The relationship between integrated reporting and real earnings management in South Africa

Amsterdam Business School

The relationship between integrated reporting and real

earnings management in South Africa

Master Thesis Name: Shain Hamid Student number: 11020237

Thesis supervisor: Dr. G. Georgakopoulos Date: 26th _{June 2017}

Word count: 15,339

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

2 Statement of Originality

This document is written by student Shain Hamid who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

3

Abstract

This study examines the possible influence of integrated reporting on real earnings management in a pre-integrated reporting setting and in a post integrated reporting setting in South Africa. There is no research that examines the effect of integrated reporting on real earnings

management. Therefore, this study aims to fill the gap into existing literature. The research is conducted on South African firms that are listed on the Johannesburg Stock Exchange (JSE), since integrated reporting has been mandatory for firms listed on the JSE. Real earnings management is measured as the sum of abnormal production costs and abnormal cash flows (Roychowdhury, 2005).

Contrary to the expectations, which are based on the studies of Zhou et al. (2016), Barth et al. (2015), and Reimsbach et al. (2017), this study found no significant relation between integrated reporting and real earnings management and there is no significant evidence that before 2010 there was more real earnings management than after 2010. This contradicts the claim that integrated reporting improves the quality of financial reporting and reduces real earnings management.

4

Content

1 Introduction ... 5

2 Literature review and hypotheses ... 8

2.1 Integrated Reporting ... 8

2.1.1 Integrated reporting motives ... 10

2.2 Earnings management ... 11

2.2.1 Earnings management motives ... 13

2.3 Relationship between integrated reporting and earnings management ... 14

2.4 Agency Theory ... 16

2.5 Hypotheses development ... 16

2.6 Oversight ... 18

3 Methodology ... 21

3.1 Data collection and sample selection ... 21

3.2 Measurement of earnings management ... 23

3.2.1 Robustness test ... 24

3.2.2 Real earnings management model ... 24

3.3 Control variables ... 26 3.4 Regression model... 28 4 Results ... 30 4.1 Descriptive Statistics ... 30 4.2 Correlation Matrix ... 31 4.3 Regression Analysis ... 32 4.4 Additional tests ... 33 5 Discussion ... 35 6 Conclusion... 38 References ... 40 Appendices ... 43

5

1

Introduction

The recent wave of corporate governance failures and accounting scandals, such as Worldcom and Enron, has raised concerns about the integrity of the accounting information provided to investors and resulted in a drop in investor confidence (Agrawal & Chadha, 2005). These failures and scandals increased regulators’ scrutiny considerable by, for example, implementing Sarbanes-Oxley Act (Sox) in 2002. SOX was implemented to increase internal control and to increase the transparency of the financial statements (Cohen et al., 2008). After the passage of Sox accrual-based earnings management declined substantially (Cohen et al., 2008). However, the increased regulators’ scrutiny had negative consequences. Firms continued to manage earnings via real manipulation activities rather than accrual-based earnings management, such as reductions and postponements of investments (Roychowdhury, 2006). According to Cohen and Zarowin (2010) and Roychowdhury (2005), the most important reason for executives and management to manage earnings through real activities than through accruals is that accrual-based earnings management is more likely to draw auditor or regulatory scrutinity than real decisions. Cohen et al. (2008) state that investors should be aware that firms which face financial distress and industry competition substitute accrual-based earnings management for real earnings management to meet short-term targets. Therefore, investors are recommended to be more skeptical and are advised to analyze the financial statements thoroughly before investing in a firm that faces high industry competition or financial distress. Recent researches (Gunny, 2005; Roychowdhury, 2006; Zang, 2012) show increased appreciation for the importance of understanding how firms manage earnings through real activities manipulation in addition to accrual-based activities. To obtain the desired earnings level, firms could choose to manage earnings through deviating from the normal business activities although this may affect the future economic performance of the firm

negatively (Roychowdhury 2006). High quality, value-relevant information is very important for firms, investors and capital markets since they depend upon this information for the efficient and effective allocation of resources, to encourage a climate for investment, and ensure transparent, ethical and sustainable business practices (Zhou et al., 2016). The information produced by firms must be reliable and show the economic reality of the firm because providers of financial capital and other users of corporate information use the information for their decision making (Zhou et al., 2016).

According to the Villiers et al. (2016) and the IIRC (International Integrated Reporting Council), integrated reporting has the potential to increase the quality of financial information. Villiers et al. (2016), Humphrey et al. (2016) and Sierra-Garcia et al. (2015) conclude in their latest study, that integrated reporting is a new concept and is not much investigated and researched. In

6 their study, Villiers et al. (2016) state that integrated reporting represents a new philosophy that focusses on providing future value creation information, related to the firm’s strategy and business model, to financial stakeholders. In another study of Villiers et al. (Villiers et al., 2014) they state that there is limited amount of research on integrated reporting. According to Villiers et al. (2014), the first practices of integrated reporting started in 2002 and the first framework was published by IIRC in 2013 to provide guidance for companies on what forms an integrated report. Many of the initial studies (Brown & Dillard, 2014; Higgins, Stubbs, & Love, 2014) advocated or criticised integrated reporting or provided advice on how to implement integrated reporting. These initial integrated reporting studies were followed by empirical studies using case studies, interviews or content analyses (Humphrey et al., 2016; Bommel, 2014; Stubbs & Higgins, 2014). Villiers et al (2016) state that integrated reporting is a recent development and therefore limited data and research is available. Given the preliminary nature of the studies and their findings, the integrated reporting measures used and the research design choices of these studies are open to critique, and there are several opportunities for further research (Villiers et al., 2014; Villiers et al., 2016). Villiers et al. (2016) conclude in their research that future research should focus on whether the adoption of integrated reporting is associated with real earnings

management. Real earnings management can affect investors’ investments and integrated

reporting provides investors with the information they need to make more effective decisions to facilitate better long-term investment returns. So it is very interesting to research if there is a connection and if integrated reporting has influence on real earnings management.

In this paper I want to examine whether firms that adopt integrated reporting, behave appropriately to constrain earnings management, thereby delivering more transparent and reliable financial information to investors as compared to other firms that do not meet the same criteria. Despite the increasing interest in- and importance of real earnings management activities, no study to date has examined the effect of integrated reporting on real earnings management and whether this has any influence on the levels of earnings management in a mandatory setting. I fill in this gap in the literature. To obtain a strong motivation for my study, I look into prior research regarding integrated reporting and economic consequences. There are some studies on the relationship between integrated reporting and economic consequences (Barth et al., 2015; Zhou et al., 2016; Bernardi & Stark, 2015), for example the impact of integrated reporting on analysts’s forecast error and firm value, but with mixed results.

Therefore the research question is:

7 This study examines the possible influence of integrated reporting on real earnings management in a pre-integrated reporting setting and in a post integrated setting in South Africa.

Three measures are used to examine the influence of integrated reporting on real earnings management: (1) the abnormal production costs, (2) the abnormal cash flows and (3) the

combined model of (1) and (2). These measures are chosen, because these measures are described as the most preferred in prior literature (Dechow et al., 1995; Roychowdhury, 2006; Cohen et al., 2008; Kim et al., 2012; Zang, 2012; Zhou et al., 2016). This research is conducted on South African firms that are listed on the Johannesburg Stock Exchange (JSE). Since March 2010, integrated reporting has been mandatory for firms listed on the JSE. The characteristics of the JSE as a market, for example in terms of liquidity and market reaction, are similar to those of developed country bourses (Villiers et al., 2014; Zhou et al., 2016). In a study setting such as the African JSE, the ‘before’ (2008-2009) and ‘after’ (2010-2016) the implementation of integrated reporting could be analysed. By using data from all the listed companies in South Africa, the sample is representative.

Contrary to the expectations, which were based on the findings of Barth et al. (2015), Zhou et al. (2016) and Reimsbach et al. (2017), this research cannot find any significant

association between integrated reporting and real earnings management. There is no significant evidence that before 2010 there was more real earnings management than after 2010. The

conclusion of this paper is that integrated reporting does not influence real earnings management in South Africa.

The study is organized as follows. Section 2 contains the literature review and hypotheses, in which the background of this study is discussed, divided into paragraphs about integrated reporting (motives), real earnings management (motives), the agency theory and an oversight of the important papers used in the literature review. Next, section 3 provides the methodology where the used measures and the sample selection are explained. In section 4 the results of the regression and additional tests are presented. In section 5 the discussion is presented. Last, section 6 concludes this study.

8

2

Literature review and hypotheses

2.1 Integrated Reporting

Social and environmental reporting has a long history, which initially took place through disclosures within corporate annual financial reports (Villiers et al., 2014). In the past two decades, however, social and environmental disclosures have been made in separate reports. These reports have become more complex and long since a greater range of issues has been disclosed to meet the expectations of stakeholders (Cho et al., 2009). Because of the length and the complexity of the stand-alone reports, there has been a move to combine social and

environmental reports with financial disclosures in single reports (Villiers et al., 2014). These recent changes and moves have sought to integrate social, environmental, financial and governance information (Hopwood et al., 2010). Villiers et al. (2014) state that these resulting practices have become to be known as integrated reporting. In its extreme form integrated reporting can be defined as the integration of social, environmental, financial and governance information. However, integrated reporting is much more than just a report that integrates social, environmental, and financial and governance information (PWC, 2013; IIRC, 2013).

PricewaterhouseCoopers (PWC) states that integrated reporting is the means by which companies communicate how value is created and will be preserved over the short, medium and long term. This information is used principally by investors to support their capital allocation decisions (PWC, 2013). According to the IIRC (2013), an integrated report is a “concise communication about how an organization’s strategy, governance, performance and prospects, in the context of its external environment, lead to the creation of value over the short, medium and long term”. The difference between the IIRC and other interpretations of integrated reporting is that, according to IIRC integrated reporting focusses primarily on providers of financial capital, particularly those with a long term view.

According to Villiers et al. (2014) and Humphrey et al. (2016) integrated reporting has rapidly gained considerable prominence and attention since the formation of the IIRC in 2010. The IIRC (2013) claims that integrated reporting aims to ‘improve the quality of information available to providers of financial capital to enable a more efficient and productive allocation of capital’ and ‘support integrated thinking, decision-making and actions that focus on the creation of value over the short, medium and long term. The IIRC was formed in 2010 and is a global coalition of regulators, investors, companies, standard setters, the accounting profession and NGOs, who all share the view that communication of value should be the next step in the evolution of corporate reporting. In 2011, the IIRC launched a global IIRC pilot do develop an

9 integrated reporting framework. In April 2013 the IIRC launched an integrated reporting

framework, the Consultation Draft 1 _{(IIRC, 2013). The mission of IIRC is to create a globally} accepted integrated reporting framework that assists organizations in presenting material information about their strategy, governance, performance in a clear, concise and comparable format (IIRC, 2013). Although the IIRC has become the dominant body globally in developing policy and practice around integrated reporting they were not the first body in this area (Cheng et al., 2014). South Africa was and is still the first country to require listed companies to produce an integrated report (Zhou et al., 2016; Barth et al., 2015; Villiers et al., 2014). Specifically, following on from the King III2 _{initiatives in March 2010, listed companies on the JSE were mandated to} provide an integrated report (Zhou et al., 2016; Barth et al., 2015; Villiers et al., 2014). At present, integrated reports are common practice among South Africa’s listed companies and larger state-owned organizations. Smaller state-state-owned organizations and non-profit organizations also

prepare integrated reports (Villiers et al., 2014). To improve the integration of current disclosures, the Integrated Reporting Committee of South Africa (IRCSA) was established in 2011 (Cheng et al., 2014). In South Africa business organizations were called to account for their non-financial performance. Zhou et al. (2016) state that the motive behind the promotion of integrated reporting in South Africa was a belief that the existing requirements for disclosures were not sufficient and organizations and directors should change the way they acted. The IRCSA initiated a discussion on the accountability of the impacts of business activities (Villiers et al., 2014). Organizations in South Africa were urged to commit to the principles of integrated thinking and organization should integrated their reporting approaches on risk and opportunities through financial and sustainability considerations. Listed companies in South Africa are required to combine financial performance information with sustainability performance information within their annual reports.

Villiers and Van Staden (2006) conclude in their report that the IIRC’s vision of

integrated reporting shares characteristics with the developments of integrated reporting in South Africa, but there are some differences. One of the main differences is that the IIRC in mainly focusses on the interests and information needs of providers of financial capital and the IRCSA was initially focused on a broader perspective that encompasses the interest of a wider range of stakeholders (Villiers & van Staden, 2006). Although Villiers and Van Staden (2006) state that there is a difference, Villiers et al. (2014) state in their latest research that the IRCSA is endorsing the IIRC’s investor value creation focused integrated reporting framework. They conclude that

1 http://integratedreporting.org/wp-content/uploads/2013/03/Consultation-Draft-of-the-InternationalIRFramework.pdf

10 the integrated reporting of South Africa and the framework of the IIRC is becoming aligned. South African companies are using much more of the integrated reporting framework of the IIRC than before (Villiers et al., 2014). For example, Cheng et al. (2014) mention that one of the key concerns noted was that the integrated reporting framework of the IIRC identifies providers of financial capital as the primary users of an integrated report, but the IIRC is also going to a wider range of stakeholders. While the IIRC states that the primary intended audience for an integrated report are the providers of financial capital, it also notes that integrated reporting will be of benefit to all stakeholders interested in an organization’s ability to create value over time, including employees, customers, suppliers, legislators, local communities and more (Cheng et al., 2014). Even though integrated reporting, as stated by the IIRC, is aimed at “providers of financial capital” as its primary audience. The South African listed companies use the same principle.

2.1.1 Integrated reporting motives

Integrated reporting is used principally by investors to support their capital allocation decisions (IIRC, 2013). In November 2013, PWC launched the ‘Annual Integrated Reporting Benchmark among Dutch top 50 AEX companies’. In this report3 _{PWC conclude that the ongoing financial} crisis has highlighted the need for more comprehensive information in corporate reporting, so that investors and other stakeholders can have a clear sense of the risks and opportunities faced by companies (PWC, 2013). PWC (2013) state in their report that investors and other

stakeholders are now demanding that management teams provide clear, unambiguous information about issues such as external drivers affecting their business, their approach to governance and managing risk, and how their business model really works. Integrated reporting has been created to enhance accountability, stewardship and trust as well as transparency of business. Integrated reporting provides investors with the information they need to make more effective decisions to facilitate better long-term investment returns (IIRC, 2013). On the contrary, Rensburg and Botha (2013) argue that, although it is assumed that stakeholders and investors benefit from the integrated reporting of companies, it is not always the case. They state that very few stakeholders use integrated reports as their main source of financial and investment

information, and that these reports are seen as additional information. They conclude that investors and other stakeholders still use the annual and interim financial reports by companies for corporate financial information. Adams and Simnett (2011) conclude in their report that the advantages for external and internal stakeholders of integrated reporting include transparency and

11 insight into the future and strategic direction of an organization. According to Adams and Simnet (2011) investors and other stakeholders will have a clear understanding of the business and financial performance of a business. The risk of this transparency is that organizations will be hesitant to provide too much disclosure for fear of risking commercially sensitive information, including details on strategy, and value drivers.

2.2 Earnings management

Earnings management occurs when managers use judgment 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). Investors, stakeholders and analysts look at earnings to determine the attractiveness of a particular stock. According to Prior et al. (2008) earnings management not only affects a firm’s owners but, it also has an impact on other stakeholders, too. Stakeholders bear risks as a result of having invested in some form of capital, human or financial, in a firm (Prior et al., 2008). Managerial actions like earnings management mislead stakeholders and have serious consequences for shareholders, creditors, employees, and society as a whole (Zahra et al., 2005). According to Davidson et al. (2004) earnings management is used to misrepresent the company’s earnings, in order to get a pre-determined target that is equivalent with the figures of the financial statements. This is the reason why managers mislead their stakeholders, in order to satisfy their own self-interests (Davidson et al., 2004). According to Davidson et al. (2004) managers and stakeholders have their own

interests, and when earnings management is applied, managers will act to their best interest and not that of stakeholders. For example, shareholders want to get a return on their investment in the company, while managers are interested in their compensation, promotion, and reputation. One means of managing earnings is by manipulation of accruals with no direct cash flow consequences, this is called accrual manipulation (Roychowdhury, 2006; Cohen et al., 2008; Dechow et al., 1995). Accruals are the difference between net income and cash flows and demonstrate the true performance of the firm by recording revenues and costs in the relevant period, rather than presenting the cash flows (Cohen et al., 2008). An example of an accrual is deferred revenue which is reported when the cash flow from the sales is received before the recording of the sales (Roychowdhury, 2005). When companies engage in earnings management, they can increase or decrease income by creating accruals; these are often referred to as non-discretionary accruals (Matsumoto, 2002). However, it is the non-discretionary accruals, accruals created to manipulate changes in reported earnings that are of concern (Matsumoto, 2002). These

12 types of accruals include using increasing or decreasing estimates of bad debt reserves, warranty costs, and inventory write-downs (Roychowdhury, 2006). Phillips et al. (2003) state that accrual-based earnings management generally is less observable than management’s choices of

accounting methods, harder to detect, and less costly to implement than altering operating cash flows. Prior to the passage of the Sarbanes Oxley Act, research found extensive evidence of accrual-based earnings management (Cohen et al., 2008; Cohen & Zarowin, 2010).

In addition to accrual manipulation, firms can manage earnings by altering real earnings activities (Cohen & Zarowin, 2010). According to Cohen and Zarowin (2010) the distinction is important, because while accrual-based earnings management activities have no direct cash flow

consequences, real earnings management does. Cohen and Zarowin (2010) refer to real earnings manipulation as actions managers take that deviate from normal business practices. Firms could deviate from operating and investing activities by, for example, altering the level of discretionary expenditures, such as research and development expenditures (R&D) and selling, general and administrative expenditures (Cohen & Zarowin, 2010). Real earnings management are

management actions that mislead stakeholders into believing that earnings benchmarks have been met in the normal course of operations (Cohen et al, 2008). An example is to postpone a certain expense regarding for example R&D. Managers manipulate real activities by for example reducing discretionary expenditures or capital investments (Roychowdhury, 2005). Real management activities can be indistinguishable from optimal business decisions, and thus more difficult to detect, although the costs involved in such activities can be economically significant to the firm (Cohen and Zarowin, 2010). Roychowdhury (2006) finds evidence that firms use multiple real earnings management methods in order to meet certain financial reporting benchmarks to avoid reporting annual losses. Managers provide price discounts to temporarily boost sales, reduce discretionary expenditures in order to improve reported margins, and overproduce to lower the cost of goods sold (Roychowdhury, 2005). In contrast to Phillips et al. (2003), Graham et al. (2005) provide evidence suggesting that managers prefer real earnings management activities compared to accrual-based earnings management because real earnings management, while more costly, have a greater probability of not being detected.

After the passage of Sox accrual-based earnings management declined substantially (Cohen et al., 2008). However, the increased regulators’ scrutiny had negative consequences. Firms continued to manage earnings via real manipulation activities rather than accrual-based earnings management, such as reductions and postponements of investments (Roychowdhury 2006). Investors should be aware that firms which face financial distress and industry competition

13 substitute accrual-bases earnings management for real earnings management to meet short-term targets.

2.2.1 Earnings management motives

Boschen et al. (2003) state in their research that unexpectedly good accounting performance is initially associated with higher CEO compensation. They offer evidence that CEO’s have financial incentives to meet earnings expectations and that this is mostly the reason behind earnings management. In their research Boschen et al. (2003) show that unexpectedly good stock performance increases the long-run cumulative financial gains of CEO’s. Matsunaga and Parker (2001) conducted a research on CEO annual bonuses and their evidence also indicates that CEO bonuses provide managers with an incentive to meet earnings forecasts. Their research also contributes to the earnings management literature by providing an incentive for CEO’s to manage earnings in order to meet benchmarks. Payne and Robb (2000) provide evidence that there are strong incentives to achieve analysts’ earnings forecast to protect a company’s stock price. Certainty is an important commodity in the stock market (Payne and Robb, 2000). When the earnings of a company don’t meet analyst forecast expectations investors will react and stock prices will fall. This is a motive for management to manager earnings. According to Payne and Robb (2000) senior executives make efforts to avoid negative earnings surprises since they have much to lose if their firms fail to meet expectations. According to Healy and Wahlen (1999), investors and financial analysts value stocks on the basis of financial accounting information. When the earnings are higher than expected than, it is presumed that the stock price of the firm will increase. Users of the financial statements become optimistic about future firm performance. So CEO’s and high management will try to keep the earnings high in order to keep the stock prices high (Healy & Wahlen, 1999). Othman and Zeghal (2006) researched the motives behind earnings management in Canadian firms and their findings provide evidence that initial public equity offerings are strong motives for earnings management. Hence Canadian firms show specific incentives matched with a dynamic capital market. Issuing equity is a strong motive for earnings management.

Cohen et al. (2008) conclude in their research that before the passage of the SOX, the increase in accrual-based earnings management was associated with an increase in option-based compensation. However, according to Cohen et al. (2008), option-based compensation has two opposing incentives with respect to current stock prices and thus earnings management. On the one hand, manager have an incentive to lower the current stock price around stock option award. On the other hand, for unexercised options, managers have an incentive to increase current stock

14 prices and, hence, to manage earnings upward. After the passage of the SOX, accrual-based earnings management declined significantly, while real earnings management increased (Cohen et al, 2008) and at the same time, option-based compensation decreased. So according to Cohen et al. (2008), option-based compensation is an incentive for accrual-based earnings management and much less an incentive for real-earnings management.

Healy and Wahlen (1999) researched the incentives for earnings management of publicly held firms. The main incentives come from capital market expectation, contracts based on terms of reported accounting numbers and governmental regulation. Public firms engage in earnings management much more than private firms (Healy and Wahlen, 1999). Givoly et al. (2010) also argue that private firms engage less in earnings management compared with public firms. Private firms have less problems due to agency problems, because private firms’ ownership is highly concentrated. Therefore, private firms have fewer incentives to cover the unsatisfied economic firm performance (Givoly et al., 2010). Also, public firms need to meet certain performance benchmark to attract investors in the market so they have more incentives to manipulate numbers on financial reports (Givoly et al., 2010).

2.3 Relationship between integrated reporting and earnings management

Reimsbach et al. (2017) conclude in their research that integrated reporting increases investors’ potential access to sustainability information, because as the researchers state: “readers could not entirely opt out of encountering sustainability information during their information processing” (Reimsbach et al., 2017, p. 17). So investors don’t want a lot of sustainability information during their information processing and want to filter only what they think is useful (Reimsbach et al., 2017). According to Reimsbach et al. (2017) a substantial number of investors showed no interest in the separate sustainability report and those participants who accessed the integrated

sustainability information showed good ability to store and recall sustainability information. Reimsbach et al. (2017) conclude that the results indicate that the assurance of sustainability information was associated with an increase in the perceived sustainability performance and led to higher investment-related judgments. These results are in line with the statement made by the IIRC (2013), that integrated reporting has been created to enhance accountability, stewardship and trust as well as transparency of business. Integrated reporting provides investors with the information they need to make more effective decisions to facilitate better long-term investment returns (IIRC, 2013).

Barth et al. (2015) investigate the economic consequences (firm value) associated with integrated report in South Africa. Barth et al. (2015) used three components for firm value: stock

15 liquidity, cost of capital, and expected future cash flows. They found a positive association between integrated reporting and stock liquidity and integrated reporting and expected future cash flows, but they didn’t find evidence of a relation between integrated reporting and cost of capital. Integrated reporting is associated with higher realized future operating cash flows and is associated with less over- and under-investment (Barth et al., 2015). Bart et al. (2015) conclude that integrated reporting improves external information and better internal decisions. Zhou et al. (2016) also investigated the economic consequences associated with integrated reporting in South Africa, but contrary to Barth et al. (2015), they used also data from before the implementation of the regulation in South Africa. Their findings show that analysts’ forecast error is lower when the quality of integrated reporting is higher, but they report no clear association between integrated reporting and forecast dispersion. Bernardi and Stark (2015) report no association between integrated reporting and analyst forecast errors prior to the integrated reporting regime in South Africa. And in contrast to Zhou et al. (2016), Bernardi and Stark (2015) show a negative and statistically significant association between integrated reporting disclosures and forecast accuracy after the adoption of integrated reporting in South Africa. Bernardi and Stark (2015) studied the impact of the reporting regime change in South Africa in 2010, on analyst forecast accuracy as a way of evaluating users’ perceptions of the value of integrated reporting and they conclude that integrated reporting does have effect on the level of transparency of environmental, social and governance disclosures. In order for integrated reporting to be useful to investors, an explicit investor focus is not required (Bernardi & Stark, 2015).

There are also prior studies that have looked into the relation between CSR (Corporate Social Responsibility) and earnings management (Kim et al., 2012; Chih et al., 2007). Kim et al. (2012) examined whether socially responsible firms behave differently from other firms in their financial reporting. They researched whether firms that exhibit CSR also behave in a responsible manner to constrain earnings management, and delivering more transparent and reliable financial information to investors. According to Kim et al. (2012) socially responsible firms are less likely to manage earnings through discretionary accruals, are less likely to manipulate real operating activities and are less likely to be subject of SEC investigations against top executives. Chih et al. (2007) investigate the effect of CSR on the quality of publicly released financial information during the 1993-2002 period in 46 countries. Chih et al. (2007) researched if there is an

association between accrual earnings management and CSR and they conclude that an increase in CSR score mitigates earnings smoothing, increases earnings aggressiveness, and mitigates

16 interested in avoiding earnings losses and decreases (Chih et al., 2007). It is however interesting that firms that engage in CSR, engage in more earnings aggressiveness (Chih et al., 2007).

2.4 Agency Theory

The agency theory is an agency relationship, which is based on a contract, in which one party (the principal) delegates work to another (the agent) who performs that work (Jensen & Meckling, 1976). Jensen & Meckling (1976) state that the agency theory is concerned with resolving problems that can occur in agency relationships. One problem that can occur, is when the goals of the principal and agent conflicts. It is often difficult and expensive for the principal to verify what the agent is doing and the principal cannot verify that the agent has behaved appropriately (Jensen & Meckling, 1976). For example management has different interests than stakeholders because stakeholders have to make decisions based on the annual report, so this must be reliable. The stakeholders could be protected, when the financial numbers provide a true view of the real economic situation of the company (Jensen & Meckling, 1976).

Davidson et al. (2004) state that there is a relationship between the agency theory and earnings management. Managers may have personal goals that compete with the goals of shareholders. Managers are given power by shareholders to make decisions, and if there is a conflict of interest this will have potential agency costs (Davidson et al., 2004). Davidson et al. (2004) conclude that earnings management is a type of agency costs if managers present financial reports that do not present an accurate economic picture of a firm. As a result investors and shareholders make bad investment decisions based on this inaccurate information (Davidson et al., 2004). For example, a new CEO may feel pressure to show improved performance and to convince the board and shareholders that their hiring decision is moving the company in a good direction.

2.5 Hypotheses development

Villiers et al. (2016) conclude in their research that future research should focus on whether the adoption of integrated reporting is associated with real earnings management because real earnings management can affect investors’ investments and integrated reporting provides investors with the information they need to make more effective decisions to facilitate better long-term investment returns. So it is very interesting to research if there is a connection and if integrated reporting has influence on real earnings management. Based on the literature (Villiers et al., 2016; Villiers et al., 2014; Zhou et al., 2016; Barth et al., 2015) and theories I want to

17 introduce hypotheses to test the relationship between integrated reporting and real earnings management. Even though there is no literature regarding integrated reporting and real earnings management, there are some interesting and important conclusions of previous studies that mostly looked into the relation between integrated reporting and the economic consequences of integrated reporting. Barth et al. (2015) conducted a research to investigate if there is a

relationship between integrated reporting and firm value and they found a positive association between integrated reporting and stock liquidity, positive association with expected future cash flow, but no evidence of a relation between integrated reporting and cost of capital. They conclude in their research that integrated reporting is associated with higher realized future operating cash flows and is associated with less over- and under-investment. Zhou et al. (2016) show that analysts’ forecast error is lower when the quality of integrated reporting is higher. So their finding is in line with the research of Reimsbach et al. (2017) and the statement made by the IIRC (2013) that integrated reporting has been created to enhance accountability, stewardship and trust as well as transparency of business. Integrated reporting provides investors with the information they need to make more effective decisions to facilitate better long-term investment returns (IIRC, 2013). On the contrary, Bernardi and Stark (2015) report no association between integrated reporting and analyst forecast errors. And in contrast to Zhou et al. (2016), Bernardi and Stark (2015) show a negative and statistically significant association between integrated reporting disclosures and forecast accuracy after the adoption of integrated reporting in South Africa. There are also studies (Kim et al., 2012; Chih et al., 2007) examining the relationship of CSR with integrated reporting. According to Kim et al. (2012) socially responsible firms are less likely to manage earnings through discretionary accruals, are less likely to manipulate real

operating activities, and are less likely to be subject of SEC investigations against top executives. Chih et al. (2007) researched if there is an association between accrual-based earnings

management and CSR and they state that CSR mitigates earnings smoothing, increases earnings aggressiveness and mitigates earnings losses avoidance.

Based on the existing literature and prior studies on integrated reporting and the

economic consequences and the claims that integrated reporting provides transparency and value for investors (Villiers et al., 2014; Villiers et al., 2016; IIRC, 2013; PWC, 2013) I expect that there is a relationship between integrated reporting and real earnings management. Before 2010, when integrated reporting was not mandatory in South Africa, I would expect that there is more real earnings management than after 2010 when integrated reporting became mandatory in South Africa. In companies that are implementing integrated reporting I expect that real earnings management declines, so integrated reporting has a negative effect on real earnings management.

18 Therefore my main hypothesis is:

Hypothesis 1

Before 2010, when integrated reporting was not mandatory in South Africa, there was more real earnings management than after 2010 when integrated reporting became mandatory for listed companies in South Africa.

Based on literature (Roychowdhury, 2006; Cohen et al., 2008; Kim et al., 2012), I use 3 proxies to measure real earnings management which are sales manipulation/abnormal cash flows,

overproduction and a combined measure of both. Therefore I have 2 additional hypotheses that answer the main hypothesis.

Hypothesis 1a

Integrated reporting has a negative effect on abnormal cash flows. When firms use integrated reporting, the abnormal cash flows reduce.

Hypothesis 1b

Integrated reporting has a negative effect on overproduction. When firms use integrated reporting, the levels of production decreases.

2.6 Oversight

In this section a depiction of the main articles reviewed in the literature review is provided in table 1.

Table 1. Literature Overview

Author(s) Year Research matter Important for this study

Villiers, Rinaldi

& Unerman 2014 Insights from accounting and accountability research into the rapidly emerging field of integrated reporting

To meet expectations of stakeholders, there has been a move to integrated reporting

Cheng, Green, Conradie, Konishi & Romi

2014 The development of integrated reporting, which is based on the International Integrated Reporting Framework

Integrated reporting differences between the IIRC and South Africa. The uses of the reports are the main differences

Zhou, Simnet, Green

2016 Effect of integrated reporting on the capital market, study of South African listed companies

Analysts’ forecast error is lower when the quality of integrated reporting is higher.

19 Rensburg &

Botha 2013 How do stakeholders in South Africa react to the new reporting standards like integrated reporting

Few stakeholders use integrated reports as their main source of financial and

investment information, these reports are seen as additional information

Adams &

Simnett 2011 Opportunities and risks of integrated reporting for companies and stakeholders in Australia

Integrated reporting has advantages and risks. For stakeholders it will provide insight into the business and financial performance of an organization. The risk of transparency is competitiveness Healy & Wahlen 1999 Which specific accruals are used to

manage earnings and their impact on the economy

Public firms engage much more in earnings management than private firms.

Prior, Surroca &

Tribo 2008 Relationship between earnings management and CSR Positive impact of earnings management on CSR. The combination of these two has a negative impact on financial performance Roychowdhury 2006 The move from accrual based to

real earnings manipulation Firms manage earnings via real manipulation activities rather than accrual based earnings management.

*Research design Cohen and

Zarowin 2010 Accrual-based and real earnings management activities around seasoned equity offerings

Real earnings management has a direct effect on cash flows and is more difficult to detect than accruals

Graham, Harvey

& Rajgopal 2005 Earnings benchmark as incentive for managers Most earnings management is achieved via real actions as opposed to accounting manipulations

Cohen, Dey &

Lys 2008 Accrual-based and real earnings management activities in the period leading to the passage of SOX and in the period following passage of SOX.

After SOX, more real earnings

management, before SOX more accrual-based. Increase in accrual earnings

management is associated with an increase in option-based compensation

Boschen, Duru,

Gordon & Smith 2003 The long-run effects of unexpected firm performance on CEO compensation

CEO’s have financial incentives to meet earnings expectations and this is mostly the reason behind earnings management Payne & Robb 2000 Managers align earnings with

market expectations established by analysts’ forecasts

Senior executives make efforts to avoid negative earnings surprises since they have much to lose if their firms fail to meet expectations.

Othman and

Zeghal 2006 Factors that influence earnings-management policy with reference to the Anglo-American and Euro-Continental accounting models

Initial public equity offerings are strong motives for earnings management

Givoly, Hayn &

Katz 2010 Differential earnings quality of private equity and public equity firms

Private firms engage less in earnings management than public firms. Private firms have less problems due to agency problems.

Reimsbach, Hahn & Gurturk

2017 The influence of integrated reporting and assurance of sustainability information on investors’ information processing

Assurance of sustainability information is associated with an increase in the perceived sustainability performance and leads to higher investment-related judgments

20 Barth, Cahan,

Chen & Venter 2015 Economic consequences associated with integrated reporting quality: capital market and real effects

Integrated reporting is positively associated with both stock liquidity and firm value Bernardi and

Stark

2015 Impact of the reporting regime change in South Africa on analyst forecast accuracy based on Integrated reporting

No association between integrated

reporting and analyst forecast errors prior to the integrated reporting regime. Kim, Park &

Wier 2012 Association between earnings quality with CSR Socially responsible firms are less likely to manage earnings through discretionary accruals; manipulate real operating

activities and are less subject of investigations against top executives Chih, Shen &

Kang 2007 Relationship between accural earnings management and CSR CSR mitigates earnings smoothing, earnings losses avoidance but increases earnings aggressiveness

Davidson, Jiraporn, Kim & Nemec

2004 The relationship between earnings management and the Agency Theory

There is a relationship between the agency theory and earnings management. Earnings management is a type of agency costs if managers present financial reports that do not present an accurate economic picture of a firm.

21

3 Methodology

In this paper, an archival database study will be conducted using publicly available data, to run a regression model to answer the research question. The independent variable of the model that will be used to assess the research question, will be whether a firm engages in integrated reporting. The dependent variable of the model will be whether firms engage in real earnings management. The empirical part of this research is based on a cross-sectional analysis, which contains cross-sectional regression models to measure real earnings management. The regression model to measure the real earnings management is based on the model of Roychowdhury (2006). This model is developed by Dechow et al. (1995) but is used by Roychowdhury (2006), Cohen et al. (2008, 2010) and Zhang (2012). Because previous researches have already used them in their studies, the confidence in the validity of these proxies has increased. Dechow et al. (1995) and Roychowdhury (2006) use the abnormal levels of cash flow from operations, production costs, and discretionary expenses as proxies to measure real earnings management. To investigate the impact of integrated reporting on real earnings management, through the real activities manipulation, the levels of earnings management before and after the adoption of integrated reporting needs to be compared. Integrated reporting became mandatory in South Africa in 2010, so there will be a pre-integrated reporting period and a post-integrated reporting period.

3.1 Data collection and sample selection

The sample consists of companies in South Africa that are listed on the Johannesburg Stock Exchange (JSE). Since March 2010, Integrated Reporting has been mandatory for firms listed on the JSE (Villiers et al., 2014; IIRC, 2013). The characteristics of the JSE as a market, for example in terms of liquidity and market reaction, are similar to those of developed country bourses (Villiers et al., 2014). In a study setting such as the African JSE you can analyse the ‘before’ and ‘after’ the implementation of integrated reporting by listed firms. By using data from all the listed companies in South Africa, I know that I have a representative sample. In their research on listed firms in South Africa, Zhou et al. (2016) used a time span of four years beginning from 2009 to 2012. Although this research is similar on some points from the research of Zhou et al. (2012) I will use data from 2008 until the latest available data which is 2016.

As mentioned before, Roychowdhury (2005) uses three proxies for real earnings

management: abnormal levels of cash flows from operations, production costs, and discretionary expenses, but I will use only cash flows from operations and production costs, because the data on discretionary expenses on South African companies is unavailable. The necessary data will be

22 gathered from Wharton Research Data Service (WRDS). WRDS is a database and business intelligence tool with over 40.000 corporate, academic, government and non-profit clients at over 400 institutions in more than 30 countries (WRDS). The necessary data on South African listed companies will be gathered from Compustad Global which is a database from WRDS. The data will be gathered on annual bases (Roychodhury, 2006, Kim et al., 2012; Zang, 2012) and will be gathered from 2008 to 2016, and the option Global Company Key (GVKEY) will be chosen. Since all the data on South African companies from the JSE are needed, the entire database will be searched. When collecting the data, the industry format of the data must be set. This code indicates whether a company reports in an Industrial format or Financial Services. The Industry Format (INDFMT) describes the general industry presentation for the associated data record which includes companies reporting manufacturing, retail, construction and other commercial operations other than financial services. The Financial Services includes banks, insurance companies, brokers, real estate and other financial services. For this research it is important to exclude financial services since these companies will have a direct influence on the earnings management. Only non-financial firms will be included in this research (Zang, 2012; Kim et al., 2012). Based on the data, gathered from Compustad Global, of listed companies in South Africa we can see that listed South African Companies all use domestic accounting standards which are fully compliant with IFRS. South Africa therefore is the most suitable mandatory integrated reporting setting. Since all the listed South African companies use IFRS and they are all from one country, these two variables will not be used as an independent variable or control variable. The only independent variable that will be used in this research is integrated reporting. Before 2010 and after 2010.

After running the data search there is an initial sample of 300 firms and 2228 firm-year observations obtained from 2008 to 20164_{. After obtaining the data, the data will be cleaned by} dropping observations with missing or non-positive assets and missing industry codes. Duplicate firm-year observations are eliminated. Firms in regulated industries (sic codes between 4400 and 5000) and financial firms, bank and institutions (SIC codes between 6000 and 6500) are

eliminated (Zang, 2012). After eliminating financial institutions and bank, variables will be generated to measure the proxies and variables with missing observations will also be dropped. After generating new variables, the data will be winsorized (Kim et al., 2012; Zang, 2012). By winsorizing the data, statistics are transformed by limiting extreme values in the statistical data to reduce the effect of possible outliers that will influence the results. The top and bottom 1% of firms are removed to minimize extreme observations’ effect on regression results. To reliably

23 measure the real earnings management model, firms with more than 15 observations for each industry group are kept in the analyses and the rest of the firms are dropt (Roychowdhury, 2006; Zang, 2012; Kim et al., 2012). After the reduction of the data there is a final sample of 802 observations.

3.2 Measurement of earnings management

In contrast to accrual-based earnings management, there are not many models that measure real earnings management. There are different studies that measure real earnings management (Cohen et al., 2008; Zang, 2012; Kim et al., 2012 and Roychowdhury, 2006) and all of these studies use the model explained by Roychowdhury (2006). Although the real earnings management model is most of the time referred to as the Roychowdhury model, Dechow et al. (1998) were the first that introduced the model. To capture real earnings management, the research design of

Roychowdhury (2006) uses a model that measures normal levels of cash flow from operations (CFO), discretionary expenses (DISEXP) and production costs (PROD) for every firm-year. As mentioned before, discretionary expenditures is not included in this research because of lack of data from R&D cost, advertising cost, and SG&A expenses of the South African companies. Zang et al. (2012) used also two proxies (DISEXP and PROD) and refered to the studies of Cohen et al. (2008) and Cohen and Zarowin (2010). In this study the abnormal levels of cash flows from operations and production costs will be estimated and if these proxies deviate from the normal levels, they become abnormal. Following Cohen et al. (2008), in order to capture the effects of real earnings management through these two variables in a comprehensive measure, a single variable will be computed by combining the two individual real earnings management variables. The third variable COMBINED_REM will be computed as the sum of the variables, CFO and PROD. However the two individual variables have different implications for earnings that may disappear when using only the COMBINED_REM alone. So the results corresponding with COMBINED_REM as well as the two individual variables will be reported. A regression model contains dependent variables, control variables and independent variables. The proxies CFO and PROD are the dependent variables of the regression model. The independent variable is integrated reporting and since all listed firms in South Africa use integrated reporting, the regression model doesn’t need proxies for integrated reporting. Dummies will be used for the regression model to capture the ‘before’ and the ‘after’ period of integrated reporting.

24 3.2.1 Robustness test

When running a regression model, it is important to understand the assumption of

homoscedasticity (same variance), since homoscedasticity is central to a linear regression model (Field, 2014). Homoscedasticity describes a situation in which the error term (that is, the “noise” or random disturbance in the relationship between the independent variables and the dependent variable) is the same across all values of the independent variables. According to Field (2014) heteroscedasticity is the violation of homoscedasticity, and is present when the size of the error term differs across values of an independent variable. When heteroscedasiticy increases, the homoscedasticity will decrease and the variability of the variable will be unequal across the range of values of the second variable that predicts it (Jarque & Bera, 1980). A robust regression will be conducted to resolve the problem of heteroskedasticity (Zang, 2012; Cohen et al., 2008). The regression standard errors will be adjusted and the results based on robust standard errors will be reported. So the standard errors will be corrected for heteroscedasticity. To solve for the cross-observation correlation in the variables a clustered standard errors will be used, which are robust standard errors that are corrected for unwanted correlation (Zang, 2012; Cohen et al., 2008). When there are such unwanted correlations in the panel data across years the standard errors based on years will be clustered (table 5). After running a robust regression based on years, another robust test will be conducted based on the firms (gvkey) and another test with additional control variables.

3.2.2 Real earnings management model

This research uses two proxies (ABN_PROD and ABN_CFO) and a combined proxy (COMBINED_REM). The first step is to examine levels of real earnings management by calculating the normal levels of CFO and PROD. The abnormal CFO and PROD are computed as the difference between the actual values and the normal levels predicted from the equations of proxy 1 and proxy 2. The abnormal levels of CFO and PROD will be predicted using the residual term, which is the error term of the formula (ε).

Proxy 1: Abnormal production costs

Overproduction, or increasing production to report lower COGS.

To manage earnings upward, managers of manufacturing firms can produce more goods than necessary to meet expected demand (Cohen et al., 2008; Roychowdhury, 2006). With higher production levels, fixed overhead costs are spread over a larger number of units, lowering fixed

25 costs per unit. The abnormal production costs will be measured by the sum of COGS and change in inventory during the year, and expenses as a linear function of contemporaneous sales

(Roychowdhury, 2006).

The production costs is defined as 𝑃𝑅𝑂𝐷𝑡 = 𝐶𝑂𝐺𝑆𝑡+ Δ𝐼𝑁𝑉𝑡

𝐶𝑂𝐺𝑆_𝑡/𝐴_𝑡−1 =









COGS = the cost of goods sold in year t, A = is the total assets;

At-1 = Lagged total assets;

S = net sales;

α0, α1,β1 and β2 = parameters;

ε = error term; the value that captures the level of abnormal operating cash flow Inventory growth:

Δ𝐼𝑁𝑉_𝑡/𝐴_𝑡−1= 𝛼₀+ 𝛼₁(1/𝐴_𝑡−1) + 𝛽₁(Δ𝑆_𝑡/𝐴_𝑡−1) + 𝛽₂(Δ𝑆_𝑡−1/𝐴_𝑡−1) + 𝜀_𝑡 ΔINV = INVt - INVt-1, change in inventories;

A = is the total assets; At-1 = Lagged total assets;

S = net sales;

S = St - St-1, change in sales; S t-1 = Lagged change in sales α0, α1,β1 and β2 = parameters;

ε = error term; the value that captures the level of abnormal operating cash flow Production cost

𝑃𝑅𝑂𝐷_𝑡/𝐴_𝑡−1 = 𝛼₀+ 𝛼₁(1/𝐴_𝑡−1) + 𝛽₁(𝑆_𝑡/𝐴_𝑡−1) + 𝛽₂(Δ𝑆_𝑡/𝐴_𝑡−1) + 𝛽₃(ΔS_𝑡−1/𝐴_𝑡−1) + 𝜀_𝑡 PROD = Production costs in period t, sum of cost of goods and the change in inventory At-1 = Lagged total assets;

S = net sales;

S = St - St-1, change in sales; S t-1 = Lagged change in sales

26 α0, α1,β1 and β2 = parameters;

ε = error term; Proxy 2: Cash flow

Sales manipulation;

Accelerating the timing of sales or generating additional unsustainable sales through increased price discounts or more lenient credit term (Roychowdhury, 2006; Cohen et al., 2008). With the use of the coefficients derived with the historical data, the ‘normal’ level of CFO is determined which is compared to the actual level of the CFO. The difference between these levels is indicative for the use of real earnings management. Sales manipulation can be measured by the abnormal cash flow from operations. In order to estimate the abnormal CFO, first the normal CFO is estimated. This is expressed as a linear function of sales (St) and the change in sales (St). With this equation, the coefficients are estimated. All the variables in the proxy are scaled by lagged total assets (At-1) so that heteroskedasticity is reduced.

t t t t t t t t A A S A S A CFO / _1











At-1 = Lagged total assets;

S = net sales;

S = St - St-1, change in sales; α0, α1,β1 and β2 = parameters;

ε = error term;

For every firm-year, abnormal cash flow from operations (AB_CFO) is the residual from the corresponding industry-year model and the firm-year’s sales and lagged assets.

3.3 Control variables

Control variables are used to determine whether earnings management through real activities manipulation has been influenced by integrated reporting. Therefore, a regression analysis is used that contains a dependent variable and control variables. There are different control variables included, which could affect real earnings management, to avoid the problem of correlated

27 omitted variables. Based on previous studies and literature (Roychowdhury, 2006; Richardson et al., 2002; Cohen et al., 2008, Cohen and and Zang, 2012) on real earnings management, there are some control variables that could influence real earnings management. The control variables included in this research are: firm size, net income (NI), leverage, sales growth and cash flows. According to Roychowdhury (2006) the size of a firm can have a significant variation in earnings management. Richardson et al. (2002) examined firm size as a determinant of earnings management, and they conclude that larger firms are subject to closer scrutiny by investors and analysts. According to Richardson et al. (2002) capital market pressures are greater for larger firms leading to the adoption of aggressive accounting policies like earnings management. In contrast to Richardson et al. (2002), Watts and Zimmerman (1990) found evidence that management of large companies can have the intention to manage earnings downwards. The reason behind this phenomenon is that large firms have more political and media attention and this leads to higher costs (Watts & Zimmerman, 1990). In order to reduce these costs, managers choose to manage their earnings downwards. Zang (2012) uses industry-adjusted log value of total assets to control for relative firm size in the industry. The logarithm of the total assets will be calculated and used as a control variable.

Roychowdhury (2006) used the net income as a control variable in the regression in his study. The net income figure is scaled by lagged total assets. The net income is similar to the return-on-investment (ROA). Net income is measured as income before extraordinary items scaled by lagged total assets, expressed as deviation from the corresponding industry-year mean. The variables are extracted from COMPUSTAT Global as a ratio between IBC and lagged total assets. Duke and Hunt (1990) used leverage as a proxy for the tightness of debt covenant restraints. According to Duke and Hunt, the higher the leverage, the more stressful the debt covenant restrain for the firm is, thus the higher probability for the firm to violate the debt covenant. Callao and Jarne (2010) state that in stressful financial situation, managers have more incentives to manipulate earnings. They provide evidence that higher leverage causes greater extents of earnings management. Leverage is calculated as the long term debts divided by the total assets of the company. Park and Shin (2004) also use financial leverage as a proxy, which they obtain as the ratio of the long term debt to total assets.

Cohen et al. (2008) used growth of sales as a proxy for earnings management in their research. The sales growth is measures as the change in sales divided by lagged sales, which will be calculated as the change in sales divided by lagged sales.

28

3.4 Regression model

The regression contains dependent variables, control variables and independent variables. The proxies abnormal cash flow, abnormal production costs and combined real earnings management are the dependent variables of the regression model. The independent variable is the dummy variable zero and one. The value of zero is appointed to the period before integrated reporting and the value of one is appointed to the period after the adoption of integrated reporting. The control variables are the variables firm size, net income, financial leverage, growth of sales and cash flows. In this research three regression analyses will be conducted. One for the combined ABN_PROD + ABN_CFO (COMBINED_REM) and the other two are the separate proxies of abnormal cash flow and abnormal production costs.

Regression 1 abnormal production costs:

𝐴𝐵𝑁_𝑃𝑅𝑂𝐷𝑡 = 𝛼0+ 𝛽1𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑𝑟𝑒𝑝𝑜𝑟𝑡𝑖𝑛𝑔 + 𝛽2𝑠𝑖𝑧𝑒 + 𝛽3𝑛𝑒𝑡𝑖𝑛𝑐𝑜𝑚𝑒 + 𝛽4𝑓𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒

+ 𝛽5𝑠𝑎𝑙𝑒𝑠𝑔𝑟𝑜𝑤𝑡ℎ

Regression 2 abnormal cash flows:

𝐴𝐵𝑁_𝐶𝐹𝑂𝑡= 𝛼0+ 𝛽1𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑𝑟𝑒𝑝𝑜𝑟𝑡𝑖𝑛𝑔 + 𝛽2𝑠𝑖𝑧𝑒 + 𝛽3𝑛𝑒𝑡𝑖𝑛𝑐𝑜𝑚𝑒 + 𝛽4𝑓𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒

+ 𝛽5𝑠𝑎𝑙𝑒𝑠𝑔𝑟𝑜𝑤𝑡ℎ

Regression 3 combined real earnings management: COMBINED_REM = ABN_CFO - ABN_PROD

𝐶𝑂𝑀𝐵𝐼𝑁𝐸𝐷_𝑅𝐸𝑀𝑡

= 𝛼0+ 𝛽1𝑖𝑛𝑡𝑒𝑔𝑟𝑎𝑡𝑒𝑑𝑟𝑒𝑝𝑜𝑟𝑡𝑖𝑛𝑔 + 𝛽2𝑠𝑖𝑧𝑒 + 𝛽3𝑛𝑒𝑡𝑖𝑛𝑐𝑜𝑚𝑒 + 𝛽4𝑓𝑙𝑒𝑣𝑒𝑟𝑎𝑔𝑒

+ 𝛽5𝑠𝑎𝑙𝑒𝑠𝑔𝑟𝑜𝑤𝑡ℎ

Table 2 contains a summary with all the variables used in the methodology section.

Table 2 Variables used in the regression

Variable

COGS Cost of goods sold in year t,

ΔINV INVt - INVt-1, change in inventories;

PROD Production costs in period t, sum of cost of goods

and the change in inventory

CFO Cash flow from operations in year t;

A Total assets;

At-1 Lagged total assets

29 S St - St-1, change in sales;

S Lagged change in sales

ε error term

α0, α1,β1 and β2 parameters

Firm size Logarithm of the total assets

Net income Net income scaled by lagged total assets

Leverage Long term debts divided by total assets

30

4

Results

In this chapter the empirical results of the model will be discussed. It will start with the descriptive statistics than the correlation and regression will be discussed. The last section will present additional tests.

4.1 Descriptive Statistics

In table 3 descriptive statistics for the sample are presented. All the variables are winsorized at the top and bottom one percent of their distribution. There are around the 802 and 723 observations of this sample. The mean of COMBINED_REM is zero and the median is also zero. The mean of ABN_PROD is zero which is not surprising since the abnormal production costs are estimated as the residual of regression proxy 1. The residual of the proxy is by definition zero. If the residual in this sample was not zero, either positive or negative, this would mean that there is a problem. The standard deviation of ABN_PROD is not high (0.182) since the variables are winsorized. The median of ABN_PROD is (almost) zero and has almost the same value as the mean of ABN_PROD, but there is a small negative difference due to the winsorizing of one percent. The mean of ABN_CFO is also zero and the standard deviation is also not so big. The median of ABN_CFO is zero, which has the same value of the mean. The variable integrated reporting stand for the year 2010 to 2016, and has an average of 75 percent which indicates that 75 percent of the sample falls in the range of the year 2010 to 2016. The rest of the sample lies in the range of the years 2008 and 2009. The average of 75 percent is a good indicator of the

integrated reporting years, since it consist of 7 years. So it is a logical spread. The mean of the variable size is 7.210, since this is a logarithm, the exponent of the 7.210 is 1353.369 this means that the average market value of a company in this sample lies around the 1,353,369,000 million South African Rand (ZAR). The average net income of the sample is 5 percent which is not very high which is in accordance with the study of Roychowdhury (2006). The net income reflects on the firm performance which could be a reason to manipulate their earnings. There is a lot of spread of the net income in the sample. The average leverage in the sample of firms is almost 10 percent. This means that for every ZAR on the asset side of the balance, the firm has 0.10 ZAR as debt on the balance. The average leverage of the sample is not so big. According to Duke and Hunt (1990), the higher the leverage, the more stressful the debt covenant restrain for the firm is, thus the higher probability for the firm to violate the debt covenant. Callao and Jarne (2010) state that in stressful financial situation, managers have more incentives to manipulate earnings. They provide evident that higher leverage causes greater extents of earnings management. In this