Earnings management in a global environment

(1)

Earnings management in

a global environment

Sander Voogt

Student number: 5863848

27 Juli 2014 (Final Version)

University of Amsterdam,

Faculty of Economics and Business Accountancy & Control

Supervisor:

Dr. Sanjay Bissesur Second supervisor: Dr. Ir. Sander van Triest

(2)

Preface

This master thesis is written to obtain the Master of Science degree in Business Administration (MScBA).

There are many people to thank for the considerable support I received during the completion of this master thesis. First of all, I would like to thanks my parents for their enormous support to continue and fulfil this last phase of my study.

At the beginning of my university studies the writing of my thesis seemed as an impossible task. Now that I have successfully accomplished this thesis I want to thank all those people who have supported me one way or another. Finally, I would like to thank my supervisor dr. S.W. Bissessur for his supervision and patience throughout the process.

Amsterdam, July 2014

(3)

Abstract

This study investigates whether the level of earnings management in Dutch listed companies is depending on the complexity of the company. As firms and markets become increasingly global, there is a need to better understand how globalization impacts the accounting values of the company to protect shareholders.

My research showed that there is significant difference in earnings management by globalized companies in comparative to national organised organisations. Research only suggests the foreign sales is controlling the earnings management. Even though there are some indications that the size and the number of operating units plays a role in earnings management.

Key words: earnings management, globalisation, culture, discretionary accruals and modified Jones model.

(4)

1 Introduction

In this chapter, the subject of the study is presented. First, the subject and the research questions are illustrated. Secondly, the motivation and contribution to prior literature are presented. Third follows the main findings and the conclusions. The chapter ends with a description of the structure of the study.

1.1 Subject and research question

The last years a number or research has been performed with regard to earnings management, as Walker (2003) noted “How far can we trust numbers”. It’s also the regular questions friends asked me after the number or incidents (enron, ahold, vestia) where management was cooking the books in disinterest of the shareholders. As an auditor I am working in a cross cultural environment and frequently facing the complexity to validate management estimates with respect to the financial statements. The general accepted accounting standards (IFRS) and accounting standards (ISA) gives examples and instructions how to deal with estimates. But commonly ends up in a discussion with management where there is no real good or false, it’s a matter of judgement and interpretation of the regulation. But how does this works out if the environment of the company is international and involves management with different views, values and culture backgrounds?

Companies have several stakeholders but preliminary have to operate in flavour of the shareholders who participate in the company. Earnings management occurs when managers use judgement in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company. Whereby the judgement can be divided in conservative, neutral, aggressive or fraudulent. From prior literature we learned that there are several motives for management to conduct earnings management. Management (variable) compensation has a positive correlation with the incentive to positive management the underlying performance of the company (Ball, Kothari and Robin (2000)) and have the tension to have more creative measures to income smoothing (Watts and Zimmerman 1986)). Pressure of financial markets and debt covenants showed a positive relation with earnings management (Dicheva and Skinner). Intentions for initial public offerings Aharony et al (1993) call for a higher initial share price. Or blaming the old management and show better results on a sort term notice, by big-bath accounting, Moore (1973).

Investors are protected for earnings management by the common shared values of the company, Doupnik (2008). Whereby the cultural dimensions, institution consequences, accounting values and accounting systems interact with each other to maintain that management operates in the flavour of the shareholders, the values protect shareholders. Researched by Gray and Hofstede shows that are several difference in the effectiveness of the shareholder protection mechanise in a cross-cultural environment. Hofstede defined culture as the collective programming of the mind distinguishing the members of one group or category of people from others. Dechow and Skinner (2000) noted that it’s essential to understand what the members of the group share as a common

(6)

value to understand investor protection. As an example House et al (2002) found evidence for the correlation between masculinity and earnings smoothing and discretion and that for example Nordic countries are more risk taking and optimistic than for example Germany. These differences are an indication that companies who are more international have potentially more difficulties to control the total organisation, or in other words management have more opportunities to carry out earnings management.

Webb et al (2008) already discovered evidence on a significant interaction between globalization and legal environment. In which he indicated that’s there is more voluntary disclosure for firms surrounded in a weak legal environment. Furthermore Barinov (2012) shows that post-earnings-announcement drift is stronger for conglomerates, despite conglomerates being larger, more liquid, and more actively researched by investors. Limitation in their research is the observation that complexity is an arbitrage variable, as we confirm, complex firms are significantly larger and their other characteristics, like trading costs, volatility, analyst coverage, and institutional ownership, suggest that complex firms should have lower limits to arbitrage.

Besides the view that complexity is an arbitrage variable I found it fascinating to analyse the relationship between complexity and earnings management. As this is already explored in the research of Barinov (2012), where post-earnings announcement can be an indicator for revenue stabilisation. Furthermore I expect that the cross board difference in legal systems and cultural dimensions decrease the effectiveness of the shareholder protection regulations. The complexity reduces the effect of control mechanism of shareholders and increase to opportunities to control the figures in their own profit, cooking the books in disinterest of the shareholders.

This study will concentrate on analysing earnings management manipulation in the Netherlands. Earnings management is examined on the complexity of the firm; foreign sales, concentration of foreign sales and number of operating units. The modified jones model will be used by measuring the discretionary accruals (abnormal) and non-discretionary accruals. Data set is based on financial data of Dutch companies listed on the Amsterdam Stock Exchange during the period 2001 until 2011, using the databank of Datastream.

Main objective of this study is to examine whether complexity has affected earnings management practices in Dutch publicly quoted companies following the research question:

“To what extent does the complexity of an entity plays a role in earnings management within Dutch public quoted companies?”

The research question is marked out by sub questions:

• To what extent is earnings management driven by the magnitude of foreign sales. • To what extent is earnings management driven by the concentration of foreign sales. • To what extent is earnings management driven by the number of operating units.

(7)

1.2 Contribution to prior research

As described in the previous section a number of research has been performed on earnings management and globalization. However no research has been performed on the interaction between globalization and earnings management. The purpose of the research is to contribute to existing knowledge about earnings management and globalized companies in the Netherlands. Further, it supports the view that earnings management might varies amongst the foreign sales, concentration of foreign sales and the number of operating units.

1.3 Structure of study

In the next chapter, prior literature is reviewed. In that chapter, the prior literature on earnings management is described. In the third chapter, the methodology is illustrated. A description of the methodology and the motivations for choosing the research method are given. The fourth chapter presents the findings of the data analysis. These findings are compared with the prior literature in the conclusion in chapter five. Chapter five contains also an analysis of the limitations of this research and the possibilities for future research. The study is concluded with the answer on the research question in the conclusion.

(8)

2 Prior literature

This chapter discusses the theoretical background of the research question; what is the definition of earnings management and what are the main motives for management? First, the chapter resumes prior literature on the earnings management. A description is given of the meaning of dysfunctional behaviour and what types of dysfunctional behaviour exist.

2.1 Earnings management

The phenomenon of earnings management has been the topic of a large body of research in the last few decades. As a result, a number of different definitions are used for earnings management:

Schipper (1989) defines earrings management as follows: “ a purposeful intervention in the external financial reporting process, with the intent of obtaining some private gain (as opposed to, say, merely facilitating het neutral operation of the process”. While Schipper preliminary focus on the intervention of the external reporting process to obtain private gains, Healy and Wahlen (1999) have a broader definition and also include other motives of managers to mislead the underlying economic performance of the company. “Earnings management occurs when managers use judgement in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company, or to influence contractual outcomes that depend on reported accounting numbers”

Scott (2009) and Roychowdhury (2006) formulated earnings management more negatively: “‘Earnings management is the choice by a manager of accounting policies as at to achieve some specific objective” and “Real activities manipulation is defined as departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations”.

The majority of the definitions above describe earnings management as a negative, and misleading the shareholder. According to Healy (1989) earnings management is a matter of judgments which are not preliminary the motive for management to mislead the company but also can be a matter of interpretation of accounting values. This is in-line with the research of Ronen and Yaari (2008) there are many definitions of earnings management and these definitions can be classified as:

• White: Earnings management is taking advantage of the flexibility in the choice of accounting treatment to signal the managers’ private information on future cash flows.

• Grey: Earnings management is choosing an accounting treatment that is either opportunistic (maximizing the utility of management only) or economically efficient.

• Black: Earnings management is the practice of using tricks to misrepresent or reduce transparency of financial reports.

(9)

“earnings management is a collection of managerial decisions that result in not reporting the true short-term, value-maximizing earning as known to management. Earnings management can be: _{Beneficial: it signals long-term value;}

_{Pernicious: it conceals short- or long-term value;}  Neutral: it reveals the short-term true performance.

The managed earnings result from taking production/investment actions before earnings are realized, or making accounting choices that effect the earnings numbers and their interpretation after the true earnings are realized”.

The first part measures earnings against short-term truth as it is known to management. The second part attaches subjective value to earnings management. Subsequently the third part describes in a broad sense how earnings management is achieved.

For this study we do not make a choice which definition we use for earnings management, because for our research we use a model to detect earnings management.

Levels of judgement

As described in the previous section accounting values are the main drivers of judgments. It’s important to have an understanding of the four different accounting values to understand the drivers for earnings management. This section describes four different levels of accounting values based on the framework of Dechow and Skinner (2000). The figure below shows that there is a difference between fraudulent accounting and earnings management by the violation of GAAP (General Accepted Accounting Principles). As described in the first section by using the definition of Healy and Wahlen this study is based on the perception that management not violates GAAP.

(10)

Following Dechow and Skinner (2000) neutral accounting results following directly the operations, no tendency from management. Conversely this implies that within neutral accounting management have no tendencies Xie at al. (2001) questioned whether neutral accounting exists based on their conclusion, that management will bring bad news earlier than good news. Therefore it’s suggested that smoothing results by neutral accounting is also a method for earnings management.

Bliss (1924) defines conservative accounting as follows: “Anticipate no profits, but anticipate all losses”. Anticipation of losses indicates that ‘good news’ is earlier reflected in the financial statements than ‘bad news’ Basu (1997). Following the statement of Basu, earnings reflect bad news more quickly than ‘good news’. For instance, unrealized losses are recognized earlier than unrealized gains and provisions have the tendency to be overstated instead-off understated`, conservative accounting.

Conservative accounting is closely related to auditor reporting conservatism, as this research by Francis and Krishnan (1999) showed that high accrual-firms are more likely to receive a modified audit report. Which indicated that auditors compensate their risk exposure to lower their threshold for modified audit reports?

Aggressive accounting is one of the main indicators for earnings management. As management is willing to understate provisions and drowning down current provisions within the boundaries of local GAAP. In chapter 4 I will further discus the main tendencies of earnings management.

2.2 Motives for earnings management

In the previous section, we discussed the definition and levels of earnings management. This section will further explain the main drivers and motives of management to conduct earnings management.

Given the definition of Healy and Wahlen (1999) earnings management occurs when management use judgements to influence the underlying economic performance or contractual outcome. The possible motives of management to apply earnings management can be categorized into three categories:

1. Revenue optimisation 2. Revenue overstatement 3. Revenue stabilisation

To optimize, stabilize or overstate revenues management has several motives, which will be explained further in this section.

Management compensation:

Management often receives a variable compensation, also known a bonus, on top of their fixed salary. This can be paid out for example in cash, options, shares, products and have the tendency to increase their personal welfare. Especially for top management this could be a substantial part of their total

(11)

remuneration as we see currently in the discussion on remuneration of bankers. This variable compensation could be based on several objectives, operational and financial.

As a result of the economic crisis a discussion started whether management should be compensated on the financial performance of the company, here we focus mainly on the period before the economic crisis where management was rewarded for the financial performance of the company.

Ball, Kothari and Robin (2000) noted in their research that there is a positive correlation between the variable remuneration and management incentive to positive manage the underlying performance of the company. Which also relates to motive for earnings optimisation of Healy (1985) that organisations have the tendency to accumulate losses to boost revenues in the coming years.

According to Watts and Zimmerman (1986), organisations with a variable compensation scheme are more likely to creative measures to smooth income between periods (manage the underlying performance of the company). Reward management by the same incentive as investors, the net result of the company. However this does not take the long-term objectives of investors, although a highly profitable organisation increases the economic welfare of the investor and management on the short run. Long-term objectives require sometimes large investments, restructurings within the organisation with less results / losses in the short run.

Pressure on financial markets:

Given the current debt crisis there is an increased focus of management to meet the debt covenants of the bank. Dichev and Skinner (2002) found strong evidence that “managers take actions to avoid debt covenants violations: we find an unusually small number of firm/ quarters with financial measures just below covenant thresholds and an unusually large number of firm/ quarters that just meet or beat covenant thresholds”. This is in-line with article of Defond and Jiambalvo (1994) who examined that companies that are close to the covenants make more use of discretionary accruals in the current and the past year than other companies. In the year of violation, they find evidence of positive abnormal discretionary accruals. This indicates that debt covenants and pressure from financial markets is a drive for management to violate the underlying figures by using discretionary accruals.

Initial public offerings:

Aharony et al (1993) noted that managers overstate earnings in the period prior to the initial public offering “The presumed goal of this manipulation is to induce outside investors to pay a higher (offer) price for the firm’s common shares than is justified by its true profitability”. The method to overstate the earnings to the initial public offering is studied by Friedlan (1994) who found out that initial public offering issuers use discretionary accrual in the period before the offering. This is also the result of the study of Teoh et al (1998) who found evidence that initial public offering firms have high

(12)

positive earnings in the issue year compared to the years before and after. Scott (2006) indicated that the net profit is an important indicator to determine the value of the shares. Therefore managers have the tendency to overstate revenue. As managers regularly have shares in the company they work for, they directly benefit from the high share price.

Small loss avoidance:

Managers do not want to report red figures, as indicated by Burgstahler and Dichev (1997) management tend to avoid especially small losses. Which result in an abnormal high number of small positive earnings in there result. In their study they found out that there are significant more companies that record a small gain in comparative to a small loss, refer to picture below.

Figure 1Avoidance small losses by Burgstahler and Dichev

Customers are willing to pay a higher price for goods because the firms are assumed more likely to honor implicit warranty and service commitments.

• Suppliers offer better terms, both because the firm is more likely to make payments due for current purchases and because the firm is more likely to make larger future purchases.

• Lenders offer better terms because the firm is less likely to either default or delay loan payments.

• Valuable employees are less likely either to leave or to demand higher salaries to stay.

Other explanation for small loss avoidance is given by De Angelo (1988) who examines accounting choices by incumbent managers during proxy fights. DeAgelo concludes that incumbents avoid earnings decreases and also reports evidence “…consistent with the hypotheses that incumbents exercise their accounting descretion to avoid reporting a net loss druing an election campaign, perhaps because of the empahasis that dissendents accord these losses”.

(13)

Change of management:

Moore (1973) examined that a potential change of management could be an indicator for earnings management: “First, the reported low earnings may be blamed on the old management, and the historical bases for future comparison will be reduced. Second, future income would be relieved of these charges, so that improved earning’s trends could be reported”. Furthermore Dechow and Sloan (1991) found out that managers which will leave the company on a short notice will have a short-term focus on the results. Through decreasing the costs for research and developments and increase capitalisation of intangible SIZE.

2.3 Protection for earnings management

In the previous section the motives for management to conduct earnings management has been discussed. This section will describe the investor protection mechanism to control management.

The framework of Doupnik (2008) is commonly used in prior literature to explain the relationship between earnings management and investor protection mechanisms. In four different clusters he described the importance of the interaction between the cultural dimension, institutional consequences, accounting values and accounting systems. Doupnik combined in his framework several previous literatures and elaborate the importance of the balance and interaction of the clusters. With regard to the research question of this study the main focus will be on the cross-cultural differences in the framework, refer to detailed explanation of the clusters below.

Figure 2 Relationship matrix by Doupnik

Culteral dimensions: Individualism Power distance Uncertainty avoidance Masculinity Long-term orientation Institutional consequences: Legal system Corporate ownership Captial markets Professional associations Education Religion Accounting systems: Authority Enforcement Measurement Disclosure Accounting values: Professionalism Uniformity Conservatism Secrecy

(14)

Cultural dimensions

The cultural dimensions of the framework for Doupnik is based on the research performed by Hofstede (1980,1983). To understand the importance of culture with respect to earnings management at first the cultural dimension of Hofstede will be explained. In his research he detected five structural elements of culture and particularly those which most strongly affect behaviours’ in work situations.

• Power distance: The extent to which less powerful members of organizations and institutions accept and expect that power is distributed unequally. The basic problem involved is the degree of human inequality that underlies the functioning.

• Uncertainty avoidance: The extent to which a culture programs its members to feel either uncomfortable or comfortable in unstructured situations. Unstructured situations are novel, unknown, surprising, and different from usual. The basic problem involved is the degree to which a society tries to control the uncontrollable.

• Individualism versus collectivism: The degree to which individuals are supposed to look after themselves or remain integrated into groups, usually around the family. • Masculinity: refers to the distribution of emotional roles between the genders; it

opposes “tough” masculine to “tender” feminine societies. Male achievement reinforces masculine assertiveness and competition; female care reinforces

Hofstede defined culture in his research as follows: “Culture is the collective programming of the mind distinguishing the members of one group or category of people from others”. With reference to the levels of judgement of earnings management and accounting values as discussed by Dechow and Skinner (2000).

House et al (2002) increased the number of dimensions based on his project Global Leadership and Organizational Behavior Effectiveness Research Program (GLOBE) project. Which gives more insight in the cultural dimensions but shows no other insights as already noted by Hofstede. With respect to the scope of this study the dimension of Hofstede will be used in this study. Based on the five dimensions as described by Hofstede, Doupnik (2008) scrutinized the relationship with earnings management in an international context, as can be seen in the figure 4 below.

Earnings management

Cultural dimensions Earnings smoothing Earnings discretion

Uncertainty avoidance Positive Positive

Individualism Negative Negative

Power distance None Positive

Masculinity Positive / Negative Positive

Long-term orientation None Negative

Figure 4 Relationship cultural and earnings management

Uncertainty avoidance, power distance and masculinity have in general a positive effect on earnings management. Organisation and management with this cultural dimension has more tendencies to

(15)

manage the underlying performance of the company than in countries with another cultural dimension.

Institutional consequences

Legal systems protect investors by conferring on them rights to discipline insiders, as well as by enforcing contracts designed to limit insiders’ private control benefits Leuz (2003). Legal systems that effectively protect outside investors reduce insiders’ need to hide their activities. From a global perspective this has the implication that countries with a weak legal institutional have more tendencies to earnings management.

Legal systems differ based on the ownership, financing and control of the company. Burgstahler et al (1997) provide evidence that earnings management is more pronounced in countries with a weaker legal system and enforcement. Strong capital markets, legal institutions reduce the level of earnings management primarily for public traded firms. Whereby more developed countries, as for example the Netherlands, can be seen as strong capital markets. As a result expansion to less developed countries with a weaker legal system increase the potential level of earnings management.

This result is in-line with the researched of Leuz et al (2003) who found out that there’s a relationship between the institutional framework to protect investors and earnings management. If managers have the incentives to conduct earnings management this have an effect on the performance of the company. By using a proxy buy La porta (1998) to measure the how strong legal systems are over countries Leuz found significant evidence for the following ”The greater the protection provided outside shareholders and the stronger the enforcement of legal rights, the lower the amount of earnings management that takes place within a country”.

This study only focused only on Dutch public firms, as a result it is expected that based on the research of Burgstahler and Leuz the level of earnings management does not differ as they all operate under the same legal system. However the level of education and religion differs within a company when the operations are more word-wide oriented.

Accounting values

In comparative to accounting standards there’s no general definition for accounting values and systems. Therefore these values differs in each countries based on their ecological influences, social values and institutional consequences. Gray (1999) relates these accounting values to the level of the accounting subculture:

Professionalism versus statutory control: A preference for the exercise of individual professional judgement and the maintenance of professional self-regulation as opposed to compliance with prescriptive legal requirements and statutory control.

(16)

Uniformity versus flexibility: A preference for the enforcement of uniform accounting practices between companies and for the consistent use of such practices over time as opposed to flexibility in accordance with the perceived circumstances of individual companies

Conservatism versus optimism: A preference for cautious approach to measurement so as to cope with uncertainty of future events as opposed to a more optimistic, laissez-faire, risk taking approach Secrecy versus transparency: A preference for confidentiality and the restriction of disclosure of information about the business only to those who are closely involved with its management and financing as opposed to a more transparent, open and publicly accountable approach.

If we relate the accounting values and systems of Gray to the cultural dimension of Hofstede this result in the following matrixes.

Figure 5 Relationship matrix measurement and disclosure by Gray

Measurement and disclosure:

Nordic countries are more risk taking and optimistic than the more developed Latin countries and Germany. These countries are also less transparent and open in there public information.

If we reflect this on the companies within the Netherlands with a substantial subsidiary in Germany, it’s hypotheses that this information is more conservative and less transparent in comparison to as subsidiary in Norway.

The matrix of measurement and disclosure shows also that within Europe there are large diversity in accounting values. Although all countries have to report since 2005 under the same International Financial Reporting Standards (IFRS) within the European Union the underlying values differ. Which potentially also explains the difficulties in the creation and adjustments on IFRS.

The same applies for companies who operation in different countries, the internal reporting standards (corporate reporting standards) applies for all subsidiaries. But accounting values transparency and conservatism could potentially differ on the accounting subculture in the countries.

(17)

Figure 6 Relationship matrix authority and enforcement by Gray

Authority and enforcement:

Nordic countries have the preference to work individually on their judgement while for example Japan is focussed on compliance with their legal requirements. Their practice is more standardised and less flexible. Reflecting on the globalisation of companies in last few decades in especially Asia there is a huge gap in the accounting values.

2.4 Complex firms

In the previous section discussed the motives for earnings management. This section will further explain the characteristics of complex, conglomerate firms.

Khanna, Kogan and Parelpu (2006) defines globalization as follows “globalization entails a lifting of barriers to the mobility of capital, products and labor to an intensification of completion for these factors across boards by firms and countries”. In the neoclassical models financial globalization generates major economic benefits. It enables investors world-wide to share risks better, allows capital to flow where its productivity is highest, and provides countries opportunities to reap the benefits of their respective comparative advantages, Stultz 2005.

This neoclassical model is discussed by Stulz and Aggarwal, saying that are limitation on financial globalization also called the twin agency problems. Which connect the first agency problems of the 16’th century to challenges of these days. Refer to appendix IIV the foreign SIZE as a function of GDP is increase significantly after the World Ware II.

The terms earnings management and globalization have been widely studied. Still, for the most part, these studies do not consider the role of globalization fact that many firms operate and raise funds in multiple countries. As firms are markets become increasingly global, there is a need to better understand how globalization impacts the accounting values of the company to protect shareholders for earnings management.

(18)

Webb et al (2008) already discovered evidence on a significant interaction between globalization and legal environment. In which he scrutinized that’s there is more voluntary disclosure for firms bordered in a weak legal environment. The results indicate that the effects of globalization are most pronounced for firms from weak legal environments. First, globalization can increase the benefits of disclosure by exposing firms from weaker legal environments to new markets where disclosure is more highly valued. Second, as firms from these countries globalize, they need to provide better governance and extra disclosure to build trust and enhance their reputation. There results contribute to the debate about convergence of corporate governance and indicates that globalization is an important variable that has been overlooked previous cross-country research.

Barinov et al (2012) contributed to the research by Webb (2008) et all by analysing the relationship between complexity to the speed of information processing. And shows the post-earnings-announcement drift is stronger for conglomerates, despite conglomerates being larger, more liquid, and more actively researched by investors. In addition they noted that stronger for new conglomerates than it is for existing conglomerates and that investors are most confused about complicated firms that expand from within rather than firms that diversify into new business segments via mergers and acquisitions. Using three measures for complexity:

• Dummy variable for conglomerates • Number of business segments

• Sales concentration based on the Herfindal index

Showing in a cross-sectional regression that post-earnings-announcement drifts are twice as strong from complex firms as it is for single-segment firms.

As discussed in the articles of Webb and Barinov firm complexity is an arbitrage variable, as we confirm, complex firms are significantly larger and their other characteristics, like trading costs, volatility, analyst coverage, and institutional ownership, suggest that complex firms should have lower limits to arbitrage. Through its gives a good illustration of how complexity have an effect on the legal and corporate governance environment of the firm.

(19)

3 Methodology

In this chapter, the methodology is discussed. First, the research method is presented. The motivations for choosing the research method are discussed. Second, the hypothesis are discussed in relation to prior literature. Third paragraph explains the methodology which will be used The fourth paragraph explains the construction of the data set and the sample. Fifth paragraph explain the control variables which will be used in the regression. The last paragraph will elaborate the regression models for the testing of the hypothesis.

3.1 Research method

As discussed in the previous chapters there are several methods and motivations to conduct earnings management. This makes it difficult to develop a method to detect earnings management based on external data sources. Based on previous research there are in general to methods to detect earnings management; discretionary line items and accrual approach.

In this study will made use of the accrual accounting as discretionary line items mainly focusses on the extraordinary items (reorganisation provision, change in accounting policies and so on). And accrual accounting includes all methods and motivations to conduct earnings management. In addition this study only includes post-IFRS financial statements within the Netherlands. Therefore the change in accounting policies will consistent for all selected items.

Healy model:

Healy (1985) was one on the first researchers in detecting earnings management, by estimating deviations from the average. This model was different because it assumes that systematic earnings management takes place in every period. Healy (1985) starts with total working-capital accruals. Total accruals (TAt) are defined by:

𝐴𝐴𝐴𝐴𝐴𝐴𝑡𝑡 =(∆𝐴𝐴𝐴𝐴𝑡𝑡− ∆𝐴𝐴𝐶𝐶𝑡𝑡− ∆𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶_𝐴𝐴 𝑡𝑡+ ∆𝐶𝐶𝑆𝑆𝑆𝑆𝑡𝑡− ∆𝑆𝑆𝐷𝐷𝐷𝐷𝑡𝑡

𝑡𝑡−1

Where,

ACR = total working capital accruals ΔCA = change in current SIZE ΔCL = change in current liabilities

ΔCASH = change in cash and cash equivalents

ΔSTD = change in debt included in current liabilities DEP = depreciation and amortization expense

(20)

Healy parts his sample by comparing three groups. In one group earnings management is presumed to be managed upwards and in the other two groups’ earnings are presumed to be managed downwards. The group of observation where it is presumed that earnings are to be managed upwards are treated as the estimation period and the group of observations where it is presumed that earnings are to be managed downwards are treated as the event period. Graphically, the compensation package of a manager who receives a base salary and a bonus is a piecewise contract as depicted in figure below.

Figure 7 Graph management compensation Healy model

The compensation is based on reported earnings, with a lowest level at A and a highest level at B. The accounting flexibility allows managers to inflate earnings by the difference between A and B, this compensation schedule can be divided into three zones:

• To the left of A • Between A and B • To the right of B

If the real earnings fall in the second zone (A-B), the manager can increase his current bonus by inflating the earnings of the firm. If the true earning fall in the first zone (to the left of A) or in the third zone (to the right of B), inflating the earnings of the firm will not yield higher compensation. The mean total accruals from the estimation period represent the measure of nondiscretionary accruals (NDA). This leads to the following model for determining NDA (Dechow et al., 1995).

𝑁𝑁𝑆𝑆𝐴𝐴𝑡𝑡=∑ 𝑡𝑡_𝑆𝑆𝑆𝑆𝐴𝐴𝑡𝑡

Where:

NDA = estimated nondiscretionary accruals; TA = Total accruals scaled by total SIZE;

T = 1, 2, … T is a year subscript for years included in the estimation period; and t = a year subscript indicating a year in the event period.

(21)

Di Angelo model

The DeAngelo (1986) model does not differ much from the Healy model. In the DeAngelo model the period of estimation for non-discretionary accruals is focused on the prior year observation. The total accruals of the previous year are the measure of discretionary accruals. This implies that non-discretionary are equal to the total accruals of the last period (Bartov et al., 2000).

𝑁𝑁𝑆𝑆𝐴𝐴𝑡𝑡= 𝑆𝑆𝐴𝐴𝑡𝑡−1

The changes between this period and the previous period are seen as discretionary accruals. Both the Healy and the DeAngelo model assume that nondiscretionary accruals are constant over time, and that changes can only be discretionary. If nondiscretionary accruals are constant over time and discretionary accruals have a mean of zero in the estimation period then the model will measure nondiscretionary accruals without error (Bartov et al., 2000). If nondiscretionary accrual will vary over time, then the both models will measure nondiscretionary accruals with error. (Dechow et al., 1995).

According to Ronen and Yaari (2008) DeAngelo defines total accruals as in Healy (1985), except that in her earlier study DeAngelo adjust earnings to reflect the impact of the equity method of accounting for intercorporate investment.

In the 1986 study, DeAngelo does not detect earnings management, but in the 1988 study she is more successful. The test results indicate that earnings increase during a contest by about 1% of total SIZE, but accruals increase by about 2%. Most studies at this day do not apply DeAngelo’s approach unless they seek to compare the efficiency of different models of discretionary accruals (Dechow, Sloan and Sweeney, 1995; Thomas and Zhang, 2000; Bartov, Gul and Tsui, 2002).

Jones Model

The Jones model (1991) improves the models of Healy and DeAngelo by controlling the effects of changes in a firm’s economic circumstances on nondiscretionary accruals. Jones abandoned the assumption that non-discretionary accruals remain constant. The Jones model takes the change in revenues (thus also the growth of the firm) into account and adds the total amount of property, plant and equipment. The Jones model includes the change in revenue and the total amount of property, plant and equipment; because Jones recognized that accruals depend on the business activities of a firm. The Jones model for determining non-discretionary accruals is:

𝑁𝑁𝑆𝑆𝐴𝐴𝑡𝑡= 𝛼𝛼1�_𝐴𝐴1

𝑡𝑡−1� + 𝛼𝛼2(∆𝐴𝐴𝐷𝐷𝑅𝑅𝑡𝑡) + 𝛼𝛼3(𝐷𝐷𝐷𝐷𝐷𝐷𝑡𝑡) Where:

NDAt = the nondiscretionary accruals in the event period

∆REVt = revenues in year t less revenues in year t-1 scaled by total SIZE at t-1 PPEt = gross property plant and equipment in year t scaled by total SIZE at t-1

(22)

At-1 = total SIZE at t - 1

α1, α 2, α 3 = firm-specific parameters

Estimates of the α 1, α 2, and α 3 are generated using the following model in the estimation period:

𝑡𝑡−1� + 𝛼𝛼2(∆𝐴𝐴𝐷𝐷𝑅𝑅𝑡𝑡) + 𝛼𝛼3(𝐷𝐷𝐷𝐷𝐷𝐷𝑡𝑡) + 𝜀𝜀𝑡𝑡

Where TA stands for total accruals scaled by lagged total SIZE, while, α 1, α 2, and α 3 denote the ordinary least squared estimates of α 1, α 2, and α 3. In the Jones model the predicted value of the regression is the normal level of accruals and is referred to as non-discretionary accruals (NDA). Discretionary accruals (DA) are the residual of TA and NDA. In this way, earnings management activities can be found.

A limitation of the Jones model is the fact that earnings could be managed through influencing revenues, by adding revenue at the end of the year that are not earned yet and for which no cash has been received. The Jones model will then be biased to zero and will make an incorrect assumption that there is no case of earnings management (Dechow et al., 1995).

Modified jones model

Dechow, Sloan and Sweeney (1995) adjusted the Jones Model to eliminate the limitation of the original Jones model. The modified Jones model estimates nondiscretionary accruals during the period in which earnings management is assumed as:

𝑡𝑡−1� + 𝛼𝛼2(∆𝐴𝐴𝐷𝐷𝑅𝑅𝑡𝑡− ∆𝐴𝐴𝐷𝐷𝐴𝐴𝑡𝑡) + 𝛼𝛼3(𝐷𝐷𝐷𝐷𝐷𝐷𝑡𝑡)

In this formula the variable ∆REC is added, which stands for net receivables in year t less receivables in year t-1 scaled by total SIZE at year t-1.

Estimates of the α 1, α 2, and α 3 are generated in the same way as in the original Jones model. The modified Jones model assumes that all variations in the credit sales in the event period are a sequence from earnings management. The reason behind this is that it is easier to manage earnings over the recognition of revenue on cash sales. If this adjustment in the Jones model is a success, than the detection of earnings management should no longer be biased towards zero as it was in the original Jones model. (Dechow et al., 1995).

Performance-matching Jones accrual model

Kothari (2005) developed a model called the performance-matching Jones accrual model. The first stage of estimating accruals is similar to the standard Jones and the modified Jones model.

(23)

𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡

𝐴𝐴𝑖𝑖𝑡𝑡−1=

𝐷𝐷𝐸𝐸𝐸𝐸𝐸𝐸𝑖𝑖𝑡𝑡 − 𝐴𝐴𝐶𝐶𝐶𝐶𝑖𝑖𝑡𝑡

𝐴𝐴𝑖𝑖𝑡𝑡−1

In this model the approximation of NDA is almost similar to the modified-Jones model, except that it includes ROAit.

𝑁𝑁𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡 𝐴𝐴𝑖𝑖𝑡𝑡−1 = 𝛽𝛽1 � 1 𝐴𝐴𝑖𝑖𝑡𝑡−1� + 𝛽𝛽2 � ∆𝐴𝐴𝐷𝐷𝑅𝑅𝑖𝑖𝑡𝑡− ∆𝐴𝐴𝐷𝐷𝐴𝐴𝑖𝑖𝑡𝑡 𝐴𝐴𝑖𝑖𝑡𝑡−1 � + 𝛽𝛽3 � 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑡𝑡 𝐴𝐴𝑖𝑖𝑡𝑡−1� + 𝛽𝛽4 � 𝐴𝐴𝐶𝐶𝐴𝐴𝑖𝑖𝑡𝑡 𝐴𝐴𝑖𝑖𝑡𝑡−1� + Where:

NDAt = the nondiscretionary accruals in the event period A it-1 = Total SIZE for firm I in period t-1

∆REV it = revenues in year t less revenues in year t-1 for firm I ∆REC it = revenues in year t less revenues in year t-1 for firm I PPEt = gross property plant and equipment in year t for firm I B1,B2,B3,B4 = regression coefficients estimates

ROA it = return on SIZE for firm i in period t

After NDA has been determined, the DA can be determined by subtracting NDA from TA 𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡 = 𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡− 𝑁𝑁𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡

Kothari (2005) suggest that this model might provide stronger results for earnings management. Dechow et. al model: a new approach

Dechow, Hutton, Kim and Sloan (2011) designed a new approach for detecting earnings management. This model is defined as follows:

𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡 = 𝑎𝑎 + 𝑏𝑏𝐷𝐷𝐴𝐴𝐴𝐴𝑆𝑆𝑖𝑖𝑡𝑡− 𝜀𝜀𝑖𝑖𝑡𝑡

Where:

DA it = Discretionary accruals for firm i in period t

PART it = A dummy variable that is set to 1 in periods during which a hypothesized determinant of earnings management is present and 0 otherwise

A,B = Regression coefficients;

E it = Regression residuals for firm i in period t Equation as follows:

𝑊𝑊𝐴𝐴_𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑡𝑡 = 𝑎𝑎 + 𝑏𝑏𝐷𝐷𝐴𝐴𝐴𝐴𝑆𝑆𝑖𝑖𝑡𝑡+ � 𝑓𝑓𝑘𝑘𝐸𝐸𝑘𝑘𝑖𝑖𝑡𝑡

(24)

Where:

WC_ACC = Non-cash working capital accruals for firm i in period t

PART = dummy variable that is set to 1 in periods during which a hypothesized determinant of earnings management is present and 0 otherwise

Xk = Controls for nondiscretionary accruals

They use non-cash working capital accruals (WC_ACC) as the measure of accruals in 𝑊𝑊𝐴𝐴_𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑡𝑡 =(∆𝐴𝐴𝐴𝐴𝑖𝑖𝑡𝑡− ∆𝐴𝐴𝐶𝐶𝑖𝑖𝑡𝑡_𝐴𝐴− ∆𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶𝑖𝑖𝑡𝑡+ 𝐶𝐶𝑆𝑆𝑆𝑆𝑖𝑖𝑡𝑡)

𝑖𝑖𝑡𝑡−1

Where:

WC_ACC it = Non-cash working capital accruals for firm i in period t ∆CA it = The change in current SIZE for firm i in period t ∆CL it = The change in current liabilities for firm i in period t ∆Cash it = The change in cash for firm i in period t

∆STD it = The change in short-term debt for firm i in period t A it-1 = Total SIZE for firm i in period t-1

To incorporate reversals, they augment through the inclusion of a second partitioning variable that identifies periods in which the earnings management is hypothesized to reverse (PARTR):

𝑊𝑊𝐴𝐴_𝐴𝐴𝐴𝐴𝐴𝐴𝑖𝑖𝑡𝑡 = 𝑎𝑎 + 𝑏𝑏𝐷𝐷𝐴𝐴𝐴𝐴𝑆𝑆𝑖𝑖𝑡𝑡+ 𝑐𝑐𝐷𝐷𝐴𝐴𝐴𝐴𝑆𝑆𝑖𝑖𝑡𝑡+ ∑ 𝑓𝑓𝑘𝑘 𝑘𝑘𝐸𝐸𝑘𝑘𝑖𝑖𝑡𝑡+ 𝜀𝜀𝑖𝑖𝑡𝑡

For the purpose of conducting their evaluation of the model, Dechow considered three scenarios regarding the timing of the reversal of earnings management.

1. Assume that the researcher has no priors regarding the reversal of the earnings management, thus ignoring the coefficient on PARTR altogether. This scenario essentially collapses to the traditional model.

2. Assume that all earnings management reverses in the year immediately following the earnings management year. This seems to be a plausible assumption when considering working capital accruals, since most working capital accruals are expected to reverse within a year.

3. Assume that all earnings management reverses over the two years following the earnings management year.

Choosing the right model

Based on the models above, it is important to choose the right model to use for this research. All the models described above have some limitations. Criticism given on the Healy model is that the changes in non-discretionary accruals should not be equal to zero, because nondiscretionary accruals

(25)

can be sensitive to performance, Ronen and Yaari, (2008). Just like the Healy model, the DeAngelo model assumes that nondiscretionary accruals are constant, which is also a limitation of this model.

The Jones model assumes that revenues are nondiscretionary; when earnings are managed through discretionary accruals the Jones model will then be biased towards zero and will make an incorrect assumption that there is no case of earnings management, since the part of the managed earnings will be removed from the discretionary accrual proxy.

The modified Jones model has been used widely in the other studies, for example the study of Becker et al. (1998). By using the modified Jones model in this study a comparison can be made with other studies that have used the modified Jones model to detect earnings management, since this model have been widely used. Consequently in this research the modified Jones model will be used.

3.2 Hypothesis

This study investigates the use of earnings management in a global perspective within the Netherlands. As mentioned in the previous chapters, it’s expected that earnings management is applied more in globalized complex companies as a result the following research question will be used in this study:

“To what extent does the complexity of an entity plays a role in earnings management within Dutch public quoted companies?”

By means of Burgstahler et al (1997) it has been noted that earnings management is more marked in countries with a weaker legal system. In respect to Webb et al (2008) it has been securitized that there is a significant interaction between globalization and legal environment. Which showed that’s the effects of globalization are most distinctive for firm’s localized in weak legal environments. This effect is supported by the Barinov et al (2012) who shows the post-earnings-announcement drift for conglomerates. Therefore it’s expected that a company which is more complex have the tension to have more earnings management. Following Webb et al (2008) there are three measures for complexity; dummy variable for conglomerates, number of business segments and sales concentration based on the Herfindal index. The measures of sales concentration and number of business segments will be used in this research. Additionally a proxy for foreign sales will be used to consider whether foreign sales relates to earnings management itself. Reflection the analysis above this results in the following hypothesis are formulated:

H1: “Total foreign sales is positively correlated to discretionary accruals” as the company with foreign sales is depending on more cultural dimensions and legal systems, Doupnik (2008), Leuz (2003). H2: “The sales concentration of the company is positively correlated to discretionary accruals” following the research of Webb et al (2008) that the complexity of a company is conditional to the number of segments, measuring sales concentration of the business.

(26)

H3: “The number of operating units positively correlated to discretionary accruals” following the research of Webb et al (2008) that the complexity of a company is conditional to the number of segments, measuring the effect of the decentralisation of the business.

3.3 Methodology

The methodology will be divided into two stages. First, the discretionary accruals will be estimated. Second, the discretionary accruals will be regressed against the variables to determine their effect on earnings management. The discretionary accruals will be estimated with use of the Modified Jones Model. The Modified Jones model will be estimated cross-sectionally as in DeFond and Jiambalvo (1994), Subramanyam, (1996) and Bartov et al., (2000). The reason in choosing the cross-sectional approach over time series, is that time series estimation base the future events on known past events. With the cross-sectional approach groups can be compared with each other. Alternative explanation for choosing the cross-section over time series is that is has been widely used in prior research to estimate Modified Jones model.

Step 1: Determining total Accruals for each firm-year.

𝑆𝑆𝐴𝐴𝑡𝑡 = (∆𝐴𝐴𝐴𝐴𝑡𝑡− ∆𝐴𝐴𝐶𝐶𝑡𝑡− ∆𝐴𝐴𝐴𝐴𝐶𝐶𝐶𝐶𝑡𝑡+ ∆𝐶𝐶𝑆𝑆𝑆𝑆𝑡𝑡− 𝑆𝑆𝐷𝐷𝐷𝐷𝑡𝑡)/( (𝐴𝐴𝑡𝑡−1)

Where,

TAit = total accruals in year t, ΔCAt = change in current SIZE, ΔCLt = change in current liabilities,

ΔCASHt = change in cash and cash equivalents,

ΔSTDt = change in debt included in current liabilities, ΔDEPt = depreciation and amortization expense, At-1 = net total SIZE in year t-1

Step 2: Estimate the parameters: a1, a2 and a3.

The estimates of industry specific parameters will be used to calculate the non-discretionary accruals. 𝑆𝑆𝐴𝐴𝑡𝑡 = 𝛼𝛼1 �_𝐴𝐴1 𝑖𝑖𝑡𝑡−1� + 𝛼𝛼2 � ∆𝐴𝐴𝐷𝐷𝑅𝑅𝑖𝑖𝑡𝑡− ∆𝐴𝐴𝐷𝐷𝐴𝐴𝑖𝑖𝑡𝑡 𝐴𝐴𝑖𝑖𝑡𝑡−1 � + 𝛼𝛼2 � 𝐷𝐷𝐷𝐷𝐷𝐷𝑖𝑖𝑡𝑡 𝐴𝐴𝑖𝑖𝑡𝑡−1� + 𝜀𝜀𝑡𝑡 Where:

At-1 = total SIZE at t-1

ΔREVt = revenues in year t less revenues in year t-1

ΔRECt = net receivables in year t less net receivables in year t-1 PPEt = gross property, plant and equipment in year t

(27)

Step 3: Determining the amount of non-discretionary accruals for each firm-year 𝑁𝑁𝑆𝑆𝐴𝐴𝑡𝑡 = 𝛼𝛼1�_𝐴𝐴1 𝑡𝑡−1� + 𝛼𝛼2� ∆𝐴𝐴𝐷𝐷𝑅𝑅𝑡𝑡− ∆𝐴𝐴𝐷𝐷𝐴𝐴𝑡𝑡 𝐴𝐴𝑡𝑡−1 � + 𝛼𝛼3� 𝐷𝐷𝐷𝐷𝐷𝐷𝑡𝑡 𝐴𝐴𝑡𝑡−1 � where,

At-1 = net total SIZE in year t-1,

ΔREVt = change in revenue from year t-1 to year t,

ΔARt = change in accounts receivable from year t-1 to year t, PPEt = gross property plant and equipment in year t, and εit = error term in year t.

α1, α2, α3 = firm-specific parameters

The variables above have all been scaled by lagged total SIZE to reduce heteroscedasticity. Estimates of the α1, α2, and α3 are generated using the following model in the estimation period: 𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡 𝐴𝐴_{𝑖𝑖𝑡𝑡−1𝑡𝑡}= 𝛼𝛼 𝐴𝐴𝑖𝑖𝑡𝑡−1+ 𝛼𝛼1� 1 𝐴𝐴𝑡𝑡−1� + 𝛼𝛼2� ∆𝐴𝐴𝐷𝐷𝑅𝑅𝑡𝑡− ∆𝐴𝐴𝐷𝐷𝐴𝐴𝑡𝑡 𝐴𝐴𝑡𝑡−1 � + 𝛼𝛼3� 𝐷𝐷𝐷𝐷𝐷𝐷𝑡𝑡 𝐴𝐴𝑡𝑡−1 � + 𝑆𝑆𝐴𝐴𝑖𝑖𝑡𝑡 𝐴𝐴_{𝑖𝑖𝑡𝑡−1𝑡𝑡}+ 𝜀𝜀𝑖𝑖𝑡𝑡 Step 4: Calculating the amount of Discretionary Accruals

When total accruals (TAt) and non-discretionary accruals (NDAt) have been estimated, discretionary accruals (DAt) will be determined by subtracting NDAt from TAt.

𝑆𝑆𝐴𝐴𝑡𝑡 = 𝑆𝑆𝐴𝐴𝑡𝑡− 𝑁𝑁𝑆𝑆𝐴𝐴𝑡𝑡

The relationship between discretionary accruals and internationalisation will be investigated using ordinary least square (OLS) regression analysis.

3.4 Sample and data

The research analyse is based on financial data of Dutch companies listed on the Amsterdam Stock Exchange during the period 2001 until 2011. Using the databank of Datastream the research sample has been established; refer to details on sample to appendix III. The total data selection exist of 122 companies in the Netherlands, for the years 2000-2011, for 33 companies the data are incomplete in the database and are therefore excluded from the population. As the year 2000 is used as comparative information for the year 2001, the observations of 2000 are excluded from the selection.

Besides six companies has been excluded of the data set to correct the outliers, applied the 1% rule. So, the final sample exists of 10 year of 83 Dutch companies and 913 company-years observations during the period 2001 until 2011. For an overview of the companies include in the data refer to appendix III.

(28)

3.5 Control variables

Prior research has shown that there are many other factors that influence earnings management. The following control variables will be included, to better isolate the on earnings management: size, total SIZE, operating cash flow, growth and loss. Firm size (SIZE) is included according to the political cost hypothesis, which states that managers of large firms have a higher incentive than small firms to decrease reported earnings in order to reduce political attentions (Watts and Zimmerman, 1978). The relation between firm size and discretionary accruals is unclear, since they may have incentives to reduce reported earnings; they also have higher disclosure level than small firms (Koh, 2003; Lang and Lundholm, 1993). No prediction can be made between the relation of size and discretionary accruals, because of the ambiguous relationship. The variable size (SIZE) is measured as the natural logarithm of total SIZE. Operating cash flow (OCF) is an alternative for earnings to measure the performance of the entity. Particularly in a situation where the report earnings are close to zero the difference between the operating cash flow and reported earnings could be an explanatory factor in analysis the discretionary accruals, Hribar and Nichols (2007). In addition the growth (GROWTH) in revenue and reporting of losses (LOSS) are included in the regression model as a control variable.

3.6 Regression models

This section will connect the hypotheses with the regression models. First the connection of the Modified Jones model and the regression models will be described. The dependent variable of the regression models used below to test the hypotheses is estimated with use of the Modified Jones model. In paragraph 5.1 four steps were described. In the last step the non-discretionary accruals (NDA) are subtracted from the total accruals (TA) and separate the part of discretionary accruals (DA).

𝑆𝑆𝐴𝐴𝑡𝑡 = 𝑆𝑆𝐴𝐴𝑡𝑡− 𝑁𝑁𝑆𝑆𝐴𝐴𝑡𝑡

The discretionary accruals are than used as the dependent variable in the regression models, as earnings management.

Model 1

Hypothesis (H1) predicts a positively relation between foreign sales and discretionary accruals. This means that large foreign sales increase the tension of management to earnings management. The following regression model will be used to assess this relation:

𝐷𝐷𝐸𝐸 = 𝛽𝛽1(𝐶𝐶𝐶𝐶𝑡𝑡) + 𝛽𝛽2(𝐴𝐴𝐶𝐶𝐶𝐶𝐷𝐷𝑆𝑆𝐶𝐶𝑡𝑡) + 𝛽𝛽3(𝐶𝐶𝐴𝐴𝐶𝐶) + 𝛽𝛽4(𝐺𝐺𝐴𝐴𝐶𝐶𝑊𝑊𝑆𝑆𝐶𝐶) + 𝛽𝛽5(𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶) + 𝜖𝜖𝑡𝑡

Where, EM is the dependent variable estimated with use of the Modified Jones model used as the measure of earnings management. The variable FS stands for foreign sales, and should test the effect

(29)

of foreign sales on earnings management. The proxy for foreign sales (FS) is 0 or 1 as an assessment whether there are foreign sales.

Model 2

Hypothesis (H2) predicts a positively relation between complexity of the company and discretionary accruals. This means that large foreign SIZE increase the tension of management to earnings management. The following regression model will be used to assess this relation:

𝐷𝐷𝐸𝐸 = 𝛽𝛽1(𝐶𝐶𝐶𝐶𝐸𝐸𝑡𝑡) + 𝛽𝛽2(𝐴𝐴𝐶𝐶𝐶𝐶𝐷𝐷𝑆𝑆𝐶𝐶𝑡𝑡) + 𝛽𝛽3(𝐶𝐶𝐴𝐴𝐶𝐶) + 𝛽𝛽4(𝐺𝐺𝐴𝐴𝐶𝐶𝑊𝑊𝑆𝑆𝐶𝐶) + 𝛽𝛽5(𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶) + 𝜖𝜖𝑡𝑡

The variable HHI stands for complexity based on the Herfindahl index computed segment analysis. 𝐶𝐶𝐶𝐶𝐸𝐸 = � = 𝑠𝑠𝑖𝑖2

𝑁𝑁 𝑖𝑖

Is the number of segments, si is the fraction of total sales generated by segment i. Model 3

Hypothesis (H3) predicts a positively relation between the number of operating units of the company and discretionary accruals. This means that large foreign SIZE increase the tension of management to earnings management. The following regression model will be used to assess this relation:

𝐷𝐷𝐸𝐸 = 𝛽𝛽0 + 𝛽𝛽1(𝐴𝐴𝐶𝐶𝑁𝑁𝐺𝐺𝐶𝐶𝐶𝐶) + 𝛽𝛽2(𝐴𝐴𝐶𝐶𝐶𝐶𝐷𝐷𝑆𝑆𝐶𝐶𝑡𝑡) + 𝛽𝛽3(𝐶𝐶𝐴𝐴𝐶𝐶) + 𝛽𝛽4(𝐺𝐺𝐴𝐴𝐶𝐶𝑊𝑊𝑆𝑆𝐶𝐶) + 𝛽𝛽5(𝐶𝐶𝐶𝐶𝐶𝐶𝐶𝐶) + 𝜖𝜖𝑡𝑡

The variable CONGLO is a dummy for conglomerate. 1 if the firm is a conglomerate, 0 if not. The firm is a conglomerate if it has business segments in more than one two-digit SIC industries in the data.

(30)

4 Findings

Next section will describe the research statistics for the whole period between 2001 and 2011, divided in three sub sections. The first section is showing the data output of the modified jones model. The second paragraph presented the descriptive statics and output of the regression models.

4.1 Modified jones model

This paragraph described the output of the modified Jones model. The model is tested by using the data set of the Dutch listed companies during the period 2001 until 2011. Table 1 below presents the summary of the regression results of Dutch companies during the period 2001 until 2011.

Table 1 Summary of regression results for the sample of Dutch Companies during the period 2001 until 2011

Variable Coefficient t-statistic P-value

Intercept -0,0385 -6,9090 0,0000

1/A 352,2353 6,0513 0,0000

ΔREV 0,0081 2,2218 0,0265

PPE -0,0212 -3,2895 0,0010

A coefficient is significant when it’s P-value is below 5%. Al coefficients are significant in the model based on a 5% significance level. Additionally it can be noted that the total assets coefficient has the most explanatory power.

Table 2 Model summary of regression results for the sample of Dutch Companies during the period 2001 until 2011

R R Square Adjusted R Square Standard Error Observations

23,05% 5,31% 5,00% 0,10346 913

Table 2 shows the R and R Square which are respectively 23,05% and 5,31%. Also it has to be noticed that the R Square, which is the proportion of variance in the dependent variable which can be predicted form the independent variables has a value of 5,31%. That indicates that the majority for the sample of the Dutch companies have an explanatory power of 5,31% for the variation between the modified Jones model and the observed values.

With reference to previous studies performed the explanatory power is low, which can be explained through the fact that limited corrections has been made to the original data set, as for example elimination of financial service companies or specials events like mergers and acquisition. However the exclusion of this item will decrease the number of observations of foreign sales significant.

Table 3 shows the ANOVA table for the regression predictors displaying a significance lower than 0,1%. A predictor is significant when it its below 5%, concluding than the regression is significant.

(31)

Table 3 Model summary of ANOVA for the sample of Dutch Companies during the period 2001 until 2011

df SS MS F Significance F

Regression 3 0,5460 0,1820 17 0,00

Residual 909 9,7306 0,0107

Total 912 10,2766

Table 4 Model summary of descriptive statics for the sample of Dutch Companies during the period 2001 until 2011

Variable N Mean Std Dev. 25% Median 95%

FS 913 0,90 0,01 0,00 1,00 1,00 HHI 913 0,56 0,01 -0,83 0,64 1,00 CONGLO 913 0,53 0,01 0,00 0,60 1,00 OCF 913 0,09 0,00 -0,80 0,09 0,74 SIZE 913 5,82 0,04 2,85 5,80 9,12 DEBT 913 0,26 0,01 0,00 0,24 1,90 GROWTH 913 -0,18 0,03 -3,61 -0,07 5,25 LOSS 913 0,82 0,01 0,00 1,00 1,00

Table 4 provides descriptive statistics on the variables. Results show that that the majority of the variables are positive, whereby the mean of the SIZE (5,820) is relatively high. The control variable GROWTH has a negative mean. The descriptive statistics also reveal that firms with foreign sales are on average more positive than the variable complexity and conglomerates. It should be noted that on average, the control variables SIZE and GROWTH are more volatile than the other regression variables. The correlations in Table 5 illustrate the relationship between the variables.

FS is positively significant related to CONGLO and SIZE. This indicates that FS do not reduce the noise in SIZE to produce earnings, but rather make earnings more noisy estimates. HHI is only positive correlated to the CONGLO, which can be explained by reason that the nature of the data has a number of similarities.

COGNL is not significant correlated with OCF which is in-line with the research of Hribar and Nichols (2007) from which we learned that in a situation where the report earnings are close to zero the difference between the operating cash flow and reported earnings could be an explanatory factor in analysis the discretionary accruals. OCF is positively correlated to LOSS, this can be explained by the reason that a LOSS is commonly related to a negative operating cash flow.

Table 5 Model summary correlation statics for the sample of Dutch Companies during the period 2001 until 2011

FS HHI CONGLO OCF SIZE DEBT GROWTH LOSS

FS 1,00 HHI -0,51 1,00 CONGLO 0,59 0,27 1,00 OCF 0,13 0,03 0,11 1,00 SIZE 0,34 0,10 0,51 0,02 1,00 DEBT -0,02 0,05 0,01 0,00 0,17 1,00 GROWTH 0,01 0,04 0,10 -0,19 0,28 0,29 1,00 LOSS 0,12 0,03 0,13 0,38 0,14 0,06 -0,01 1,00

(32)

4.2 Testing the models

The regression in table 6 examines the relationship of FS (model 1), HHI (model 2) and CONGLO (model 3) with earnings management. The adjusted R2_{value for the relationship between the foreign} sales and earnings management is 38,23%. Conform the expectation the coefficient FS is positive (-0.0489) and significant at the 5% significance level. HHI is also positively related (0,0259) but not significant. CONGLO is contrary to the expectation negatively (-0,-0259) and only significant on a 10% significant level.

Table 6 regression results model 1,2,3 for the sample of Dutch companies during the period 2001 until 2011

Variable Expected sign Coefficient t-statistic P-value

Intercept - -0,1633 -6,634 0,000 FS + 0,0489 2,242 0,025 HHI + 0,0259 1,387 0,166 CONGLO + -0,0396 -1,983 0,048 OCF - -0,5198 -18,766 0,000 SIZE ? 0,0084 2,860 0,004 DEBT ? -0,0127 -0,854 0,393 GROWTH ? -0,0124 -3,939 0,000 LOSS ? 0,1499 19,348 0,000 n 913 Adjusted R2 _38,23% Testing model 1

Hypothesis one predicted that foreign sales are positively related to discretionary accruals, hence foreign sales will increase earnings management.

H1: “Total foreign sales is positively correlated to discretionary accruals”

The result does support the hypothesis that earnings management is positively related to earnings management. The results suggest that earnings management not increasing with the level of foreign sales. The coefficient of the control variable OCF (-0,5198) is significantly negative. This suggests that firms with high level of operating cash flows have a negative effect on earnings management. This finding is not in line with the theory provided in prior literature of Hribar and Nichols (2007) which considered that high cash flows increase the number of discretionary accruals. Coefficient SIZE (-,00084) is significant and positive. No predictions are made on this variable, given the fact that this is a significant factor in the regression it’s revealed that the size of the company is negatively related to earnings management. DEBT (-0,0127) negative but not significant, no contribution to the regression model. Coefficients GROWTH (0,0124) is significantly negative related to the regression, LOSS is positive and significant, no predictions are made on this control variables.

Earnings management in a global environment