistic approach. According to O’Gorman and MacIntosh (2014), a positivistic approach assumes that the procedures of natural science can be applied to studies of human behaviours. This is true of this study since we conduct a quantitative study with the goal of gathering data in order to reach a conclusion.

In this study, the value relevance of goodwill impairments following the mandatory adoption of IFRS within the European Union will be examined. More precisely, this study will

investigate if the value relevance of goodwill impairments changes over time or if it remains constant (hypothesis 1) as well as if the value relevance is greater in the UK than it is in Sweden (hypothesis 2). The first hypothesis will be tested by simply performing similar tests as those of the two previous studies (Hamberg & Beisland, 2014; AbuGhazaleh et al., 2012). Apart from this the sample was also divided into two different periods in order to test if a difference can be found between the earlier part and the latter part of the sample. If the results are the same as these previous studies, as well as there being no difference between the two different periods of this study, then the conclusion can be made that there is no learning effect. The second hypothesis will test how institutional settings affect the outcome by

(32)

conclusion. All the data in this paper has been collected from secondary sources. More precisely, the data has been collected from Datastream, which in turn has collected the data from the annual reports of each firm.

The methodology used in this study can be summarised as the following: There will be two different kinds of models and four models used in total, which will replicate the models used by Hamberg and Beisland (2014) as closely as possible. These are two price models and two return models. The model used by AbuGhazaleh et al. (2012) is essentially the same as the one used by Hamberg and Beisland (2014) though a bit simplified as stock returns are not included in the model of AbuGhazaleh et al. (2012) and their model is essentially the same as the price model used by Hamberg. We therefore opted for the models of Hamberg and

Beisland (2014) to provide more depth in analysing the different results.

As the previous studies indicated that goodwill impairments in Sweden were not value relevant but were relevant in the UK. This therefore gives us cause to expect that the value relevance in the UK will still be present for impairments. We also expect results indicating that goodwill impairments are value relevant for Sweden as well, since we believe that there has been a learning effect associated with IFRS, as research have shown (Filip et al. 2021; Hwang et al., 2018). Our expectations are therefore in contrast with Hamberg and Beislands (2014) findings. They did not examine the learning effect in their study, since the study was conducted in the most recent years following IFRS adoption, where the learning effect could not yet be observed. However, we do expect that the value relevance is stronger in the UK than in Sweden since this is in line with previous findings and we have no reason to believe that this relationship has changed. It is also important to note that the value relevance for goodwill impairments is supposed to show a negative association between goodwill

impairments and the dependent variable (stock returns or stock price). If the association is not negative, the relationship can technically be considered to be value relevant, but as this fails to meet basic standards of logic, this would indicate that higher impairments leads to higher stock price or returns. Taylor (1990) argues that it is a fallacy to assume that a statistically significant result can prove the research hypothesis if this result goes against the logical and expected result. Therefore, we are exclusively interested in a negative association between the two variables.

It is also important to note that it is not the purpose of this study to find and explain the various institutional settings which may affect the difference between the two countries.

(33)

Instead, the purpose of the study is to find if there are substantial differences between the two countries, but in order to conclude what specific institutional differences causes this

relationship, further research is necessary.

3.2 Data sample

The data selection is based on the data needed in order to answer the hypotheses and purpose stated within this paper. One part of this study is to examine if a learning effect has been present in the market over time, thus, the data used starts where previous literature ends. We have identified Hamberg and Beisland (2014) as the main source of literature in Sweden dealing with the value relevance of goodwill. Their empirical results are based on data from the years 2001 to 2010, and as such, our selection of years cover the available data after this period. The years 2011 to 2019 are therefore the natural sample period for this paper. Data from the year of 2010 is however also included but only in order to see the change for the year 2011. AbuGhazaleh et al. (2012), who wrote about the value relevance of goodwill in the UK used data from 2005 and 2006, therefore, collecting newer data will prove useful in order to examine any change over time as well as compared with the Swedish data.

3.2.1 Swedish Data

In collecting the data, the database Datastream was used. For the purpose of this study, firms that were listed on the Stockholm Stock Exchange and that were based in Sweden were selected. They also had to include data from the entire sample period. After excluding all other firms, 345 firms remained. A further 18 firms that belong in the financial sector, where reporting processes tend to differ from other industries (AbuGhazaleh et al., 2012), were then excluded. After these steps, 327 firms remained. Out of these 327 firms, 228 firms that did not have any reported goodwill impairments in the ten years of our sample period were

(34)

3.2.2 UK Data

The collection process of the UK data was similar to that of the Swedish data; Datastream was again used to gather the data. The London Stock Exchange (LSE) was selected, and after excluding firms that did not have data from the sampling period or were not based in the UK, 846 firms remained. 72 firms were then excluded due to belonging to the financial sector, with 774 firms remaining. Due to the large numbers of firms on the LSE, the remaining firms were sorted by size, based on total revenue over the sampling period, thus keeping the 515 largest out of the 774 firms. 15 out of these 515 were then removed after finding that numbers for several years were missing. We manually investigated why these numbers were missing and we found that there was no published annual report for one or more years. Therefore, a decision was made to remove these corporations in order to gather more complete data. This meant that 500 firms remained and these were then limited down to the 222 firms that had reported goodwill impairments in the years of the sample period. A further four firms were excluded due to reporting in another currency than the UK pound and an additional two were excluded as they had dual shares similar to that of some of the Swedish firms. This selection process produced a final sample of 216 firms or 2,160 firm year observations.

3.2.3 Outliers

Before the tests were conducted, the sample was also adjusted for outliers. Several outliers that were so severe that they were likely to have a high impact on the results of the tests were discovered. Therefore, a decision was made to adjust for extreme outliers by removing the top 1% of observations which were the most deviating. First, the 1% of firm-year

observations were trimmed with respect to each dependent variable (price and return variable). This implied removing nine firm-year observations from the Swedish sample and 22 firm-year observations from the UK sample. The outliers varied depending on the dependent variable that was used, therefore, the removed outliers differ to some extent between the two different types of models. After removing the outliers, a total of 2,138 firm-year observations remained for the UK and 841 firm-firm-year observations remained for the Swedish sample.

(35)

3.3 Models and Measures

As this study will utilise the models used by Hamberg and Beisland (2014) to the highest degree possible, this study will attempt to replicate the previous study in order to best

compare the findings. Beginning with Hamberg and Beisland (2014), this study examined the value relevance on the Swedish stock market following IFRS adoption in 2005. The study used four different models in order to reach a conclusion: two price models and two return models. The foundation behind the return- and price model is found in the Ohlson/RIV model, which was developed in order to measure how market value is affected by accounting information (Ohlson, 1995). This is currently the reference model which is most commonly used when measuring value relevance of accounting information and its effect on market value. Ota (2003) argues that both models have their own weaknesses. For example, the return model has a problem with accounting recognition lag. This refers to the fact that when an event occurs, it is not necessarily recognised by users of the financial information right away and therefore the returns for a specific period might not reflect the accounting

information for the same period. The problem with the price model according to the author is the scale effect. This scale effect refers to the phenomena that large firms will also have large accounting numbers which might lead to the results only showing the size difference between the firms, and not if the accounting information is value relevant or not. Although, Kothari and Zimmerman (1995) determined that the price model is the least biased of the different models, both models were used by Hamberg and Beisland (2014) and therefore, both of these will also be used in this study in order to be able to compare our results.

The price models have a company’s stock price (PRICE) as the dependent variable and have the following independent variables: earnings (EARN), earnings less goodwill impairments

(36)

have been excluded is simply because their study includes results from both before and after IFRS implementation, whereas, this study focuses exclusively on post-IFRS implementation. Before the IFRS implementation, companies could choose between amortisation and

impairments, and currently, impairments are the only option. Therefore, the amortisation related variables are unnecessary. The reason for excluding the dummy variable is that it is simply not relevant for this study. While Hamberg and Beisland (2014) used this variable in order to compare observations with firms exclusively with positive net results, this is not considered to be within our scope and will not contribute towards our hypotheses.

The return models have stock return (RET) as the dependent variable and have the following independent variables: goodwill impairments (GWIMP), earnings less goodwill impairments (EARN-GWIMP), earnings (EARN), changes in earnings less goodwill impairments

(ΔEARN-GWIMP), changes in goodwill impairments (ΔGWIMP), and changes in earnings (ΔEARN). Once again, all of the variables are not present in both return models. In R1, EARN and ΔEARN are being tested and in R2 GWIMP, ΔGWIMP, EARN-GWIMP and ΔEARN-GWIMP are included.

3.3.1 Variables

Dependent variables

𝑅𝐸𝑇(𝑖,𝑡) Stock return of firm i, period t

𝑃𝑅𝐼𝐶𝐸_(𝑖,𝑡) Stock price of firm i, period t.

Independent variables

𝐺𝑊𝐼𝑀𝑃(𝑖,𝑡) The amount of goodwill impairment by firm i, period t.

(𝐸𝐴𝑅𝑁 − 𝐺𝑊𝐼𝑀𝑃)_(𝑖,𝑡) Earnings reduced by goodwill impairments by firm i, period t. 𝐸𝐴𝑅𝑁(𝑖,𝑡) The net earnings for firm i in year t.

𝛥(𝐸𝐴𝑅𝑁 − 𝐺𝑊𝐼𝑀𝑃)_(𝑖,𝑡) Changes in earnings reduced by goodwill impairments by firm i, period t. 𝐺𝑊𝐼𝑀𝑃(𝑖,𝑡) Goodwill impairment in the balance sheet of firm i, from period t-1 to t.

𝛥𝐺𝑊𝐼𝑀𝑃_(𝑖,𝑡) Changes in the amount of goodwill impairment in the balance sheet of firm i, from period t-1 to t.

𝛥𝐸𝐴𝑅𝑁_(𝑖,𝑡) Changes in earnings recognised between years of firm i, from period t-1 to t. 𝐵𝑉𝐸(𝑖,𝑡) Book value of equity of firm i, period t.

(37)

(𝐵𝑉𝐸 − 𝐺𝑊)(𝑖,𝑡) Book value of equity excluding goodwill for firm i, period t.

𝐺𝑊_(𝑖,𝑡) Goodwill balance for firm i, period t.

When gathering information about the earnings during the period, the company’s EBITDA was used since it has been considered as an accurate tool in predicting market values (Nissim, 2019). EBITA was considered to be the most accurate measurement but since one of our variables includes earnings subtracted by goodwill impairments, we considered it appropriate to select EBITDA, since depreciation has not yet been subtracted.

In calculating the change from the previous year, the following formulas has been used:

𝛥𝐸𝐴𝑅𝑁 = (𝐸𝐴𝑅𝑁_𝑡− 𝐸𝐴𝑅𝑁_𝑡−1)/𝑃𝑅𝐼𝐶𝐸_𝑡−1 EQ1

𝛥𝐺𝑊𝐼𝑀𝑃 = (𝐺𝑊𝐼𝑀𝑃𝑡− 𝐺𝑊𝐼𝑀𝑃𝑡−1)/𝑃𝑅𝐼𝐶𝐸𝑡−1 EQ2

Since the value relevance of goodwill impairments following the adoption of IFRS in 2005 will be examined, it is important that the variables have close resemblance with those used by Hamberg and Beisland (2014). If many of the same variables are used, the results of this study can be compared with those of their study.

In this study, two different kinds of tests will be performed. As Kothari & Zimmerman (1995) state, there are mainly two different ways of measuring value relevance. First, there are the price models and the return models. The authors came to the conclusion that the price models are the most unbiased of the two models, but it is also true that return models are commonly preferred rather than the price models. Since this is the case, and because

Hamberg and Beisland (2014) have used both these types of models, an effort to compare the results of this study will be made as well.