Earnings Management pre- and post IFRS Adoption: Results from Sweden, Finland and Norway

Earnings Management pre- and post IFRS Adoption -

Results from Sweden, Finland and Norway

Master’s Thesis 30 credits

Department of Business Studies Uppsala University

Spring Semester of 2017

Date of Submission: 2017-05-30

Victor Dahlén Daniel Lindberg

Supervisor: Cecilia Lindholm

This paper examines the behaviour and use of accruals management (AM) and real activities manipulation (RAM) under local Generally Accepted Accounting Principles (GAAP) and International Financial Reporting Standards (IFRS) and if it has been altered following IFRS implementation in Sweden, Finland and Norway. The paper takes inspiration from Roychowdhury (2006) and Zang (2012) and use previously de- veloped frameworks for earnings management. It provides empirical results with data from 1997 to 2016, focusing on companies around zero earnings and zero earnings growth. The results are mixed where the use of RAM through production suggests in- creased earnings manipulation. RAM, through discretionary expenditures, on the other hand are positive, which suggests that companies do not engage in these activities. The results regarding AM suggest downward adjustments. However, neither discretionary expenditures nor AM have a significant change following IFRS implementation. Over- all, earnings management behavior in the sample appears to be limited. Although, increased RAM has been found regarding production costs, which suggest adjustments following IFRS adoption. Lastly the paper finds that studying one type of earnings management behavior, as often previously done in research, is insufficient in order to fully estimate earnings management.

Keywords. Earnings management; Accruals management; Real activities manipula-

tion; IFRS implementation; Information asymmetry

We would like to express our gratitude to our supervisor Cecilia Lindholm for the continuous feedback and guidance that she assisted with during the writing of this thesis. We would also like to thank our peer reviewers who gave us valuable comments and constructive criticism.

May 30, 2017

Victor Dahl´en & Daniel Lindberg

1 Introduction 1

1.1 Information Asymmetry . . . . 3

1.2 Managing Earnings . . . . 4

1.3 Summary . . . . 5

2 Aim & Research Questions 7 2.1 Contribution . . . . 7

3 Prior Research and Related Literature 8 3.1 Information Asymmetry and Value Relevance . . . . 8

3.2 Earnings Management Incentives: Agency Theory . . . . 10

3.3 Definition and Approaches to Earnings Management . . . . 11

3.3.1 Accruals Based Earnings Management . . . . 12

3.3.2 Real Activities Manipulation . . . . 13

3.3.3 The Cost of Accruals Management and Real Activities Manipulation . . . 15

3.4 IFRS and Earnings Management . . . . 15

3.5 Estimating Earnings Management . . . . 17

3.6 Summary of Literature and Formulation of Hypotheses . . . . 23

3.6.1 Hypotheses . . . . 24

4 Method 25 4.1 Research Approach . . . . 25

4.2 Literature Review . . . . 27

4.3 Sample . . . . 28

4.4 Data Collection . . . . 29

4.5 Industry Categorization . . . . 30

4.6 Statistical Assumptions . . . . 30

4.7 Suspected Earnings Management . . . . 31

4.8 Estimation Models . . . . 32

4.8.1 Accruals Management . . . . 32

4.8.2 Real Activities Manipulation . . . . 34

4.9 Fama-Macbeth Methodology . . . . 35

4.10 Method Summary . . . . 36

6 Discussion 42

6.1 Method Discussion: Limitations and Strengths . . . . 46

6.1.1 Criticism: Estimation Models . . . . 46

6.1.2 Reliability and Replicability . . . . 48

6.2 Validity . . . . 48

7 Conclusion 49

References 50

Appendices 56

A Description of Variables 56

B Literature Review Process (Adapted from Prisma Guidelines) 57

C Literature Review 58

1 Models to Measure Earnings Management . . . . 18

2 Literature Review - Relevant Papers . . . . 21

3 Sample of Companies - Number of Industry Years per Sector under IFRS and local GAAP . . . . 30

4 Summary of Models to Estimate Normal Levels . . . . 38

6 Comparison Suspect Firm Years Against Rest of the Sample . . . . 40

9 Literature Review - Included Papers . . . . 58

Figures 1 Structure of Study . . . . 6

2 Possible Scenarios . . . . 23

3 Flowchart Method . . . . 37

4 Literature Review Process . . . . 57

1 Introduction

This study examines earnings management in the light of two strands, the much researched accruals management (further denoted AM) and the newer area of real activities manipulation (further de- noted RAM). AM relates to adjustments within accounting, specifically areas where manager judg- ment is vast, which for instance are timeliness of recognition, impairment decisions and provisions of bad debt (Trejo-Pech, Weldon, & Gunderson, 2016). RAM on the other hand, is operational activities where managers depart from normal business practice to reach their targets, which for instance can be unmotivated price-cuts, less funds for R&D and overproduction (Roychowdhury, 2006). The paper is done in the context of IFRS adoption, which provides an environment where one accounting standard is replaced by another. The change of accounting standard is an interest- ing context since it, in some way, should render an actual or believed change in accounting practice that could have spurred a change in earnings management behavior. Callao and Jarne (2010), Gray, Kang, Zhiwei, and Qingliang (2015) and Doukakis (2014), who all reviewed earnings man- agement before and after IFRS in Europe, contribute to this study by concluding that earnings management is present in Sweden, Finland and Norway to some extent following the transition to IFRS, which provides the study with a setting. The study takes inspiration from mainly three stud- ies. Roychowdhury (2006), Zang (2012) and Dechow, Sloan, and Sweeney (1995) contribute with frameworks for examining RAM and AM respectively. Zang (2012) also provides a view of earn- ings management where RAM and AM are considered a trade-off. Further, it should be mentioned that the chosen countries are not studied separately, but rather as a cohesive sample. Mentioned later, ideally one country should be used when performing these tests, however one, or even two, of these countries would provide an insufficient sample size.

Earnings is a measure that provides an indication of the inherent value transformation of a com- pany, in essence how well a company transform their assets, knowledge, experience and expertise into monetary values. It is an estimate that grants owners and stakeholders an indication of the company’s ability to generate value for the shareholders, which means that it is a principal point in determining share price (Cotter, 2009). Hence, an adjustment of earnings could change the view of a company and how it performs. Therefore, not surprisingly, Burgstahler and Dichev (1997) argues that earnings is the single most important measure for managers and therefore represents a focal point. Burgstahler and Dichev (1997) continues by stating that consistent earnings with a slight upward trend is a desirable pattern for managers, in other words surprises are not suitable.

DeAngelo, DeAngelo, and Skinner (1996) argues that, on average, a company interrupting their

upward trend could expect a 14% negative return the same year as the interruption. One could

therefore suggest that there are a lot of incentive to keep earnings positive since a negative message to the market could have sincere ramifications on the company’s performance in the stock market.

Hence, earnings is an essential part of financial reporting, it is a measure that communicate to own- ers, stakeholders and the market how well the company is managed and performs, which in turn has a profound impact on the company’s share price. Therefore, the incentives to keep earnings within the ”correct” interval is great, keeping in mind the downside.

A prime example of earnings management is, the now defunct company, Enron who managed earnings in an effort to keep up appearances with the end goal of a strong share price (Gladwell, 2007). Gladwell (2007) raises an interesting point, mainly that while Enron surely manipulated earnings to a large extent, all information of their fallacious actions were presented in the company’s financial report, it was simply an issue of interpreting the information that was there. This would suggest that stakeholders had the information but were unable to understand it and realize the company’s complete financial situation. This implies that management had more information than present and potential owners. Hence, one could therefore argue that by manipulating earnings an asymmetry between a company, its owners and its stakeholders grows. The presented financial information should represent the economic reality of the company. Surely, in the case of Enron, the information was there but bundled with a lot of other information, which suggest that stakeholders focus on the earnings figure that management adjusted as to keep the upward trend. This can be related to agency theory and the principal/agent issue, mainly that the management’s custodian objectives was not aligned with that of the owners due to the inherent opportunism of management (Jensen & Meckling, 1976). According to Gladwell (2007), the earnings management done by Enron was not illegal but rather unethical by presenting the company in a brighter light than what the situation actually was. This leads one to the question of what is permitted, in essence how the companies are compelled to present earnings and what adjustments they are allowed to do.

Therefore, one important factor could be what the current regulation, or accounting standards, allow. This begs the question if they could and would have manipulated earnings if the accounting standards had not allowed it, or if they had found alternative ways to do it.

In Europe, a new environment relating to earnings management has emerged since the countries within the union have had to abandon their local generally accepted accounting principles (further denoted local GAAP) in favor of the global standard ’International Financial Reporting Standard’

(further denoted IFRS) with regards to listed companies. This was done in an effort to enhance

quality, comparability and convergence (Barth, Landsman, Danqin, & Zhuang, 2014). Hence, in

this context the question posed above becomes whether earnings management have increased or

decreased with the entry of IFRS and if it has spurred any alternative ways to conduct earnings

management. This study examines how earnings management behaviors have changed in the con- text of IFRS adoption in Sweden, Finland and Norway. According to Zhang, Uchida, and Bu (2013), the incidence of earnings management is to a large extent dependent on accounting stan- dards, which in turn is why this study use the context of a change in accounting regulation that the IFRS adoption contributed with.

1.1 Information Asymmetry

Burgstahler and Dichev (1997) emphasize the importance of the earnings measure in business to- day, how managers continue to highlight the attention to earnings throughout. How does this vast earnings focus affect the information provided to stakeholders and owners? By modifying the earnings measures, management is manipulating the view that is put forward to owners and stake- holders. In essence they are depicting a picture that suits their, or the company’s, needs better. An important factor to note in this situation is that management often have more information concern- ing the shape of the company compared to the others. Hence, an information asymmetry is present between management and present and potential owners. This relates to the agent/principle conun- drum, which concerns the separation of ownership and control, explained by Fama and Jensen (1983) and Jensen and Meckling (1976). One way to decrease information asymmetry and remedy the principal/agent problem is the publication of financial statements. In essence, owners can con- trol that management work towards their goal by monitoring them, one factor in this monitoring is the financial information that management provide through financial reports, such as the annual and quarterly reports (Fama & Jensen, 1983). However, what happens when these reports are not reflecting the performance of the business but rather the picture that management wants to depict?

An asymmetry arises between management and present and potential owners and stakeholders.

Hence, one could question if those reports communicate relevant information. This in turn relates

to another factor within information asymmetry, namely value relevance. Value relevance is a part

of information asymmetry that explores how well information provided by companies depict the

financial state of the company (Kargin, 2013). Hence, the information asymmetry apparent implies

that it is not perfect information. Perfect information (no information asymmetry), which is a sit-

uation where owners and stakeholders have the same knowledge of the company as management,

implies a perfect value relevance. However, this is surely not commonplace. An interesting point is

provided by Jiraporn, Miller, Yoon, and Kim (2006) who discuss if earnings management actually

provides more relevant information to the market. Since accounting is a set of rules that all compa-

nies have to follow, regardless of what industry they are situated in or in what region they operate, it

is not adapted to the specific circumstances surrounding each type of company. This means that it is more suited to some companies than other. Then one could question if, by managing earnings, they present a more suited picture of the company and if that does not actually decrease the information asymmetry?

1.2 Managing Earnings

As explained earlier, earnings management relates to situations where managers manipulate earn- ings in order to provide more or less information to owners and stakeholders. Earnings management is a well debated subject and it has become a popular topic after several corporate scandals. Accord- ing to Beneish (2001) there are several definitions of earnings management, generally it appears to represent a situation when managers alter earnings by adjusting financial figures. The end goal of these can in some instances be to mislead investors. It should be mentioned that in some situations managers provide more relevant information by adjusting earnings. Nevertheless, more often than not it seems to be the prior. A popular choice among academics have been to study the discretionary and non-discretionary accruals in order to arrive at a conclusion as to whether the earnings has been altered due to opportunistic behavior or not (Callao & Jarne, 2010; Mostafa, 2017; Jones, 1991;

Dechow et al., 1995). However, this approach only takes earnings management within the realm of adjusting various accounting measures, for instance timeliness of recognition, provisions and im- pairments. These represent pure accounting decisions and not management decisions. According to Roychowdhury (2006), management decisions is another way to manage earnings. These deci- sions could for example be related to funding a project or seasonally cutting down expenditures.

Out of context they could be regarded as day-to-day decisions that are impartial and necessary for continued operations, however if they are related to opportunistic behavior they should be viewed as a way to manage earnings. This could have similar effects on the overall earnings figure as the discretionary alternatives, albeit not as immediate. This way of adjusting earnings are referred to as ”real activities manipulation” (Roychowdhury, 2006).

Since IFRS relies a lot on managerial choice and, in some cases, subjective measurement, some argue that earnings management have increased as a result of the change (Callao & Jarne, 2010).

Therefore, in the context of IFRS adoption, discretionary earnings management could be done by

altering accounting decisions related to measurement and choice. However, the RAM should not

be affected in the same way as the discretionary approach since it does not relate to an intrinsic

financial accounting choice per se, but rather to a managerial accounting choice, i.e. whether to

cut prices or the designation of funds. Therefore, a change to IFRS could have an impact on dis-

cretionary accruals through intrinsic financial accounting choices and not on RAM or the opposite.

However, the relation between the two strategies of earnings management could be affected by a change in choice regarding choosing one method over the other since the advantages related to the different methods might change due to the change from local GAAP to IFRS. This in turn could affect the economic picture presented to the market, meaning that the information asymme- try could be affected depending on what kind of earnings management strategy the given company employs.

1.3 Summary

Roychowdhury (2006) stated that there are mainly two classifications of earnings management; (1)

AM and (2) RAM. This paper studies the prevalence of these two strategies in the context of the

transition from local GAAP to IFRS. Callao and Jarne (2010), Gray et al. (2015) and Doukakis

(2014) have concluded that earnings management is present in Sweden, Finland and Norway af-

ter the transition to IFRS to some extent. However, they only screened for AM behavior. Since

earnings management is not an occurrence that, most likely, will not disappear due to the change,

this study reviews if the change of accounting standard have altered earnings management behav-

ior. The change could have spurred an increase or decrease in one or the other strategy, which

in turn could mean an altered information asymmetry depending on the result. A flowchart, pre-

sented below as figure 1, has been constructed in an effort to explain the relation that the study

assumes.

Figure 1: Structure of Study

The flowchart describes the positioning of the study. First, the figure explains that agency theory can be viewed as the overarching theory, which the study assumes affects information asymmetry.

Further, the study presumes that this should be affected by earnings management, which is depicted

by the box to the right. This is the focus of the study. The box located to the left displays the

transition from local GAAP to IFRS, which should mean an altered situation regarding earnings

management since accounting practice between local GAAP could differ from IFRS. This is the

context of the study. One view of this transition is provided by Doukakis (2014), who concluded

that the quality of accounting should increase if management discretion decreased. Therefore, if

IFRS reduces the amount of management choice compared to local GAAP, earnings management

behavior should be altered. This in turn should mean a changed relationship between AM and RAM

since the ability to use AM strategies largely depends on the amount of discretionary space for the

managers. Further, since strategies regarding AM are more visible compared to RAM (Doukakis,

2014), an altered relationship between these two could impact information asymmetry, which in

turn can affect value relevance.

2 Aim & Research Questions

The study is based on previous knowledge, deduced from other contexts, regarding earnings man- agement and the transition from local GAAP to IFRS. The paper studies earnings management behavior in Sweden, Finland and Norway. Further, the study use previously developed estimation models to test the occurrence of AM and RAM before and after IFRS implementation. Hence, the objective is not to create new theory, merely use previous literature to test if theory deduced in other contexts captures the Swedish, Finnish and Norwegian examples as well. The aim is therefore to examine the behaviour and use of AM and RAM under local GAAP and IFRS and if it has been altered following the IFRS implementation in Sweden, Finland and Norway. Hence, the research questions are phrased as follows:

Was earnings management present during 1997-2016 in the form of RAM and AM in Sweden, Finland and Norway?

Did the level of RAM and AM change following the move from local GAAP to IFRS in Sweden, Finland and Norway?

2.1 Contribution

Previous research within earnings management has, too a large extent, focused the adjustments of accruals. There are a multitude of different models that try to estimate AM. However, according to Fields, Lyz, and Vincent (2001), in order to fully get to grasp with earnings management within a given population one has to estimate both AM and RAM. Therefore, this paper will study earn- ings management behavior with regards to both types of earnings management strategies. AM is estimated through a model originally constructed by Jones (1991), later modified by Dechow et al.

(1995), that singles out discretionary accruals from industry-averages. RAM focuses on operational activities such as reducing expenses or offering large discounts during a certain period in order to increase earnings, and are estimated by comparing RAM to industry-averages. (Roychowdhury, 2006). The context is before and after IFRS adoption in the Sweden, Finland and Norway.

Defond (2010) reviews the research within earnings management and argues that earnings man-

agement is a function that is used to a large extent and that manipulating real activities such as

the expenditures regarding R&D is a strategy that has become part of ”business as usual”, almost

plain. However, despite the wide use of this strategy, research regarding RAM is scarce. The large

majority of research regarding earnings management is within the realm of AM. Defond (2010)

therefore argues for further research within RAM since it is an element in business which have not been thoroughly scrutinized. Roychowdhury (2006) further argues that RAM might be of greater damage to a firm than AM. Studies also show that financial executives prefer to manipulate earn- ings through RAM rather than AM. This could be due to auditors focusing more on purely financial adjustments, in other words AM, than RAM, since those activities are more in line with managerial accounting, such as pricing and production decisions (Roychowdhury, 2006).

As mentioned, this study focuses on earnings management before and after IFRS adoption. This is done in an effort to be able to determine if the incidence of earnings management has increased as a part of the adoption, and if it manifests itself in a different manner in Sweden, Finland and Norway.

Therefore, this study contributes with an understanding of the use and manifestation of earnings management in Sweden, Finland and Norway before and after IFRS adoption, and discuss if it has incurred a change in the information asymmetry between companies and the market. This could be useful for the participants of the financial markets since it both provides insight into how extensive the use of earnings management is and what earnings management strategies are used.

3 Prior Research and Related Literature

Earnings management is a subject that relates to a variety of topics within the realm of information asymmetry, finance and accounting. First, there is the study of why earnings management exist, in essence the incentives for earnings management. Further, there is also the question of how it is performed and what the different strategies are that can be used. This also relates to incentives, what spurs a certain type of behavior. In this paper, since Callao and Jarne (2010), Gray et al.

(2015) and (Doukakis, 2014) have concluded that earnings management is present to some extent in Sweden, Finland and Norway with regards to AM, the focus is on earnings management behavior.

Therefore, previous literature regarding AM and RAM is approached in an effort to gather previous findings related to the subject. Further, information asymmetry and agency theory are also explored in order to approach what kind of reasons and incentives that play a part in earnings management behavior.

3.1 Information Asymmetry and Value Relevance

As mentioned previously, opportunistic management adjust earnings in a way that often is not up

to par with the objectives of the owners. This creates a divide between the principal and the agent

which increases information asymmetry (Callao & Jarne, 2010). Richardson (2000) and Jiraporn et al. (2006), who both studied the relationship between information asymmetry and earnings man- agement, claims that there are two ways to view earnings management and information asymmetry.

First, one could view earnings management as a factor that implies a higher information asymme- try. Second, information asymmetry can be seen as an enabler to increased earnings management.

This means that an environment with high information asymmetry facilitates a higher occurrence of earnings management since the ability for owners and stakeholders to monitor management are more difficult. These findings concur with what Mostafa (2017) concludes. He investigated the re- lationship between earnings management and value relevance of earnings and found that earnings management affects value relevance of earnings negatively. Implying that more earnings manage- ment increase information asymmetry. Another take on it is provided by Marquardt and Wiedman (2004), who focused on opportunistic earnings management by concentrating on the time period where the management sells their equity holdings in the company. They argued that a decrease in value relevance of reported earnings was a result of opportunistic earnings management.

Regarding information asymmetry and IFRS adoption, one benefit with IFRS is that it generally

should provide more relevant accounting information according to Barth et al. (2014). At the same

time, Callao and Jarne (2010) who found that IFRS generally has meant more earnings management

in Europe, argues that higher relevance is created through increased disclosure requirements. This

can be put in relation to a point made by Dechow and Skinner (2000) who question if earnings man-

agement actually is a problem if all information is available. However, according to Ball (2006) an

additional amount of information is not always accompanied by an increasing value relevance since

the multitude of information can be a source of inefficiency. It could be argued that an example of

when information can be overwhelming is Enron. That said, a lack information is not a desirable

environment either, hence companies have to adhere to standards and what is demanded by owners

and stakeholders. Viewing the different findings, the opinions are divided regarding information

asymmetry and IFRS implementation. For instance, Brad, Ciobanu, and Dobre (2016) found that

companies who changed to IFRS from the local accounting standard in Romania decreased the

information asymmetry. At the same time, Callao and Jarne (2010) who studied AM following

IFRS implementation in Europe, found that it had escalated. Viewing the literature related to the

relationship between AM and RAM after IFRS adoption, most of the papers, not all, find that the

use of earnings management have increased and also that the use of RAM have increased in com-

parison to AM. For instance, Ferentinou and Anagnostopoulou (2016) found that RAM increased

in the Greek market, and Ho, Liao, and Taylor (2015) found a similar result in China. However,

Doukakis (2014) did not find either an increase in earnings management or any shift from AM to

RAM. These various findings could be the result of country-specific differences.

3.2 Earnings Management Incentives: Agency Theory

Agency theory represents a central model in this context since it deals with the distinction between ownership and management. It also deals with opportunism from management, which is why information asymmetry is apparent. Agency theory targets the possible contradictory objectives between principal and agent (Morris, 1987). In a perfect world, the agent’s objective should be to work towards utility maximization for the principal, hence the behavior of its management should be in line with this overall goal (Jensen & Meckling, 1976). However, if both principal and agent could be regarded as individuals who seek maximization of wealth, their objectives might not be in line since the agent’s strategy to incur wealth might not par with the principal’s (Jensen &

Meckling, 1976). This becomes a reality in earnings management behaviors since they generally relate to opportunistic behavior from the agent and are not usually attributed to higher wealth for the company in the long run (Marquardt & Wiedman, 2004). Hence, it could be argued that earnings management can be regarded as a manifestation of agency theory since the agent acts in a way that maximizes his or her own wealth rather than the principals, which implies that they do not primarily act in the interest of the principal but rather to their own gain (Callao & Jarne, 2010).

Another view is provided by Jiraporn et al. (2006), who contradicts the negative connotation of earnings management. They argue that in some occasions earnings management can be viewed as beneficial for the principal, despite it being regarded as opportunistic for the agent, since it might concur with the overall goal of increased utility for the principal. Also, by examining the extent of agency costs and earnings management they also found that earnings management does not have to imply opportunistic management since, on average, agency costs were lower in firms that manage earnings. In a situation like this it seems like the objectives are aligned between the two, which could imply that the earnings management are not done in an opportunistic manner. According to Ewert and Wagenhofer (2005), this depends on the situation, if the principal wishes to display earnings as high as possible or rather display the company’s financial situation as accurately as possible.

As mentioned earlier, an argument could be made that earnings management might be a result of

the separation of ownership from management, i.e. the principal from the agent. This should indi-

cate that earnings management behavior most likely should continue even with another accounting

standard. Management will continue to act opportunistically and adjust earnings because it is rather

the inclination, or incentive, of management to adjust earnings in order to portray the desired pic-

ture of the company (Burgstahler & Dichev, 1997). Based on this notion, one could suggest that even if IFRS, for example impose stricter regulations regarding financial accounting, earnings man- agement will continue but with a higher focus on RAM instead of AM. The opposite could be said if IFRS has a less strict regulation compared to local GAAP, which then might lead to a higher use of AM strategies rather than RAM. This means that, even if regulation and standards diminish the possibility for using one strategy of earnings management, another will appear if an opportunity presents itself. This concurs with the findings of Zang (2012), who concludes that there is a trade- off between the two strategies based on the respective strategies inherent costs. One could suggest that these inherent costs can be viewed as agency costs if the chosen strategy is not aligned with the ownership’s wishes.

Another view is provided by Burgstahler, Hail, and Leuz (2006), they argue that incentives to manage earnings are coupled with the fact that, at least in Europe, we share a common reporting standard for all listed companies. This means that all companies, no matter the industry have to follow the same rules with no regard taken to industry-specific differences. Burgstahler et al.

(2006) also argue that the rigidity of a common standard makes the incentives to manage earnings rise due to the need for presenting a result that is more in line with the company’s performance, seeking an upward slope in earnings. In this light, earnings management could be viewed as a necessity in order for the reporting to provide a representative picture of the company’s finances, albeit somewhat subjective. A common denominator seem to be that earnings should remain stable with an upward trend, implying that if earnings is not pointing in the proper direction they are managed so as to remedy this. This concurs with Burgstahler and Dichev (1997) who argues that earnings represent one of the most important measures for managers. They also contend that managers seek consistency in their companies’ earnings patterns. An interesting point is provided by Gray et al. (2015). They studied earnings management and cultural influences by classifying countries as more or less individualistic based on work done by Geert Hofstede. They contend that countries that are depicted as more individualistic, based on Geert Hofstedes classification, generally use earnings management more compared to countries depicted as less individualistic.

Gray et al. (2015) use the Scandinavian countries as an example of less individualistic countries, which would imply that these countries generally use earnings management strategies less.

3.3 Definition and Approaches to Earnings Management

According to Beneish (2001), earnings management is an occurrence that has several definitions,

consensus regarding a common definition seems difficult to formulate. However, one definition is

that earnings management occurs when managers use their assigned discretion to alter earnings in a way that is beneficial to them (Mostafa, 2017). Roychowdhury (2006) defines earnings manage- ment as an occurrence when management use their own discretion to adjust financial reporting in an effort to mislead or to determine outcomes. A common denominator seems to be occasions where managers use their possibility to adjust earnings in a way that is beneficial to them. Jiraporn et al. (2006) provides another take on earnings management, they mention that earnings management can be defined as both a positive and negative occurrence; positive in a sense that earnings man- agement could imply a smooth reporting of earnings, meaning no surprises, which in turn should mean a stable share price. Further, it is negative if it is used in an opportunistic manner, displaying the company’s finances in a brighter light than what is actually the case, which implies that the management is not working in the interest of the owners.

According to both Kothari, Mizik, and Roychowdhury (2016) and (Fields et al., 2001), there are two types of behaviors related to earnings management. First, AM, which is the conventional approach that involves adjustments such as timeliness, provisions and impairments. Second, RAM, which involves managing actual activities such as R&D, advertising and production decisions. An important point to highlight is that, according to Roychowdhury (2006), using RAM incurs a higher information asymmetry. The two approaches are described below.

3.3.1 Accruals Based Earnings Management

The quality of earnings is, according to Christensen, Frimor, and Sabac (2013), to a large extent,

dependent on the quality of the accruals. According to Callao and Jarne (2010), accruals can

be divided into discretionary and non-discretionary accruals. A main difference between the two

is that non-discretionary accruals typically do not allow for management altercations, as the ac-

counting models limit management choice. Instead, non-discretionary accruals are dependent on

the company’s specific business environment and its business model. Hence, the change in non-

discretionary accruals represents the company’s organic growth (Christensen et al., 2013). Dis-

cretionary accruals do allow management modifications and are more accessible for modifications

by management compared to the non-discretionary accruals, hence they are more exposed to op-

portunism and in turn manipulation. Roychowdhury (2006) describes the change in discretionary

accruals as a means to alter earnings without affecting cash flow, which can be done by for instance

delaying write-offs regarding assets and bad debt expenses that are under-provisioned. Therefore,

many papers have used an approach examining the balance of accruals, and specifically the indus-

try normal levels of accruals against the actual levels to identify abnormal accruals (Mostafa, 2017;

Callao & Jarne, 2010).

Most of the models used to measure AM focus on the estimation of abnormal levels of discretionary accruals, which can be estimated by subtracting non-discretionary from total accruals, according to Defond (2010) and (Dechow et al., 1995). Hence, according to these models, actions affecting AM are limited to areas where managers have the ability to use their discretion. This implies that managers have the ability to increase or decrease the total amount of accruals. According to (Trejo-Pech et al., 2016), overall, a popular choice seems to be manipulation of timeliness in accounting. For instance, by modifying the time of write-offs earnings can be adjusted (Graham, Campbell, & Rajgopal, 2005; Trejo-Pech et al., 2016). Trejo-Pech et al. (2016) also mention the provisioning of bad debt as an area prone to adjustments, this since it is reviewed yearly and is a measure largely made up by managers judgment, which should be based on experience and historical data. By increasing or decreasing provisions, earnings can be adjusted with no effect on cash flow. Another possibility mentioned by Trejo-Pech et al. (2016) is for management to manipulate impairment of goodwill. Goodwill impairments implies subjectivity due to the inherent need for managements judgment, which in turn means that there is an apparent opportunity for managers to adjust measurement. Finally, Trejo-Pech et al. (2016) also mentions restructuring charges as an area prone for manipulation. Simply put, recognizing the cost only to, later on, reversing part of in order to increase earnings.

3.3.2 Real Activities Manipulation

As mentioned, AM can be viewed as the conventional approach to earnings management, whereas manipulating real activities seems to be a more recent occurrence in academia. However, accord- ing to Graham et al. (2005) and Defond (2010) who both explored financial reporting and earnings quality, the use of this strategy is, and have been, extensively used in business. Roychowdhury (2006) defines RAM as management activities that differs from normal business practice, which have the final intention of displaying a misleading picture of the company to its owners and stake- holders. RAM implies manipulating the companies on-going operational activities. The on-going operational activities could for example be defined as price adjustments and expenses related to for instance R&D, maintenance and advertising (Roychowdhury, 2006). In essence, management ma- nipulate actual managerial decisions by for instance adjusting the timing of decisions (postponing to the next year) and the weight of the decisions (amount of money funded) (Roychowdhury, 2006).

One could argue that the actions classified as RAM are part of a managers normal operations, not

implying earnings manipulation. However, since the model constructed by Roychowdhury (2006)

use industry-year averages, the abnormalities exceeding average are activities that differ from nor- mal business practice and are not economically motivated.

Prior research within RAM is quite scarce, the majority have found evidence of firms offering price discounts to temporarily increase their sales, aggressively reduce discretionary expenditures to boost margins, and alter production levels (overproduce to lower cost of goods sold) (Roychowdhury, 2006; Dechow & Skinner, 2000; Ferentinou & Anagnostopoulou, 2016). Another take on RAM is provided by Graham et al. (2005). They take a qualitative stance, investigating earnings manage- ment choices by interviewing multiple executives. They found that management actually preferred to use strategies classified as RAM instead of AM despite the fact that it implies a higher cost for the companies in the long run compared to AM. They explain that the real activities decisions might be fallacious and misleading but that the alternative of a lower earnings measure are more unfavourable.

Roychowdhury (2006) recognizes three types of reactions to RAM. He categorizes them in to sales,

overproduction and manipulation of discretionary expenses. He explains that sales, for instance,

can be manipulated through real activities by the time of purchase. This can be done by offering

large discounts or very agreeable terms of credit, all in an effort to persuade customers to purchase

now rather than later. He also argues that some companies, which at the end of the year realize

they might undershoot last years sales, use discounts that are limited to the current year in order

to ”steal” from the coming year and boosting sales this year. Roychowdhury (2006) argues that,

overall, these actions should increase earnings. They should be recognizable through abnormally

low cash flow from operations due to the discounts offered, and also abnormally high production

costs due to increased amount of sales. Further, Roychowdhury (2006) explains that overproduc-

tion is done in an effort to decrease cost of goods sold. This is largely the case for companies

that actually manufacture products. By increasing production, overhead costs decrease, which in

turn means a lowered cost of goods sold in total. However, this relies on the condition that the

variable costs does not rise with increased production. It should be mentioned that overproduc-

tion does not increase earnings per se, however it can increase a company’s margins. According

to Roychowdhury (2006), actions related to overproduction should be recognized by a lower cash

flow from operations due to increased costs as a result of overproduction. The last category, dis-

cretionary expenditures, relates to expenditures such as research and development and advertising

according to Roychowdhury (2006). Further, he explains that the costs related to discretionary

expenditures are normally expensed as they are arise, which makes them ideal to manipulate in

order to manage earnings in a short perspective. Therefore, in order to meet earnings targets the

discretionary expenditures should be diminished in order to boost earnings, this means that in or-

der to recognize earnings management within this category discretionary expenditures should be abnormally low (Roychowdhury, 2006).

3.3.3 The Cost of Accruals Management and Real Activities Manipulation

Another view of RAM and AM is provided by Zang (2012), who performed a quantitative exam- ination of the two earnings management behaviors. She reviewed the subject by putting the two approaches on opposite sides, viewing them as a trade-off. Implying that when the companies de- crease the use of one behavior they increase the other, hence they are each others replacements. She argued that each approach had its inherent costs, which could be used to predict earnings manage- ment behavior. In other words, if one knows the specific inherent costs, the choice of strategy could be extrapolated. Zang (2012) also mentions that the inherent costs of AM was related to regulation, whereas the costs, and use, of RAM is associated to a company’s relative competitive position in the market. Therefore, Zang (2012) also argued that a change in accounting rigidity, practice and rigor should change the inherent costs of each strategy, which might alter previous behavior. This indi- cates that the use of AM and RAM is interconnected and the use of one should diminish the other and that they are related to accounting standard according to Zang (2012). Therefore, one could ar- gue that the inherent information asymmetry, in part, depends on the inherent costs of the earnings management strategies, which in turn relies on its competitive position and regulation.

Another argument made by Zang (2012) is that the AM approach seems to be used as an offset towards the RAM approach. Implying that managers use RAM continuously and based on the outcomes of those measures adjusts earnings to further reach their goals. Zang (2012) argues that this is due to the fact that AM has a more immediate response compared to real activities. This view concurs with the findings of Graham et al. (2005), who concluded that managers are more inclined to use RAM before AM.

3.4 IFRS and Earnings Management

Companies listed on a European stock exchange have to follow IFRS according to an EU directive

decided in early 2000. In Sweden and Finland, the IFRS precondition meant that companies had to

implement it in 2005 (IFRS, 2016c, 2016a), however with a possibility to voluntarily implement it

before that as well. Norway, which is not a part of the European Union, opted for IFRS implemen-

tation as well IFRS (2016b). According to Cotter (2009), the purpose of IFRS is to make financial

reporting more accurate, transparent and accessible for the targeted stakeholders. However, there

are contradicting views if the adoption actually has led to an increased quality. The actual or be- lieved change in strictness due to the IFRS adoption is of interest since it could indicate if certain earnings management strategies still are appealing, and if it can reveal what earnings management strategy that has become most attracting according to Zang (2012). Barth, Landsman, and Lang (2008) argue that the quality surely has improved by more relevant measures. They argue that it has meant a decreased information asymmetry due to a more rigorous need for disclosure, despite it giving room for more managerial decision-making since it at the same time limits the possibility of alternative ways of accounting. However, there are a number of papers that state the opposite.

According to Callao and Jarne (2010), the IFRS adoption has led to an increase of earnings man- agement due to more room for management choice. This is in line with what Ball (2006) concludes, mainly that the possibility of manipulation have increased. Armstrong, Barth, Jagolinzer, and Riedl (2010) also raises the question whether the adoption actually has increased the quality of financial information for users within all regions. They describe that the success of the adoption depends on preconditions such as legal environment, rule enforcement and incentives for companies to report financial information as well as a cultural aspect (Armstrong et al., 2010). This means that the pos- sibility for success is, in part, dependent on how well the previous standard functioned compared to IFRS. This is supported by Ball (2006) who questions if IFRS actually can achieve convergence since a lot of local forces still remains in the countries such as different regulations and political influences. According to Martinez, Martinez, and Lin (2014), this means that the success of the adoption is dependent on country specific conditions. This is also supported by Gray et al. (2015) who reviewed earnings management after IFRS adoption and argues that it persevered despite the application of IFRS. Further, they explained this by referring to a cultural dimension, implying that earnings management is related to each country’s specific preconditions (Gray et al., 2015). Hence, earnings management is a factor that differs depending on the varying contexts. Another view in this matter is provided by Jeanjean and Stolowy (2008), who reviewed earnings quality in the light of IFRS adoption in France, Australia and the United Kingdom. They found that earnings manip- ulation was largely unchanged, with the exception of France where it actually increased following the adoption.

Armstrong et al. (2010) also argue that the adoption of IFRS can be viewed in two ways, either in a

positive manner; the market responding to companies providing financial information with a higher

quality or, negatively; responding to accounting information with lower quality. Hence, the reac-

tion is dependent on investor perception of the information provided. Thus, if the adoption of IFRS

provides more relevant and useful information in comparison to the previous regulation, reducing

the information asymmetry between the company and its stakeholders, the market should react pos-

itively to the change (Armstrong et al., 2010). Another possible scenario is that the adoption could lead to a lowered perceived quality of information, thus increasing the information asymmetry ap- parent between the company and its stakeholders. Therefore, the quality of information depends on what is considered relevant for the targeted users (Armstrong et al., 2010). The question becomes if the targeted users actually believe that more relevant information is provided to them.

As mentioned, Callao and Jarne (2010) argue that earnings management is an occurrence that largely depends on how large manager discretion is. If managers have a larger ability to affect financial measures, earnings management has a larger presence and vice versa. Another study that agrees with this notion is Ewert and Wagenhofer (2005), who argue that stricter accounting stan- dards increase accounting quality. They argue that diminishing management judgment is the key.

Further, they also claim that an effect of the more strict accounting standard is less use of earnings management strategies related to AM. However, instead of a decrease in earnings management they contend that a lower use of AM means a higher use of RAM. Suggesting that they are opposites in a trade-off is supported by Zang (2012). How that affects information asymmetry is difficult to capture. That said, Ewert and Wagenhofer (2005) determined that the diminished use of AM and increased use of RAM should, in the long run, decrease firm value since the use of RAM is more costly than AM due to the nature of the strategies. However, Doukakis (2014) disputes the trade-off notion. He agrees with the concept that strict accounting standard implies less use of AM, however not that real activities increases as a result. He did not find significant results that confirmed that suggestion. On the other hand, (Ferentinou & Anagnostopoulou, 2016) found the opposite, mainly that with the introduction of IFRS AM decreased while the use of real activities increased. In sum- mary, previous literature regarding IFRS implementation and earnings management behavior are both scarce and fragmented.

3.5 Estimating Earnings Management

There are a multitude of literature regarding AM, hence there are also a myriad of estimation models

with various pros and cons. Presented below are the ones that are most commonly used (Dechow

et al., 1995). Since RAM is a relatively new occurrence within earnings management research the

amount of literature is limited. Most of the paper utilize the framework constructed Roychowdhury

(2006), hence this is the only model regarding RAM presented below.

Table 1: Models to Measure Earnings Management The Healy Model

(Dechow et al., 1995)

This model employs a mean of accruals, which is then scaled in relation to total assets, lagged with a specific time-period. The model compares differ- ent periods since it assumes that manipulations always becomes rectified later. Hence, the model assumes that earnings management is conducted continuously and that adjustments are corrected later.

The Jones Model (Jones, 1991)

The Jones model assumes that non-discretionary accruals are non con-

stant, rather that they change in conjunction with the economic outlook

of the company. The model constructed by Jones attempts to adjusts for

the change in a company’s economic situation by estimating the change

in non-discretionary accruals, which should capture the change in accru-

als that cannot be altered by managers. The change in non-discretionary

accruals could later on be compared to total accruals, providing a measure

of how the discretionary accruals have changed. In essence, the model

attempts to adjust total accruals by modelling the behavior of each com-

pany’s non-discretionary accruals and later clear the total accruals measure

to arrive at a measure of the discretionary accruals and compare that to pre-

vious periods. According to Dechow et al. (1995), one criticism of Jones’s

model is that it assumes that revenues are non-discretionary, despite the

fact that these can be managed. This skews the findings more towards a

null result according to Dechow et al. (1995).

The DeAngelo Model (Dechow et al., 1995)

This model resembles the model constructed by Healy, however this model assumes that earnings management is a random process, not necessarily done continuously. Further, this model assumes that non-discretionary ac- cruals should be constant, meaning that under the premise of no earnings management the variation in accruals between two periods should be the same. It is important that the non-discretionary accruals are constant since they are put in relation to total accruals as to be able to compute the vari- ation of discretionary accruals. A large criticism of this model is that it relies on the assumption that non-discretionary accruals remain constant, Dechow et al. (1995) argues that this is a misconception and that the non- discretionary accruals vary in conjunction with the company’s economic reality.

The Modified

Jones Model

(Dechow et al., 1995)

This model resembles the original Jones model, however the modified ver- sion is adjusted to take into account revenues. The main criticism of the original model was the fact that it did not take into account possible adjust- ments of revenues, which this model does by comparing earnings between two event periods. This is done by hypothesizing that revenue management is done in the recognition phase of credit payments rather than payments done in cash, therefore the revenue measure is modified in relation to net receivables to take this hypothesis into account.

The Industry Model (Dechow et al., 1995)

This model, like the Jones model and the modified Jones model, assumes

that non-discretionary accruals are not constant but rather that they change

in conjunction with the economic outlook of the company. However, in-

stead of determining the non-discretionary accruals on a single-company

level this model assumes that the variances in non-discretionary accruals

is similar on a industry level. Hence, this model relies heavily on the as-

sumption that the variances are the same within each industry.

The Real Ac- tivities Manip- ulation Model (Roychowdhury, 2006).

Unlike the models mentioned above, the model constructed by Roychowdhury (2006) examines real activities manipulation instead of ac- cruals management. This is done by first estimating normal level of sales, inventory, production costs and discretionary expenses. The estimation is done using a model constructed by Dechow et al (1998) to estimate normal cash flow from these measures, which is a regression between sales and the change in sales in a given time period. This is then put in relation to the actual levels of the actual measures in order to examine the differences. In other words, a normal state is estimated through the models, which then are used as a proxy in comparison with the actual value. Roychowdhury (2006) argues that manipulation is present if the measures have abnormal levels. By examining sales, inventory, production costs and discretionary expenses Roychowdhury (2006) contend that sales manipulation through for instance an extreme generation of sales provided by unwarranted dis- counts can be found, and the same for inventory that can be manipulated by increasing or decreasing production in order to adjust cost of goods sold, and finally also discretionary expenditures by for instance adjusting funding to R&D and advertising.

Table 2 displays some of the various topics, regions and time periods studied regarding earnings

management, including both AM and RAM (table 9 displays the full version of the literature re-

view). Based on the literature review, it is evident that the results from the studies are mixed, in

some cases even in the same region (Callao & Jarne, 2010; Doukakis, 2014). The difference be-

tween the findings in similar regions could be due to different estimation models regarding earnings

management such as RAM presented by Roychowdhury (2006) or discretionary accruals Dechow

et al. (1995). Further, as argued earlier in the paper, looking at earnings management on a regional

level might be better because of country specific attributes, cultural differences and different tax-

laws. The literature review reveals that little research has been made with regards to the Swedish,

Finnish and Norwegian market, and that the models used was limited to the AM, which could lead

to misleading evidence of increased or decreased earnings management, where one might increase

and the other decrease such as the findings in Ho et al. (2015).

V ictor Dahl ´en and Daniel Lindber g

References Countries Topic Results

Ang and Pinnuck (2011)

Australia 2003 - 2006

Do employee share options give rise to earnings management (AM) under ac- counting regulation change

No evidence of increased earnings man- agement (AM) after regulation

Brad et al. (2016) Romania 2010-2012

The change of earnings management (AM) after adopting IFRS

Earnings management (AM) increased af- ter the adoption of IFRS

Callao and Jarne (2010)

European Union

The change of earnings management (AM) after adopting IFRS

Earnings management (AM) increased af- ter the adoption of IFRS

Doukakis (2014) European Union 2000-2010

The change of earnings management (AM/RAM) after adopting IFRS

No significant impact of IFRS on earnings management through RAM or AM

Evans, Houston, Peters, and Pratt (2015)

US, Europe

& Asia

US GAAP and IFRS, US and non-US firms and if it affects earnings manage- ment (RAM) or (AM)

US firms applying US GAAP use RAM over AM for earnings management more than non-US firms using IFRS or US GAAP. Likelyhood and amount of earn- ings management did not differ between regions and accounting standards

Ferentinou and Anagnostopoulou (2016)

Greece 2001-2008

Earnings management (RAM/AM) before and after the adoption of IFRS

Significant move to RAM from AM

Hamberg, Paana- nen, and Novak (2011)

Sweden 2001-2007

IFRS 3 impact on AM and market reac- tions

Capitalized goodwill increased post adop- tion of IFRS and investors interpreted this as higher future cash flows

Hellman (2011) Sweden 2004-2005

The impact of IFRS on financial state- ments

The findings indicate that firms gained dis- cretion which was used for earnings man- agement (AM)

21

Mana g ement pr e- and post IFRS Adoption - Results fr om Sweden, F inland and Norway Ho et al. (2015) China 2002-

2011

Earnings management before and after IFRS

AM decreased while RAM increased post IFRS

Rathke, Santana, Lourenco, and Dalm´acio (2016)

Latin Amer- ica 2010- 2012

Amount of earnings management (AM) after the adoption of IFRS of US cross listed firms

Cross listed Latin American firms shows a high level of earnings management (AM) compared to continental Europe and Anglo-Saxon firms

Van Tendeloo and Vanstraelen (2005)

Germany 1999-2001

Does voluntary adoption of IFRS reduce earnings management (AM)

Earnings management (AM) under IFRS was not different form earnings manage- ment under local German GAAP

AM = Accruals management; RAM = Real manipulation

22

3.6 Summary of Literature and Formulation of Hypotheses

Information asymmetry implies that one party has more information than the other, which is a sit- uation that can be translated to the condition between management and owners. This information asymmetry is remedied by financial statements, which relies on the information provided by the company and its management. How well this information captures the financial state of the com- pany can be quantified by value relevance. Earnings management is an occurrence that increases the information asymmetry and decreases the value relevance since it implies that owners does not get the correct picture of the company’s financial state. Zang (2012) argued that AM and RAM works as counterparts in a trade-off. This means that if the possibility of using AM decreases, the use of RAM will increase and vise versa. This paper does not take the trade-off into account, merely the notion that both earnings management methods exists and might be used. This relates to another notion, agency theory. Earnings management is an opportunistic behavior, which coincides with the definition put forward by Jensen and Meckling (1976). This means that managers will choose a path that concurs with their own gain instead of the company, due to their profit maximization nature. This suggests that managers will continue to manage earnings even if one possibility is diminished, most likely they will search for alternative ways to achieve their goal. For instance, if accounting regulation increases with IFRS, making it more difficult to manage earnings through AM, managers will instead find other ways to achieve their goal. The managers opportunistic be- havior will surely not diminish, instead other venues will be searched for. In other words, if the ability for managers to adjust accruals diminishes they will most likely look towards RAM instead, which concurs with what Zang (2012) argues. However, as mentioned above, previous literature are quite fragmented when it comes to the impact of IFRS on earnings management behavior.

Figure 2: Possible Scenarios

As Zang (2012) concludes, both AM and RAM are activities that imply a high cost for the company.

However, they do come with various inherent costs that varies depending on the given situation,

these situations could be altered due to the IFRS entry. This means that there are a multitude of

potential scenarios. Either AM has increased compared to RAM, or vice versa. Another scenario is that there is no change following the IFRS entry. There is also a possibility that earnings manage- ment simply have decreased due to the change, making findings related to earnings management behavior difficult to interpret and generalize. Therefore, the hypotheses are phrased as to encom- pass the variety of ways earnings management behavior could have been affected as a result of the entry.

3.6.1 Hypotheses

Following Roychowdhury (2006), the hypotheses are set up in a similar fashion. This implies singling out suspect firms, i.e. firms that just beat/meet prior years earnings and firms that just beat/meet zero threshold. Based on Roychowdhury (2006) and (Zang, 2012) these companies are more likely to manage earnings.

Hypothesis 1 (H1): Earnings management behaviour has changed in suspect firm-years among Swedish, Norwegian and Finnish firms following IFRS adoption.

This hypothesis state that that an overall change in earnings management behavior can be viewed after the introduction of IFRS in the sampled countries.

Hypothesis 2a (H2b): Suspect firm-years under IFRS exhibit abnormally high accruals.

Hypothesis 2b (H2b): Suspect firm-years under local GAAP exhibit abnormally high accruals.

These two hypotheses state that unusually higher accruals can be viewed during local GAAP and/or during IFRS. An abnormal amount of accruals can imply earnings management by adjusting ac- cruals according to Dechow et al. (1995) and Zang (2012).

Hypothesis 3a (H3b): Suspect firm-years under IFRS exhibit abnormally low cash flow from op- erations.

Hypothesis 3b (H3b): Suspect firm-years under local GAAP exhibit abnormally low cash flow from operations.

These two hypotheses state that unusually low cash flow from operations can be viewed during local

GAAP and/or during IFRS. According to Roychowdhury (2006) this is a sign of RAM.

Hypothesis 4a (H4b): Suspect firm-years under IFRS exhibit abnormally low discretionary ex- penditures.

Hypothesis 4b (H4b): Suspect firm-years under local GAAP exhibit abnormally low discretionary expenditures.

These two hypotheses state that unusually low discretionary expenditures can be viewed during local GAAP and/or during IFRS. According to Roychowdhury (2006) this is a sign of RAM.

Hypothesis 5a (H5b): Suspect firm-years under IFRS exhibit unusually high production costs.

Hypothesis 5b (H5b): Suspect firm-years under local GAAP exhibit unusually high production costs.

These two hypotheses state that unusually high production costs can be viewed during local GAAP and/or during IFRS. According to Roychowdhury (2006) this is a sign of RAM.

4 Method

4.1 Research Approach

This study contributes to the debate by investigating if there has been a change in earnings manage- ment behavior before and after IFRS adoption with regards to AM and RAM. Examining earnings management behavior is done through regression analysis, adopting the Modified Jones model and the Roychowdhury model to estimate AM and RAM respectively. The first model was originally constructed by Jones (1991) and later adjusted by Dechow et al. (1995), while the other model was devised by Roychowdhury (2006).

The study explores this subject by using a quantitative deductive approach to research. The quanti-

tative approach is fitting since the study does not intend to increase the understanding of the reasons

behind earnings management, merely uncover the magnitude of, and pattern behind, earnings man-

agement behavior. A qualitative study would have been suitable if the study sought to understand

the drivers behind earnings management behaviors, however this study attempts to quantify that be-

havior. A deductive approach was chosen as it entails using existing literature and theory to deduce

hypotheses and measures to the specific concepts, and testing these using statistical analysis with

the collected data (Bryman & Bell, 2011). This was deemed appropriate since the paper intends to test if previous findings can be generalized to the Swedish, Finnish and Norwegian example.

Two important ingredients in quantitative research are the definitions of concepts and measurement (Bryman & Bell, 2011). A concept represents a part of theory, it is a basis in determining what should be measured and how. Simply put, it is a way of labeling the elements that are studied. In this case, the concepts are limited to earnings management, specifically AM and RAM. Measurement is a central part since it provides the research with a means to distinguish differences, and a way to construct estimations (Bryman & Bell, 2011). An importance aspect in the deductive process is that the research is based on existing theories since these later are applied to the data and tested (Bryman

& Bell, 2011). Therefore, the study emanates from the previous literature that has concluded that earnings management is present in Sweden, Norway and Finland after IFRS adoption (Callao &

Jarne, 2010; Gray et al., 2015; Doukakis, 2014). Further, the study also use findings from Zang (2012), who states that there is a trade-off between AM and RAM within earnings management.

This brings a perspective on earnings management where the two behaviors relates to each other.

This study aims to use this as a backdrop for discussion. Hence, by using a deductive approach, the study intends to see if existing literature and theories can explain earnings management behaviors or circumstances in Sweden, Finland and Norway.

Since the paper approach this subject by analyzing the prevalence of the different concepts before and after a specific event (IFRS implementation), employing an analytic retrospective longitudi- nal cohort research design was deemed appropriate. According to Bryman and Bell (2011) the longitudinal design is suitable when the aim is to outline a transition. This study is dependent on the time aspect since it tries to compare an occurrence before and after a specific event. Previous studies within this field have often used a cross-sectional design when investigating the occurrence of earnings management. However, these studies have not estimated the occurrence of earnings management before and after. They have focused on the actual occurrence, hence the factor of time has not been of importance. One previous study do have a similar aim as this study, and used a longitudinal design (Callao & Jarne, 2010). Compared to the cross-sectional design, the longitudinal design controls for time, ensuring that a change in the dependent variable has occurred following a change in the independent (Bryman & Bell, 2011). However, it is still not possible to prove that a change is due to IFRS implementation. By using the cross-sectional design it would have been difficult to say that differences in AM and RAM followed due to change in accounting standard.

An issue with longitudinal designs is loss of data during follow-up, in this study some companies

dropped out due delistings during the sample time period. As this was a cohort study, the subjects