The effect of IFRS adoption and institutional characteristics on accrual-based earnings management and real activities manipulation

The Effect of IFRS Adoption and Institutional

Characteristics on Accrual-based Earnings

Management and Real Activities

Manipulation

Name:

Liselotte Willig

Student number:

6045170

Final version:

June 19

₂₀₁₄

First supervisor:

dr. Alexandros Sikalidis

Second examiner: dr. Bo Qin

Master thesis Accountancy & Control, variant Accountancy, 2013/2014

Faculty of Economics and Business, University of Amsterdam

Table of Content

2.2.1 Accrual-based Earnings Management p. 8 2.2.2 Real Activities Management p. 9

2.2.3 IFRS p. 9

2.2.4 Legal Institutions p. 9

2.2.5 Investor Protection p. 10

2.3 Hypothesis Development p. 10

3 Research Methodology

3.1 Data and Sample Description p. 12

3.2 Variable Estimations

3.2.1 Variables of Interest p. 13 3.2.2 Discretionary Accruals p. 13 3.2.3 Real Activities Manipulation p. 14

3.3 Empirical Models p. 14

4 Empirical Results

4.1 Descriptive and Univariate Analysis p. 16

4.2 Multiple Regression Analysis p. 24

5 Robustness Check p. 28

6 Conclusion p. 29

7 References p. 31

8 Appendix A

8.1 Discretionary Accruals p. 33

8.2 Real Activities Manipulation p. 33

The Effect of IFRS Adoption and Institutional

Characteristics on Accrual-based Earnings

Management and Real Activities Manipulation

Abstract

This study examines the effects of the adoption of International Financial Reporting Standards (IFRS) and institutional characteristics on both accrual-based and real activities manipulation. Especially the institutional characteristics legal system, investor protection by the level of legal enforcement, and the importance of the equity market are important in this analysis. While prior literature has mainly examined the effects of IFRS adoption and institutional characteristics on accrual-based earnings management, no study has focused on the effects of adopting IFRS combined with institutional characteristics on real activities manipulation. Using a sample of 5429 firm-year observations from 22 members of the European Union, between 1999 and 2012, this study applies a univariate and a multiple regression analysis. The results suggest that the adoption of IFRS had a significant effect on real activities manipulation, and when we control for the recent financial crisis also on accrual-based earnings manipulation. Mixed results for the effect of institutional characteristics suggest the level of legal enforcement and the importance of the equity market have a significant effect on the level of accrual-based earnings management, and the legal system and the importance of the equity market have a significant effect on the level of real activities manipulation. In addition, our findings suggest the existence of a negative relationship between accrual-based earnings manipulation and real activities manipulation.

1.

This paper starts with examining whether the mandatory implementation of International Financial Reporting Standards (IFRS) in 2005 in the European Union (EU) has an impact on the method used of managing earnings, accrual-based or real earnings manipulation. And then especially whether institutional characteristics affect the level of accrual-based earnings management and real activities manipulation despite the adoption of a common set of standards. Especially the institutional characteristics legal system, investor protection and importance of the equity market are important in this analysis. Two different proxies for earnings management are used, namely: (1) discretionary accruals and (2) real activities manipulation.

Prior research mainly focused on the exploitation of within Generally Accepted Accounting Principles (GAAP) discretion over the level of accruals (Walker 2013). Following Healy and Wahlen (1999), accrual based earnings management is defined as the alteration of firms’ reported economic performance by insiders to either mislead some stakeholders or to influence contractual outcomes. Less attention in literature is drawn to earnings management through management of operational activities, namely real earnings management. However, Graham, Harvey and Rajgopal (2005) provide evidence through a survey that far more attention needs to be paid to real earnings management than had previously been the case. Managers would rather take economic actions that could have negative long-term consequences, than make within-GAAP accounting choices to manage earnings (Graham et al., 2005). Following Roychowdhury (2006), real earnings management is defined as management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds.

Also the effects of IFRS adoption on earnings management are attempted to assess in the literature. The implementation of IFRS can be seen as a tighter set of accounting standards relatively to prior national accounting standards. Regulators expect to enhance the comparability of financial statements, improve corporate transparency, and increase the quality of financial reporting by implementing IFRS (European Commission, 2002). If so, this would lead to a decrease in accrual-based earnings management practices.

Although accounting standards become tighter, earnings management remains (Jeanjean & Stolowy, 2008; Ewert & Wagenhofer, 2005; Cohen, Dey and Lys, 2008; Cohen, Dey, Lys and Zarowin, 2010, Zang, 2012; Doukakis, 2013). This indicates the

need for further investigation of the impact of mandatory IFRS adoption on earnings management. Accounting standards only play a limited role in determining observed reporting quality. Many forces shape the quality of financial reporting, and accounting standards should be viewed as one of those forces (Holthausen, 2009). According to Jeanjean & Stolowy (2008) the mandatory introduction of IFRS standards did not have an impact on the pervasiveness of accrual-based earnings management. Accounting standards make it more costly to apply accrual-based earnings management. Because this causes an increase in the marginal benefit of real earnings manipulation, managers will increase their use of real activities manipulation (Ewert & Wagenhofer, 2005; Cohen et al., 2008). Manager’s trade-off decisions between accrual-based and real earnings management are influenced by costs and timing of earnings management activities (Zang, 2012). Taking these studies together, managers are trading off earnings management methods based on their relative costliness.

The application of standards involves judgement and underlying measurements are often based on private information (Watts & Zimmerman, 1986). In addition to regulation, institutional characteristics seem to be of great influence on earnings management (Ball, Kothari and Robin, 2000; Daske, Gebhardt and McLeay, 2006; Leuz, Nanda and Wysocki, 2003; Ball, 2006). Like any set of other accounting standards, IFRS involves considerable judgement and the use of private information. Institutional factors shape firms’ incentives to report informative earnings (Burgstahler, Hail and Leuz, 2006).

Demand for accounting earnings is systematically different depending on national legal institutions. Countries can be classified in code-law countries or common-law countries. According to LaPorta, Lopez-de-Silanes, Shleifer and Vishny (2000) differences among legal origins are best described by the proposition that some countries protect all outside investors better than others. LaPorta et al. (2000) suggest common-law countries have the strongest protection of outside investors, whereas code-law countries have the weakest protection. Earnings management and loss avoidance practices appear to be more prevalent in companies from code-law countries compared to companies from common-law countries (Ball et al., 2000; Daske et al., 2006). This can be explained by greater private control benefits and stronger incentives to obfuscate firm performance in case of weak outside investor protection. The level of outside investor protection depends on several factors, namely the equity market development, ownership structures, investor rights and legal enforcement. The results of Leuz et al. (2003) indicate that countries with developed equity markets, dispersed

ownership structures, strong investor rights and legal enforcement engage in less accrual based earnings management.

Because the role of accounting standards is only limited and there exists a trade-off in the way managers practice earnings management, it is relevant to investigate whether there exist a change in the use of accrual-based earnings management and real activities manipulation due to the implementation of IFRS, and whether institutional characteristics play an important role in this change. We therefore propose accrual-based earnings management will decrease and real earnings management will increase across the adoption, and that institutional characteristics will affect the level of accrual-based and real earnings management despite the adoption of IFRS.

This investigation is based on the EU before and after the implementation of IFRS. The focus lays on the EU for several reasons. First, the EU has a relatively strong legal system and enforcement regime (Li, 2010). However, although being member of the EU, differences in institutional factors such as a nation’s traditional law remain. This makes the focus on members of the EU an interesting choice. Second, harmonization efforts within the EU have largely focused on eliminating differences in accounting standards across countries by adopting a common set of standards (Van Hulle, 2004). The IFRS adoption in 2005 is regarded as an important milestone towards achieving a common EU market (Tweedie, 2006). Also the EU seems particularly suitable for an international analysis. As sample sizes are similar, the EU is large enough to draw comparisons with prior evidence from the United States (US) (Daske et al., 2006).

Our analysis is based on financial accounting data from 1999 to 2012 for over 845 unique firms, which consist of 5429 firm-year observations, from 22 members of the EU. To measure the pervasiveness of earnings management of a firm, we follow Kim, Park and Wier (2012) in estimating the two models of accrual-based earnings management and real earnings management. Following the model developed by Dechow, Kothari and Watts (1998) as implemented in Roychowdhury (2006) we calculate four proxies for real earnings manipulation. Investor protection is measured by the quality of legal enforcement, which consists of the mean of an estimation of control of corruption, rule of law and government effectiveness. The legal system of a country is determined by the historical origin of their laws (LaPorta, Lopez-de-Silanes, Shleifer and Vishny, 2002). The importance of the equity market is measured by the number of listed domestic companies per country divided by its population.

Prior research has pointed out the importance of understanding how firms manage earnings and how managers make trade-off decisions between accrual-based

earnings and real activities manipulation with respect to accounting standards and institutional characteristics (Ball et al., 2000; LaPorta et al., 2000; Leuz et al., 2003; Ewert & Wagenhofer, 2005; Daske et al., 2006; Roychowdhury, 2006; Cohen et al., 2008, 2010; Holthausen, 2009; Jeanjean & Stolowy, 2008; Zang, 2012). To my knowledge, it has not been investigated yet whether the implementation of IFRS in 2005 caused a change in the use of earnings management, both accrual-based and real manipulation, and whether this is influenced by institutional characteristics as a country’s legal system, the level of legal enforcement and the level of importance of the equity market. This study aims to fill in the gap in the IFRS, institutional characteristics and earnings management literature by examining the impact of mandatory IFRS adoption and institutional characteristics on both accrual-based and real earnings management. In this respect, the study contributes to the regulatory issue of accounting harmonization and the debate on accounting convergence.

To examine whether the mandatory implementation of IFRS has an impact on the method used of managing earnings and if institutional characters play a significant role in this, we undertake a univariate and a multiple regression analysis. Our results suggest that mandatory IFRS adoption has a significant impact on the level of real activities manipulation, and when we control for the financial crisis also for the level of accrual-based earnings manipulation. Firms located in common-law countries make more use of real activities manipulation compared to firms located in code-law countries. Also the importance of the equity market has a significant affect on the level of earnings management, both accrual-based and real activity-based. Although legal enforcement affects the level of accrual-based earnings management significantly, it is not the case for the level of real activities manipulation.

This study builds on recent advances in the earnings management literature on the role of the mandatory IFRS adoption and institutional characteristics (Ball et al., 2000, 2003, 2006; LaPorta et al., 2000; Leuz et al., 2003; Ewert & Wagenhofer, 2005; Jeanjean & Stolowy, 2008; Li, 2010; Doukakis, 2013). We extend this literature in several respects. First, this is the first study to provide an examination of the role that institutional characteristics play in determining the level of accrual-based and real activities manipulation of firms after the adoption of IFRS. Second, different from most prior earnings management studies this study includes real activities manipulation. Finally, evidence from this study can help standard-setters and regulators better understand the effect of institutional factors on earnings management, despite the adoption of a common set of standards.

The remainder of the paper will be organized as follows; Section 2 presents the literature review. In Section 3 the research methodology is described. The empirical tests and results are presented in section 4. Section 5 presents the robustness test and Section 6 concludes the study.

2. Literature Review

2.1 Agency Theory

The relationship between stockholders (outsiders) and managers (insiders) of a corporation fits the definition of a pure agency relationship. Jensen & Meckling (1976) define an agency relationship as a contract under which one or more persons (the principal(s)) engage another person (the agent) to perform some service on their behalf, which involves delegating some decision-making authority to the agent. There are issues associated with this type of “separation of ownership and control”, the so-called principal-agency problem.

The principal-agency problem holds the divergence between shareholders decisions and those decisions, which would maximize the welfare of the manager. From a conflict of interest between firms’ insiders and outsiders arise incentives to misrepresent firm performance or real manipulation of performance (Leuz et al., 2003). Earnings management is one way of expropriation by managers and can be done by discretionary accounting choices of accruals or real activities manipulation (Roychowdhury, 2004; Walker, 2013).

2.2 Definitions

2.2.1 Accrual-based Earnings Management

According to Walker (2013) the exploitation of within Generally Accepted Accounting Principles (GAAP) discretion over the level of accruals is by far the most studied earnings management method. Earnings management trough accruals occurs when managers use judgement in financial reporting and in structuring transactions to alter financial reports to either mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that

depend on reported accounting numbers (Healy & Wahlen, 1999). Accounting depreciation rates; deferred tax assumptions; stock (inventory) valuation assumptions; bad debt provisions and revenue recognition are all examples of discretionary accounting choices (Walker, 2013).

2.2.2 Real Activities Management

Manipulating earnings can be done in several ways. It does not only refer to exercising judgement in the accounting process, but also to real activities manipulation. Examples include speeding up sales or cutting expenses etcetera to increase earnings.

Real activities manipulation is defined as management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds (Roychowdhury, 2006). Because real activity manipulation imposes real costs on firms, it is necessary to make a distinction between the two. Also real activity manipulation has a stronger negative effect on performance than accrual based earnings management does (Graham et al., 2005; Cohen et al., 2010).

2.2.3 IFRS

IFRS are accounting rules (‘standards’) issued by the International Accounting Standards Board (IASB). The IASB purport to be a set of rules that ideally would apply equally to financial reporting by public companies worldwide. Under its constitution, the IASB is a standard-setter and does not have an enforcement mechanism for its standards. It cannot penalise individual companies or countries that adopt its standards, but in which financial reporting practice is of low quality because managers, auditors and local regulators fail to fully implement the standards. (Ball, 2006)

2.2.4 Legal Institutions

Demand for accounting earnings is systematically different depending on national legal institutions. Countries can be classified as either a code-law or a common-law country. In general, differences among legal origins are best described by the proposition that some countries protect all outside investors better than others (LaPorta et al., 2000).

Common-law takes its name from the process whereby laws originate: its pure form, common law arises from what is commonly accepted to be appropriate practice. Common law originated in England and spread to its former colonies. It tends to be more market-oriented, supports a proportionately larger listed corporate sector, is more litigious, tends to presume that investors ‘at arm’s-length’ from the company, and

hence is more likely to presume that investors rely on timely public disclosure and financial reporting. Financial reporting practice (and rules) emphasises timely recognition of losses in the financial statements. (Ball, 2006)

Code-law countries generally are less market-oriented, have proportionately larger government and unlisted private-company sectors, are less litigious and are more likely to operate an ‘insider access’ model with less emphasis on public financial reporting and disclosure. There is less emphasis on timely recognition of losses in the public financial statements, and earnings have lower volatility and lower informativeness. (Ball, 2006)

2.2.5 Investor Protection

When investors finance firms, they typically obtain certain rights or powers that are generally protected through the enforcement of regulations and laws. Some of these rights include disclosure and accounting rules, which provide investors with the information they need to exercise other rights. In different jurisdictions, rules protecting investors come from different sources, including company, security, bankruptcy, takeover, and competition laws, but also from stock exchange regulations and accounting standards. (La Porta et al. 2000)

Investor protection turns out to be crucial because, in many countries, expropriation of minority shareholders and creditors by the controlling shareholders is extensive. When outside investors finance firms, they face a risk, and sometimes near certainty, that the returns on their investments will never materialize because the controlling shareholders or managers expropriate them. (La Porta et al. 2000) In general, expropriation is related to the agency problem described by Jensen and Meckling (1976), who focus on the consumption of “perquisites” by managers and other types of empire building. Enforcement is as crucial as their contents.

2.3 Hypothesis Development

The mandatory adoption of IFRS in 2005 brings mixed expectations. Regulators expect the use of IFRS to enhance the comparability of financial statements (European Commision; 2002). The European Commission (2002) suggests that mandatory adoption of IFRS reporting leads to significant benefits in terms of accounting quality often start from the premise that IFRS reporting increases transparency and improves the comparability of financial reporting. IFRS could be expected to reduce information

asymmetry between insiders and outsiders, because IFRS are more detailed, require more disclosures, allow fewer options and are broader in scope compared to many national accounting standards. Therefore it should lead to a decrease in accrual-based earnings management practices. This brings us at the first hypothesis:

H1: Mandatory IFRS adoption is associated with a decrease in accrual earnings

management practices.

Although accounting standards become tighter, earnings management remains (Ewert & Wagenhofer, 2005; Jeanjean & Stolowy, 2008; Cohen et al., 2008; Zang, 2012; Doukakis, 2013). Besides earnings management through discretionary accruals, earnings management can be done through manipulating real activities. According to Graham et al. (2005) managers would rather take economic actions that could have negative long-term consequences, than make within-GAAP accounting choices to manage earnings.

According to Zang (2012) a direct substitutive relation exists between real activities manipulation and accrual-based earnings management. Managers fine-tune their accrual accounts based on the outcomes of real activities manipulation. The manager’s trade-off decision between accrual-based and real earnings management are influenced by their relative costliness. The implementation of IFRS in 2005 makes it more costly to apply accrual-based earnings management. Because this causes an increase in the marginal benefit of real earnings manipulation, managers will increase their use of real manipulation (Ewert & Wagenhofer, 2005; Cohen et al., 2008; Zang, 2012). This brings us at the second hypothesis:

H2: Mandatory IFRS adoption is associated with an increase in real earnings

management practices.

The harmonization efforts within the EU have largely focused on eliminating differences in accounting standards across countries by adopting a common set of standards (Van Hulle, 2004). The adoption of IFRS is regarded as an important step towards achieving a common EU market (Tweedie, 2006). Although the relatively strong legal system and enforcement regime of the EU, differences in institutional characteristics remain (Li, 2010).

From an economic perspective there are reasons to be sceptical about the premise the adoption of IFRS will lead to a decrease in earnings management. The role

of accounting standards in influencing earnings management practices may be limited relative to other forces, such as institutional factors (Holthausen, 2009; Doukakis, 2013). Powerful economic and political forces determine how managers, auditors, courts, regulators and other parties influence the implementation of rules (Ball, 2006). According to Jeanjean & Stolowy (2008), the underlying argument is that the application of accounting standards involves considerable judgement and the use of private information. Countries’ legal institutional structures play an important role in explaining accounting quality (Ball et al., 2000, 2003; Leuz et al., 2003; Burgstahler et al., 2006; Jeanjean & Stolowy, 2006). Ball et al. (2000) argue that demand for accounting earnings is systematically different between code-law and common-law countries, which can be explained by the level of outside investor protection (LaPorta et al., 2000). All the above arguments lead to the formation of the third and last hypothesis:

H3: Institutional characteristics affect the level of earnings management despite the

adoption of IFRS.

3. Research Design

3. 1 Data and sample description

Financial firm-level data is collected from 1999 to 2012 from the Worldscope Database via Datastream. This database contains historical financial data from annual reports of publicly traded companies around the world. Financial country-level data is collected from 2000-2012 from the World Development Indicators via WorldBank. The initial sample consists of all publicly listed companies in EU member states that mandatorily adopted IFRS in 2005. The study excludes financial institutions (ICB codes 8000 – 8999) because characteristics of accruals differ in these firms and the possibility that earnings figures are affected by specific accounting requirements, government intervention and regulation. Also firm-years that lack accounting information are excluded from the sample. For these reasons Estonia, Latvia, Lithuania, Malta and Slovakia are excluded from the sample.

Because the focus of this study is on the adoption of IFRS in Europe, the sample is divided in two groups. The pre-adoption period constitutes of the years 2001-2004 and the post-adoption period constitutes of the years 2006-2012. The year 2005 is

excluded from the multiple regression analysis, because the estimation of real earnings management proxy AB_PROD, requires two lags of data under the same standard for the sales variable. Moreover, this exclusion removes any adoption year effect.

Governance data is derived from the Worldwide Governance Indicators via WorldBank. This data includes Rule of Law, Control of Corruption, Governmental Effectiveness and Regulation Quality. Legal System data is derived from the JuriGlobe – World Legal Systems from the University of Ottawa.

3.2 Estimations

3.2.1 Variables of Interest: Investor Protection by Legal Enforcement and Importance of the Equity Market

Following La Porta et al. (1997, 1998) the Legal Enforcement measure for each country is the average score across three variables, namely (1) an estimate of control of corruption; (2) an estimate of the rule of law, and (3) an estimate of the government effectiveness. All three variables range from -2.5 to 2.5.

The number of listed domestic firms relative to the population measures the Importance of the Equity Market.

3.2.2 Discretionary Accruals

As a proxy for accrual-based earnings management discretionary accruals are used. As in DeFond & Subramanyam (1998) and Kim et al. (2012) we use a cross-sectional version of the modified Jones model due to its superior specification and less restrictive data requirements. Following Kothari, Leone and Wasley (2005) and Kim et al. (2012) return on investment (ROA) is included in the prior year as a regressor in the estimation model to control for the effect of performance on measured discretionary accruals (see Appendix A for details). Because earnings management can involve either income increasing or income decreasing accruals, the absolute value of discretionary accruals (ABS_DA) is used for our main analyses.

3.2.3 Real Activities Manipulation

Following Kim et al. (2012) we use the following four measures to detect real activities manipulation: (1) abnormal levels of operating cash flow (AB_CFO), (2) abnormal production costs (AB_PROD), (3) abnormal discretionary expenses (AB_EXP), and (4) a combined measure of real activities manipulation (COMBINED_RAM). These

proxies rely on prior studies, namely Roychowdhury (2006), Cohen et al. (2008), Cohen & Zarowin (2010), Badertscher (2011) and Zang (2012). Abnormal levels of the first three real activities manipulation measures are measured as the residual from the relevant models estimated by year (see Appendix A for details).

Following Cohen et al. (2008) and Kim et al. (2012), three individual proxies are used as well as a combined proxy (COMBINED_RAM). Considering the expected directions of the first three variables, we calculate COMBINED_RAM as AB_CFO – AB_PROD + AB_EXP. Following Kim et al. (2012) the combined real activities manipulation proxy decreases as firms engage in more aggressive earnings management through real activities. The higher AB_PROD, the larger is the amount of inventory overproduction and the greater is the increase in reported earnings through reducing the cost of goods sold.

3.3 Empirical Models

Following Kim et al. (2012) we estimate the following two models to capture the relation between accrual-based earnings management and real earnings management and investor protection:

𝐴𝐴𝐴𝐴𝐴𝐴_𝐷𝐷𝐴𝐴𝑡𝑡 = 𝛼𝛼0𝐿𝐿_𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 + 𝛼𝛼1𝑃𝑃𝑃𝑃𝐴𝐴𝑃𝑃2005 + 𝛼𝛼2𝐿𝐿_𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡∗ 𝑃𝑃𝑃𝑃𝐴𝐴𝑃𝑃2005 + 𝛼𝛼3𝐿𝐿_𝐸𝐸𝐸𝐸𝐸𝐸𝑃𝑃𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑃𝑃𝑡𝑡+ 𝛼𝛼4𝐼𝐼𝐸𝐸𝑃𝑃𝑃𝑃𝐸𝐸𝑃𝑃𝐴𝐴𝐸𝐸𝐸𝐸𝐸𝐸_𝐸𝐸𝐸𝐸𝑡𝑡 + 𝛼𝛼5𝐸𝐸𝑃𝑃𝐸𝐸𝐴𝐴𝐼𝐼𝐸𝐸𝐸𝐸𝐷𝐷_𝐸𝐸𝐴𝐴𝐸𝐸𝑡𝑡+ 𝛼𝛼6𝐴𝐴𝐸𝐸𝑡𝑡−1+ 𝛼𝛼7𝐴𝐴𝐷𝐷𝐴𝐴_𝐸𝐸𝑃𝑃𝐴𝐴𝑡𝑡−1+ 𝛼𝛼8𝐿𝐿𝐸𝐸𝐿𝐿𝑡𝑡−1 + 𝛼𝛼9𝐸𝐸𝐷𝐷_𝐼𝐼𝐸𝐸𝑃𝑃𝑡𝑡+ 𝜀𝜀𝑡𝑡; 𝐸𝐸𝐴𝐴𝐸𝐸_𝑃𝑃𝐸𝐸𝑃𝑃𝑃𝑃𝑃𝑃𝑡𝑡= 𝛼𝛼0𝐿𝐿_𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡 + 𝛼𝛼1𝑃𝑃𝑃𝑃𝐴𝐴𝑃𝑃2005 + 𝛼𝛼2𝐿𝐿_𝐴𝐴𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑡𝑡∗ 𝑃𝑃𝑃𝑃𝐴𝐴𝑃𝑃2005 + 𝛼𝛼3𝐿𝐿_𝐸𝐸𝐸𝐸𝐸𝐸𝑃𝑃𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝑃𝑃𝑡𝑡+ 𝛼𝛼4𝐼𝐼𝐸𝐸𝑃𝑃𝑃𝑃𝐸𝐸𝑃𝑃𝐴𝐴𝐸𝐸𝐸𝐸𝐸𝐸_𝐸𝐸𝐸𝐸𝑡𝑡+ 𝛼𝛼5 𝐴𝐴𝐴𝐴𝐴𝐴_𝐷𝐷𝐴𝐴𝑡𝑡 + 𝛼𝛼6𝐴𝐴𝐸𝐸𝑡𝑡−1+ 𝛼𝛼7𝐴𝐴𝐷𝐷𝐴𝐴_𝐸𝐸𝑃𝑃𝐴𝐴𝑡𝑡−1+ 𝛼𝛼8𝐿𝐿𝐸𝐸𝐿𝐿𝑡𝑡−1+ 𝛼𝛼9𝐸𝐸𝐷𝐷_𝐼𝐼𝐸𝐸𝑃𝑃𝑡𝑡 + 𝜀𝜀𝑡𝑡; where:

ABS_DA = absolute value of discretionary accruals (signed discretionary

accruals), where discretionary accruals are computed through the cross-sectional modified Jones model adjusted for performance;

RAM_PROXY = AB_CFO, AB_PROD, AB_EXP, or COMBINED_RAM: AB_CFO = the level of abnormal cash flows from operations;

(1)

AB_PROD = the level of abnormal production costs, where production costs are defined as the sum of cost of goods sold and the change in inventories;

AB_EXP = the level of abnormal discretionary expenses, where

discretionary expenses are the sum of R&D expenses and SG&A expenses;

Variables of Interest:

L_System = binary variable to classify countries as either code-law or common-law, takes the value of zero for code-law countries and one for common-law countries;

POST2005 = binary variable that equals one for observations after 2005, to capture any changes in earnings management that are

independent of the classification of legal system;

L_System*POST2005 = interaction term, takes the value of one for common-law countries in the post-adoption period, to capture any incremental change in earnings management for common-law countries compared to code-common-law countries.

L_Enforcement = measured as the mean score across three legal variables from the Worldwide Governmental Indicators

(WorldBank); (1) the control of corruption, (2) an assessment of rule of law, and (3) the governmental effectiveness;

Importance of Equity Market = measured as the number of Listed Domestic Companies divided by the Population;

Control variables:

BM = book-to-market equity ratio, measured as BVE/MVE, where BVE is the book value of equity and MVE is the market value of equity; ADJ_ROA = adjusted ROA, where ROA is measured as income before

extraordinary items, scaled by lagged total assets;

LEV = total capital to total asset ratio determines whether a company has sufficient capital to support its assets;

RD_INT = R&D intensity (R&D expenses/net sales) for the year;

Equations (1) and (2) are estimated with multiple regressions. It is possible for firms to use a mix of accrual based earnings management and real manipulation. However, according to Cohen et al. (2008) and Zang (2012) firms will use the technique that is less costly to them. Zang (2012) finds that managers’ trade off the two earnings

management methods based on their relative costs and that managers adjust the level of accrual-based earnings management according to the level of real activities manipulation realized. Cohen et al. (2008) control for the substitutive nature of these types of earnings management by including the proxy for accrual-based earnings management (ABS_DA) as a control variable in the real activities manipulation (RAM_PROXY) regressions. Also the accrual-based earnings management (ABS_DA) regressions are controlled for this substitutive nature by including the proxy for real activities manipulation (COMBINED_RAM). We follow this method of Cohen et al. (2008) to control for the substitutive nature of these two earnings management methods.

Various control variables are included in the regressions to avoid the problem of correlated omitted variables. According to Roychowdhury (2006) and Summers & Sweeney (1998) firm-specific growth opportunity can potentially explain significant variation in earnings management. Therefore a proxy for growth opportunities, namely BM is included. Hall & Oriani (2006) find that stronger investor protection could reduce the reinforcing effect of R&D investments on potential insider expropriation and consequently increase R&D valuation by financial markets. Therefore R&D Intensity is included to control for a firm’s R&D expenditure. Following Kim et al. (2012) a proxy for leverage, LEV, is included to control for the leverage incentives for earnings management. To be able to isolate the effect of investor protection on earnings management industry ROA (ADJ_ROA) is included in the regressions to control for the potential effect of financial performance.

4. Empirical results

4.1 Descriptive Statistics and Univariate Analysis

Table 1 presents the distribution of the test sample by country. The overall sample includes 5429 firm-year observations (845 unique firms) from 22 countries with significant representation from the United Kingdom and France. The test sample includes 426 firm-year observations of the year 2005. The sample consists of 4237 (78%) code-law countries and 1192 (22%) common-law countries.

Table 1

Sample Description: Distribution of Firm-Year Observations by Legal System Country Firm-year observations

Code-law Common-law % of Sample

Austria 190 3.50 Belgium 201 3.70 Bulgaria 4 0.07 Croatia 13 0.24 Cyprus 3 0.06 Czech Republic 21 0.39 Denmark 245 4.51 Finland 222 4.09 France 607 11.18 Germany 1364 25.12 Greece 138 2.54 Hungary 40 0.74 Ireland 63 1.16 Italy 358 6.56 Luxembourg 11 0.20 Netherlands 281 5.18 Poland 31 0.57 Portugal 20 0.37 Romania 27 0.50 Slovenia 12 0.22 Spain 128 2.36 Sweden 321 5.91 United Kingdom 1129 20.56 Total 4237 1192 100

Table 2 presents descriptive statistics relating to variables used in the analysis (except for LS_SYSTEM since this variable is a dummy variable). To explore potential differences a univariate comparison across the pre- and post-IFRS adoption period for the full sample is included in Tabel 2. Years 2001 – 2004 constitute the pre-adoption period and years 2006 – 2012 constitute the post-adoption period. All continuous variables are winsorized at the top and bottom 1% of their distributions to mitigate the influence of outliers.

The means of ABS_DA, AB_CFO and AB_EXP are not significantly different across the pre- and post-IFRS adoption periods, whereas the means of AB_PROD and COMBINED_RAM have significantly changed. The mean of AB_PROD is significantly higher after 2005 and the mean of COMBINED_RAM is significantly lower after 2005. Following Kim et al. (2012) this finding indicates firms engage in more aggressive earnings management through real activities. Moreover, it is consistent with our second hypothesis in which we state the adoption of IFRS is associated with an increase in real earnings management practices. Although the mean of AB_PROD remains negative, it

Table 2

Descriptive Statistics: Selected Variables of Full Sample Full sample

Mean Difference Tests : p-value

N Pre-2005 Post-2005 t-test

Dependent Variables ABS_DA 5003 -.00271 -.00227 .8271 AB_CFO 5003 .00339 .00449 .6829 AB_EXP 5003 .00285 .00018 .6516 AB_PROD 4880 -.03870 -.01721 .0006*** COMBINED_RAM 4880 .04435 .01999 .0484** Variables of Interest L_Enforcement 4631 1.63601 1.55038 .0000*** IM_EQ 5003 .00002 .00002 .0751* Control Variables B/M 4747 .01553 .01448 .4618 ADJ_ROA 5003 .07295 .09622 .5933 LEV 4631 .14812 .17246 .0129** RD_INT 3140 .07725 .06397 .1329

The t-test tests the null hypothesis that the mean difference on the variable is zero.

*** Indicates statistical significance at the 1% level, ** indicates statistical significance at the 5% level, and * indicates statistical significance at the 10% level.

significantly increased with a value of 0.02149 relative to the years before the adoption of IFRS. The mean of the combined measure of real earnings manipulation, COMBINED_RAM, decreased significantly with a value of 0.02436. Also the mean of L_Enforcement is significantly lower across the pre- and post-IFRS periods. To be more specific, the mean decreased with a value of 0.08563 and a significance level of 1%. The mean of IM_EQ remained significantly the same and the mean of LEV increased significantly. In sum, these results suggest that firms are more likely to engage in more aggressive earnings management through real activities after the adoption of IFRS.

Table 3 presents Pearson correlation coefficients for selected variables. L_System is positively and significantly correlated with AB_CFO, AB_EXP and COMBINED_RAM, and negatively correlated with AB_PROD. L_Enforcement is positively correlated with AB_CFO and IM_EQ is negatively correlated with AB_EXP and positively correlated with COMBINED_RAM. These results suggest that these institutional characteristics do affect earnings management, especially earnings manipulation through real activities. These findings are consistent with prior literature (Ball et al., 2000, 2006; Gebhardt & McLeay, 2003; Leuz et al., 2003; Nanda & Wysocki, 2003). Since L_System is a dummy variable, and takes the value of zero if the firm is located in a code-

law country and takes the value of one if the firm is located in a common-law country, firms located in common-law countries are less likely to engage in earnings management through manipulation of production costs and more likely to engage earnings management through manipulation of operating cash flows compared to firms located in code-law countries. Although there exists a negative correlation between L_System and AB_PROD, the results suggest that firms located in common-law countries make significantly more use of real earnings manipulation since the overall measure of real activities manipulation, COMBINED_RAM, is significantly and positively correlated with L_System. In addition, firms with a high level of L_Enforcement, are more likely to engage in real activities manipulation through manipulating operation cash flows. Although there are no significant results for the relation between ABS_DA and L_SYSTEM, there are significant results for L_SYSTEM and real earnings management proxies. Because legal enforcement is stronger in common-law countries, and the equity market is more important, it becomes relatively more costly to manage earnings through discretionary accruals, therefore consistent with prior literature (Ewert & Wagenhofer, 2005; Cohen et al., 2008; and Zang, 2012), managers will increase their use of real activities manipulation.

A strong positive and significant correlation exists between L_Enforcement and IM_EQ. Moreover, there exists a strong positive and significant correlation between

L_System and IM_EQ as well. Since L_System is a dummy variable, this evidence suggests that firms located in commom-law countries have a more important equity market compared with firms located in code-law countries. Also, these countries face a stronger legal enforcement. These findings are consistent with prior literature. Common-law countries have the strongest protection of outside investors (LaPorta et al., 2000) and tend to be more market-oriented relatively to code-law countries (Ball, 2006).

Table 4 compares descriptive statistics of variables across the pre- and post- IFRS adoption periods as well, but makes a distinction between code-law and common-law located firms. For the code-law and common-law sample, the means of ABS_DA, AB_CFO and AB_EXP are not significantly different for both the adopters. However, firms located in code-law countries face a significant increase in AB_PROD, whereas this is not the case for common-law countries. This finding indicates that firms located in code-law countries engage in more aggressive earnings management after the adoption of IFRS compared to firms located in common-law countries. In addition, the mean of L_Enforcement decreased significantly in code-law countries across the pre- and post-IFRS adoption period, whereas this is not the case for common-law countries. The mean of IM_EQ of common-law countries decreased significantly after the adoption of IFRS,

Table 4

Descriptive Statistics by Common-law versus Code-law Code-law Common-law

Mean Difference

p-value Mean Difference p-value

N Pre Post t-test N Pre Post t-test

Dependent Variables ABS_DA 3903 -.0012 -.0005 .7436 1100 -.0068 -.0092 .5550 AB_CFO 3903 -.0011 .0008 .5222 1100 .0160 .0188 .6202 AB_EXP 3903 -.0028 -.0076 .4760 1100 .0186 .0304 .3529 AB_PROD 3806 -.0252 -.0038 .0013*** 1074 -.0762 -.0691 .6323 COMBINED_RAM 3806 .0214 -.0053 .0473** 1074 .1082 .1176 .7446 Variables of Interest L_Enforcement 3604 1.6079 1.5077 .0000*** 1027 1.7152 1.7126 .7216 IM_EQ 3903 .1640 .1722 .1186 1100 .3807 .3681 .0072*** Control Variables B/M 3696 .0173 .0149 .1725 1051 .0107 .0128 .3836 ADJ_ROA 3902 .0305 .0337 .9535 1100 .2015 .2378 .7028 LEV 3604 .1395 .1694 .0066*** 1027 .1725 .1842 .5791 RD_INT 2466 .0806 .0677 .2434 674 .0678 .0490 .1170

The t-test tests the null hypothesis that the mean difference on the variable is zero.

*** Indicates statistical significance at the 1% level, ** indicates statistical significance at the 5% level, and * indicates statistical significance at the 10% level.

whereas the change of the mean of code-law countries is insignificant.

Table 5 presents two Panels, Panel A and B. Panel A presents a comparison before the adoption of IFRS between firms located in code-law and common-law countries, while Panel B presents a comparison after the adoption of IFRS.

As you can see in Panel A, the mean of L_Enforcement is significantly higher for firms located in common-law countries compared to firms located in code-law countries (e.g., mean L_Enforcement = 1.6079 for the code-law located firms and 1.688 for the common-law located firms). The mean of IM_EQ is significantly higher for firms located

in common-law countries compared to firms located in code-law countries (e.g.,

mean IM_EQ = 0.1640 for the code-law located firms and 0.3807 for the

common-law located firms). These results are consistent with the findings of

LaPorta et al. (2000), Ball et al. (2006). Common-law countries have the

strongest protection of outside investors and tend to be more market-oriented

relatively to code-law countries. The mean and the median of all proxies for

earnings management, variables of interest and almost all control variables

(except for RD_INT) are significantly different between the two groups.

Table 5

Panel A: Descriptive Statistics by Common-law versus Code-law Pre-IFRS Adoption Code-law Common-law Difference Tests:

p-value

N Mean Median N Mean Median t-test Wilcoxon Test Dependent Variables ABS_DA 928 -.0012 -.0005 334 -.0068 -.0047 .1168 .0467** AB_CFO 928 -.0011 -.0048 334 .0160 .0120 .0015*** .0001*** AB_EXP 928 -.0028 -.2331 334 .0186 -.0097 .0699** .0480** AB_PROD 881 -.0252 -.0037 317 -.0762 -.0505 .0001*** .0000*** COMBINED_RAM 881 .0214 -.0309 317 .1082 .0514 .0008*** .0000*** Variables of Interest L_Enforcement 866 1.6079 1.6760 307 1.7152 1.688 .0000*** .2481 IM_EQ 928 .1640 .1115 334 .3807 .3880 .0000*** .0000*** Control Variables B/M 882 .0173 .0056 327 .0107 .0047 .0072*** .0002*** ADJ_ROA 927 .0305 .0090 334 .2015 .0606 .0542** .0000*** LEV 866 .1395 .0000 307 .1725 .0000 .0985* .0059*** RD_INT 602 .0806 .0231 212 .0678 .0271 .3806 .1658

The t-test and Wilcoxon test tests the null hypothesis that the means are of the variables are equal.

*** Indicates statistical significance at the 1% level, ** indicates statistical significance at the 5% level, and * indicates statistical significance at the 10% level.

Panel B: Descriptive Statistics by Common-law versus Code-law Post-IFRS Adoption

Code-law Common-law Difference Tests: p-value

N Mean Median N Mean Median t-test Wilcoxon Test Dependent Variables ABS_DA 2975 -.0005 .0001 766 -.0092 -.0037 .0005*** .0009*** AB_CFO 2975 .0008 -.0026 766 .0188 .0116 .0000*** .0000*** AB_EXP 2975 -.0076 -.0287 766 .0304 -.0047 .0000*** .0000*** AB_PROD 2925 -.0038 .0136 757 -.0691 -.0258 .0000*** .0000*** COMBINED_RAM 2925 -.0053 -.0500 757 .1176 .0339 .0000*** .0000*** Variables of Interest L_Enforcement 2738 1.5077 1.6499 720 1.7126 1.6878 .0000*** .0000*** IM_EQ 2975 .1722 .1243 766 .3681 .3525 .0000*** .0000*** Control Variables B/M 2814 .0149 .0056 724 .0128 .0045 .1808 .0000*** ADJ_ROA 2975 .0337 .0127 766 .2378 .0492 .0019*** .0000*** LEV 2738 .1694 .0000 720 .1842 .0000 .2490 .0072 RD_INT 1864 .0677 .0207 462 .0490 .0134 .0093 .0001

The t-test and Wilcoxon test tests the null hypothesis that the means are of the variables are equal.

***Indicates statistical significance at the 1% level, ** indicates statistical significance at the 5% level, and * indicates statistical significance at the 10% level.

This finding is consistent with prior literature and indicates that the legal institutional structure of a country is of importance. The mean of the combined real activitites manipulation (COMBINED_RAM) is significantly bigger for firms located in common-law countries relatively to firms located in code-law countries. This result is consistent with prior literature; common-law countries have stronger investor protection, what makes it relatively more costly to manage earnings through accruals (Ewert & Wagenhofer, 2005; Cohen et al., 2008; and Zang, 2012).

Panel B of Table 5 presents descriptive statistics of variables between code-law located firms and common-law located firms after the adoption of IFRS in 2005. Just as before the adoption of IFRS, the mean and the median of all proxies for earnings management are significantly different between the two groups. This means the classification of legal system remains of importance for explaining earnings management. This finding is consistent with our third hypothesis, in which we state institutional characteristics affect the level of earnings management despite the adoption of IFRS. It is interesting to see that also after the adoption of IFRS, firms located in common-law countries still engage less in accrual-based earnings management relatively to firms located in code-law countries. Moreover, the means of ABS_DA for both legal institutions decreased relatively to the means of pre IFRS adoption ABS_DA. However, when combining this result with the result from Table 4,

the change of the means of ABS_DA is not significant. This finding is inconsistent with the expectation of the regulators. The regulators expected by implementing IFRS a decrease in accrual-based earnings manipulation, which would lead to an increase of the quality of financial reporting (European Commision, 2002). The mean of AB_PROD for firms located in code-law countries increased significantly (Table 4) and remains significant different from firms located in common-law countries. Consistent with LaPorta et al. (2000) L_Enforcement and IM_EQ are significantly different with respect to the legal institution. Across the adoption of IFRS, the legal enforcement, (L_Enforcement) of code-law countries decreased significantly, while this is not the case for common-law countries. Although the mean of IM_EQ for common-law countries decreased significantly across the adoption of IFRS, it remains significantly bigger compared to the mean of code-law countries.

4.2 Multiple Regression Analysis

Table 6 presents the empirical findings of the multiple regression models for accrual and real earnings management. The differences-in-differences design allows assessment of the incremental effects of common-law countries after the IFRS adoption relative to that of code-law countries. The individual real earnings management proxies are included as dependent variables in column (b) to (d). The combined proxy for real earnings management is presented in column (e).

Column (a) presents the results using the absolute value of discretionary accruals (ABS_DA) as a dependent variable. Consistent with prior literature, we find a significant negative relation between L_Enforcement and the magnitude of accrual-based earnings management, ABS_DA. More specifically, the estimated coefficient on L_Enforcement is negative and significant at a 10% significance level, indicating that firms based in countries with a strong legal enforcement manage earnings less through accruals. From this follows the level of legal enforcement is of that importance, also after the adoption of IFRS that it leads to a support of H3. The coefficient of POST2005 is negative, but insignificant, indicating the adoption of IFRS in 2005 did not lead to a significant decrease in accrual-based earnings management. This finding is in contrast with our first hypothesis, in which we posit the mandatory IFRS adoption is associated

with a decrease in accrual-based earnings management practices. In addition, the insignificant positive sum of the coefficients of POST2005 and L_SYSTEM*POST2005 does not suggest a decrease in accrual-based earnings management practices for firms located in common-law countries relatively to firms located in code-law countries. This indicates that it does not matter whether the firm is based in a code-law or common-law country with respect to earnings management through accruals and is inconsistent with prior literature. LaPorta et al. (2000) suggest the outside protection of investors in common-law countries is better relatively to code-law countries, which would lead to less accrual-based earnings management practices relatively to code-law countries. A positive and significant relation is found between IM_EQ and ABS_DA. To be precise, the estimated coefficient on IM_EQ is positive and significant at a 5% significant level. This finding is consistent with our third hypothesis, indicating that the importance of the equity market affects the level of absolute discretionary accruals also after the adoption of IFRS. We find a negative and strongly significant coefficient for the combined proxy of real earnings manipulation COMBINED_RAM. This indicates that firms that are making more use of real earnings manipulation will manage earnings less through accruals. This result is consistent with the findings of Zang (2012), who suggests managers make a

Table 6:

Multiple Regression of Accrual-Based and Real Earnings Management

(a)

ABS_DA AB_CFO (b) AB_PROD (c) AB_EXP (d) Combined_RAM (e)

(1) L_System -.0064 (-1.28) (-1.07) -.0055 (1.20) .0206 -.0546*** (-3.36) -.0832** (-2.55) (2) POST2005 -.0014 (-.58) .0004 (.16) .0218*** (2.63) -.0196*** (-2.49) -.0424*** (-2.70) (3) L_SYSTEM*POST2005 .0027 (.50) (1.26) .0070 -.0642*** (-3.51) .0677*** (3.90) .1434*** (4.13) Test of (2) + (3) = 0 [p-value] .27 (.785) (.135) 1.49 -2.60*** (.009) 3.11*** (.002) 3.27*** (.001) L_Enforcement -.0022* (-1.39) -.0045*** (-2.83) -.0020 (-.39) -.0044 (-.88) -.0065 (-.65) IM_EQ .0226** (2.50) .0572*** (6.02) -.1558*** (-5.03) .1331*** (4.49) .3395*** (5.79) BM .0226 (.89) -.0176 (-.67) -.0614 (-.70) (-1.63) -.1351 -.0560 (-.34) ADJ_ROA .0002 (.21) .0000 (.04) (-1.12) -.0037 -.0027 (-.88) .0022 (.36) LEV .0060* (1.75) .0011 (.30) -.0042 (-.36) .0061 (.55) .0111 (.50) RD_INT -.0104* (-1.80) -.0188*** (-3.17) (1.40) .0278 .0103 (.56) (-1.01) -.0379 Combined_RAM -.0531***

(-19.30)

ABS_DA -.9711***

(-52.50) .9276*** (15.01) -.3678*** (-6.37) -2.258*** (-19.30) Country and industry fixed

effects Included Included Included Included Included

R² adjusted .1218 .5039 .1204 .0285 .1508

N 2745 2815 2745 2815 2745

*** Indicates statistical significance at the 1% level, ** indicates statistical significance at the 5% level, and * indicates statistical significance at the 10% level.

trade-off decision between accrual-based and real earnings manipulation.

Column (e) presents the results using the combined proxy for real earnings management (COMBINED_RAM) as a dependent variable. In this model, the coefficients of L_System and POST2005 are significantly negative. However, the sum of the coefficients of POST2005 and L_SYSTEM*POST2005 is significantly positive (-0.0424 + 0.1434 = 0.101). This finding suggests an increase in real activities manipulation practices for firms located in common-law countries from the pre- to the post IFRS adoption period relatively to firms located in code-law countries. Moreover, this finding supports H2, in which we state the mandatory adoption is associated with an increase in real earnings management practices. However, firms located in code-law countries face a decrease in real activities manipulation across the IFRS adoption, which is inconsistent with H2. From this follows that legal system classification is of that importance that it affects the level of real activities manipulation and thus supports of H3. The coefficient of ABS_DA has a significant negative value. This finding reflects the negative relationship between accrual-based and real earnings management and is consistent with the findings of Zang (2012), who suggests managers make a trade-off decision between accrual-based and real earnings manipulation. Combining the significant positive sum of POST2005 and L_SYSTEM*POST2005 and the significant negative value of the proxy for accrual-based earnings management, the substitution between accrual and real earnings management is related to IFRS adoption and the legal system of the country a firm is based in. Moreover, the coefficient IM_EQ has a positively and significant value, indicating that a firm will make more use of real earnings manipulation if the firm is located in a country with an important equity market. This finding supports our third hypothesis as well. However, regarding the small and insignificant value of L_Enforcement, it does not seem of influence for the combined proxy for real earnings manipulation, which is inconsistent with our third hypothesis.

In case of dependent variable AB_CFO, the value of the coefficient L_Enforcement is significantly negative, while the coefficient of IM_EQ is significantly positive.

Consistent with H3, these findings indicate countries’ characteristics are important for the real earnings management through the level of operating cash flows. However, the coefficients of L_SYSTEM, POST2005, L_SYSTEM*2005 and the sum of coefficients of POST2005 and L_SYSTEM*2005 are small and insignificant. Therefore they do not seem of influence for the proxy real earnings management through the level of operating cash flows. Consistent with prior literature, the coefficient ABS_DA is negative and significant, indicating a negative relationship between earnings management through discretionary accruals and real activities manipulation through the level of operating cash flow.

In case of dependent variable AB_PROD, the significantly negative sum of the coefficients of POST2005 and L_SYSTEM*POST2005 (0.0218 – 0.0642 = -0.0424) suggests a decrease in earnings management practices through manipulation of production costs for firms located in common-law countries after 2005. This finding is inconsistent with H2. However, firms located in code-law countries experience a significant increase in earnings management practices through manipulation of production costs, since the coefficient POST2005 has a significant positive value. Moreover, since the legal institutional structure of a country makes such a significant difference, H3 is supported. Although the coefficient of L_Enforcement is small and insignificant, the coefficient of IM_EQ has a significantly and negative value, which leads to a mixed support of H3. However, in contrast to prior literature (Ewert & Wagenhofer, 2005; Cohen et al., 2008; and Zang, 2012), in case a firm is located in a code-law country, both the proxy for real earning management through abnormal production costs and discretionary accruals (ABS_DA) increases, which indicates there is no such thing as a trade-off.

The results of the regression of the AB_EXP proxy are presented in column (d). The coefficients of the significant variables are quite the same compared to the results of the regression of COMBINED_RAM. Although, the coefficients of L_System and POST2005 are significantly negative, the sum of the coefficients of POST2005 and L_SYSTEM*POST2005 is significantly positive (-0.0196 + 0.0677 = 0.0481). This finding suggests an increase in real activities manipulation practices through the level of abnormal discretionary expenses for firms located in common-law countries from the pre- to the post IFRS adoption period relatively to firms located in code-law countries. These findings support H2 and H3. However, firms located in code-law countries experience a significant decrease in earnings management practices through manipulation of discretionary expenses since the coefficient POST2005 has a significant negative value, which is in contrast to H2. The coefficient of ABS_DA has a significant negative value. This finding reflects the negative relationship between accrual and real

earnings management and is consistent with the findings of Zang (2012). Combining the significant positive sum of POST2005 and L_SYSTEM*POST2005 and the significant negative value of the proxy for accrual-based earnings management, the substitution between accrual and real earnings management is related to IFRS adoption and the legal system of the country a firm is based in. Although the coefficient of L_Enforcement is insignificant and small, the coefficient of IM_EQ has a positively and significant value. Taken these findings together leads to a mixed support of the third hypothesis.

Overall, the results in Table 6 provide no significant evidence of a decrease in accrual-based earnings management across the mandatory IFRS adoption, so H1 cannot be supported. However, more evidence is found for a significant impact on real earnings management between the pre- and post-mandatory IFRS adoption periods, although this evidence gives mixed results. Real activities manipulation practices for firms located in common-law countries increased across the adoption of IFRS, which supports H2. But in case of firms located in code-law countries, the use of real activities manipulation decreased from the pre- to the post-IFRS adoption period, which does not supports H2. This means the classification of the legal system of the country in which the firm is located in is important for the level of real activities manipulation, and thus supports H3. Although legal enforcement is not significant for the combined measure of real activity manipulation, it is significant for accrual-based earnings manipulation and real earnings management through manipulating the level of abnormal cash flow from operations. In addition, the importance of the equity market is significant for all the proxies for earnings management. Taken these findings together it leads to a mixed support of H3.

5. Robustness test

To account for the possibility that the empirical results are driven by the recent financial crisis we conduct a robustness check. Following Doukakis (2013) we excluded all observations after 2008. The empirical results are presented in Table 7.

The inference relating to accrual-based earnings management changed compared to Table 6 column (a). POST2005 becomes significantly negative by excluding the years after 2008. This finding indicates that abnormal discretionary accruals decreased across the adoption of IFRS in 2005, which is consistent with H1. Although the estimated coefficient on L_Enforcement remains significant and becomes even more

negative for the level of abnormal operating cash flows, it is not longer significant for the level of absolute discretionary cash flows. Although the importance of the equity market remains significant for all four real activities management proxies, it is not longer significant for the level of absolute discretionary accruals. The assessment of the incremental effects of common-law countries after the IFRS adoption relative to that of code-law countries is not longer significant for any of the proxies of earnings management. However, the legal system of the country the firm is based in remains of significant influence for the combined real earnings management proxy and the individual proxy for the level of abnormal discretionary expenses. Moreover, L_SYSTEM*POST2005 remains positive and significant for the real activities management proxy as well. This indicates firms located in common-law countries face increase their use of real activities manipulation after the adoption, thus H2 remains supported. However, the significant and negative value of POST2005 indicates, just as in table 5, a decrease in the use of real activities management by firms located in code-law countries, which is inconsistent with H2. H3 remains mixed supported.

Table 7:

Robustness check, excluding 2009 - 2012

(a)

ABS_DA AB_CFO (b) AB_PROD (c) AB_EXP (d) Combined_RAM (e)

(1) L_System -.0070 (-1.22) (-0.87) -.0051 (1.43) .0271 -.0591*** (-3.28) -.0984*** (-2.77) (2) POST2005 -.0056*** (-1.82) -.0010 (-.33) .0277*** (2.74) -.0170* (-1.75) -.0477** (-2.51) (3) L_SYSTEM*POST2005 .0062 (.94) -.0068 (-.99) -.0479** (-2.20) .0438** (2.10) .0931** (2.28) Test of (2) + (3) = 0 [p-value] .11 (.916) (.196) -1.29 (.296) -1.05 (.146) 1.45 (.210) 1.25 L_Enforcement -.0012 (-0.62) -.0060*** (-2.90) -.0046 (-.70) -.0048 (-.77) -.0055 (-.45) IM_EQ .0190 (1.35) .0600*** (4.13) -.1634*** (-3.53) .1592*** (3.59) .3887*** (4.48) BM .0769** (1.97) -.0201 (-.50) -.2817*** (-2.19) -.0098 (-.08) (1.12) .2719 ADJ_ROA .0002 (.16) -.0010 (-.75) (-1.43) -.0067 -.0033 (-.78) .0044 (.50) LEV .0091* (1.90) .0045 (.90) .0012 (.08) .0040 (.26) .0036 (.12) RD_INT -.0181** (-2.24) (-1.32) -.0107 .0645** (2.42) -.0750*** (-3.03) -.1532*** (-3.06) Combined_RAM -.0581*** (-14.91)

ABS_DA -.9415***

(-38.49) .9201*** (11.52) -.3492*** (-4.67) -2.234*** (-14.91) Country and industry fixed

effects Included Included Included Included Included

R² adjusted .1313 .4915 .1199 .0233 .1519

N 1500 1553 1500 1553 1500

*** Indicates statistical significance at the 1% level, ** indicates statistical significance at the 5% level, and * indicates statistical significance at the 10% level.

6. Conclusion and discussion

This study examines the effect of the adoption of IFRS in 2005 on earnings management across 22 members of the European Union. We hypothesize that due to the adoption of IFRS accrual-based earnings management will decrease and real activities manipulation will increase, and that institutional characteristics affect the level of earnings management despite the adoption of IFRS. With institutional characteristics are meant: the classification of the legal system, investor protection by the level of legal enforcement and the importance of the equity market.

Prior literature investigated the impact of mandatory IFRS adoption and institutional characteristics on accrual-based earnings management, but not so much on real activities manipulation. The present study contributes to the literature by investigating the impact of the adoption of IFRS on both accrual-based earnings manipulation and real activities manipulation and the effect of institutional characteristics on the magnitude of these earnings manipulation techniques. To examine this we undertake a multiple regression analysis.

Our findings give mixed support for our three hypotheses. Our results provide no significant evidence of a decrease of the use of accrual-based earnings management due to the adoption of IFRS, thus IFRS adoption is not associated with a decrease in accrual-based earnings management practices. Although the use of real activities manipulation decreased in case of firms located in code-law countries, the use increased for firms located in common-law countries. This finding indicates the legal system of a country is of importance for the level of real activities manipulation. Our findings indicate that institutional characteristics do affect the level of earnings management, also after the adoption of IFRS. Especially the legal institutional structure and equity market are of importance, while legal enforcement gives mixed results. Not all findings are robust when we control for the financial crisis by excluding the years after 2008.

Consistent with our first hypothesis, the findings now indicate that abnormal discretionary accruals decreased across the adoption of IFRS.

Our empirical findings are subject to several caveats. First, besides accrual-based and real earnings manipulation, managers might apply other earnings management techniques. According to Athanasakou, Strong and Walker (2009) another way to manipulate earnings is for instance engaging in classification shifting or expectations management in attempt to meet analysts’ forecasts. Second, because other institutional factors correlate with investor protection they may also affect insiders’ earnings management incentives. Often, institutional factors are complementary and this makes it difficult to fully control for the potential impact of other factors and to disentangle them from the direct effect of investor protection. Finally, earnings management is difficult to measure and may be subject to criticism as lack of power, correlated omitted variables, etc. However, for examining this type of research question the proxies for earnings management seem to be a suitable methodology.

7. References

Athanasakou, V., Strong, N.C., Walker, M., (2009). Earnings management or forecast guidance to meet analyst expectations? Accounting and Business Research 39, 3- 35.

Badertscher, B.A., (2011). Overvaluation and the choice of alternative earnings management mechanisms. The Accounting Review 86(5), 1491 – 1518.

Ball, R., (2006). International Financing Reporting Standards (IFRS): pros and cons for investors.

Accounting and Business Research 36, 5 – 27.

Ball, R., Kothari, S.P. and Robin, A. (2000). The effect of international institutional factors on properties of accounting earnings. Journal of Accounting and Economics, 29(February), 1 – 51.

Ball, R,. Robin, A., Wu, J.S., (2003). Incentives versus standards: properties of accounting income in four East Asian countries. Journal of Accounting of Economics 36, 235 – 270.

Burgstahler, D.C., Hail, L. and Leuz, C., (2006). The Importance of Reporting Incentives: Earnings Management in European Private and Public Firms. The Accounting

Review, 81(5), 983 – 1016.

Cohen, D.A., Dey, A. and Lys, T.Z., (2008). Real and accrual-based earnings management in the pre- and post- Sarbanes-Oxley periods. The Accounting Review, 83(3), 757 – 787.

Cohen, D.A., Dey, A., Lys, T.Z. and Zarowin, P., (2010). Accrual-based and real earnings management activities around Seasoned Equity Offerings. Journal of Accounting and

Economics, 50(1), 2 – 19.

Daske, H., Gebhardt, G., and McLeay, S., (2006). The distribution of earnings relative to targets in the European Union. Accounting and Business Research 36(3), 137 – 167. Dechow, P.M., and Dichev, I.D., (2002). The Quality of Accruals and Earnings: The role of

Accrual Estimation Errors. The Accounting Review Vol. 77, 35 – 59.

Dechow, P.M., Kothari, S.P., and Watts, R., (1998) The relation between earnings and cash flows.