Faculty of Economics and Business
MSc Accountancy and Control, Accountancy track
Could a highly independent board of directors reduce real earnings
management in a financially-distressed firm?
Student: Ioana Padureanu Student number: 10622780
Date of submission: 23rd of June 2014 Supervisor: Prof. Dr. Vincent O’Connell
2
Abstract
This study investigates whether a highly independent board of directors could reduce managers’ attempts to engage in real earnings management when the company experiences financial distress. Using a sample of 261 U.S. non-financial companies from three different industries (pharmaceuticals, information technology and industrials) and a time frame from 2008-2012, I test the association between real activities manipulation, board’s characteristics and financial distress. Real earnings management is measured using three models: abnormal level of cash-flow from operations, abnormal level of production costs and abnormal level of discretionary expenses. For board of directors’ characteristics I use a composite index score, while financial distress is determined with Z Score Model. The results of the thesis reveal that real earnings management and financial distress are positively related, while an independent board can reduce real activities manipulation when they are expressed by mean of overproduction.
3
Acknowledgements
First of all, I would like to thank Professor Vincent O’Connell, my research supervisor, for guiding me through the whole research. His support and feedback helped me finish my thesis and deliver a good outcome on time.
Secondly, I would like to address special thanks to Razvan Rosculet, who has supported me through my entire research, providing me with valuable remarks and suggestions for the thesis. Also, many thanks I address to Andrei Padureanu, who reviewed my results, too.
Last but not least, I would like to thank my parents and my beloved ones for their moral support throughout the whole research process, which helped me complete my work.
4
Table of Contents
1. Introduction ... 5
2. Literature review and hypotheses development ... 9
2.1. Theoretical background ... 9
2.1.1. Agency Theory ... 9
2.1.2. Positive Accounting Theory ... 10
2.1.3. Stakeholder and Legitimacy Theory ... 10
2.2. Explaining real earnings management ... 11
2.3. The role of board of directors in reducing real activities manipulation ... 13
2.4. Real earnings management in firms that experience financial distress ... 16
3. Research methodology ... 18
3.1. Sample... 18
3.2. Real earnings management metrics ... 20
3.3. Board metrics ... 22
3.4. Financial distress metrics ... 24
3.5. Control variables ... 25
3.6. Empirical model ... 26
3.7. Data analysis ... 27
4. Empirical results ... 29
4.1. Descriptive statistics and univariate analysis ... 29
4.1.1. Descriptive statistics ... 29
4.1.2. Univariate analysis ... 30
4.2. Real earnings management models ... 33
4.3. The influence of board of directors and financial distress characteristics on real earnings management ... 37
4.3.1. Results of the association between real earnings management and board of directors metrics 37 4.3.2. Results of the association between real earnings management and financial distress metrics . 41 4.3.3. Results of the association between real earnings management, board of directors and financial distress metrics ... 45
5. Conclusions ... 50
References ... 52
5
1. Introduction
The purpose of this thesis is to consolidate the research area of real earnings management by examining the relationship between real activities manipulation, board of directors’ independence and financial distress.
While prior literature has broadly documented earnings management by means of accruals, real activities manipulation has been paid relatively little attention. Real earnings management represents actions taken by managers that affect earnings, in order to achieve a certain earnings objective. It is a topic that deserves investigation, because it is not based on manipulation of accounting numbers, but on economic decisions. This is why real earnings management cannot be easily identified in practice, compared to accrual-based earnings management. If it is disclosed properly in the financial statements, real earnings management cannot influence the auditor’s opinion.
For a better understanding of the core concepts presented in the thesis, I examined a number of economic theories. Firstly, I identified the principal-agent relationship that exists between shareholders (principal) and management (agent) and the related agency problems occurred in a company, according to Jensen and Meckling (1976). Based on this situation and considering the interest of managers to maximize their own wealth, managers could be tempted to manipulate earnings in order to meet the performance targets, just to gain the related “promised prize”. Secondly, positive accounting theory is another underlying theoretical background for earnings manipulation, which shows that managers choose accounting methods that show a true representation of company’s performance. Due to the subjectivity of the selection process, Watts and Zimmerman (1978) support the perspective that individuals act to maximize their own utility by being resourceful and innovative. The last theories that I referred to are the stakeholder theory and the legitimacy theory. They highlight another reason for earnings manipulation, showing that firms try to obtain and maintain a positive image, being constrained to meet stakeholders’ expectations.
Graham et al. (2005) found that most earnings management is realized via real actions as opposed to accounting manipulation. The preference for shifting from accrual-based earnings management to real earnings management is a consequence of the blemish attached to accounting fraud in the post-Sarbanes-Oxley period. Due to major accounting scandals that
6
happened at the beginnings of the 2000s (Enron – 2001, WorldCom – 2002, AOL – 2002), the government oversight agencies in USA, like Securities and Exchange Commission (SEC), have tightened the regulation in order to prevent fraud. This fact determined a major decrease in accrual-based earnings manipulation.
Roychowdhury (2006) identified three ways of engaging real earnings management: increasing sales through increased price discounts, overproduction which leads to lower cost of goods sold and decreasing discretionary expenses, including advertising, R&D and SG&A expenses. All these activities affect cash-flows.
The question that arises is how can this real earnings management practice be diminished? Can the corporate governance mechanisms determine such a decrease? In order to answer to these questions I examined the board of directors’ characteristics, focusing on board independence.
The board of directors is responsible for instituting an appropriate control system within the organization and for monitoring top management’s compliance with this system (Beasley 1996). Prior researches documented that the composition of individuals who are part of the board is an important factor for the effectiveness of monitoring management actions (Fama 1980, Fama and Jensen 1983). This effectiveness is determined by both inside and outside members. In order to avoid managerial discretion, it is preferred to have a greater fraction of outside directors as members of the board. These independent directors mitigate serious agency problems by acting as mediators in conflicts that appear between internal managers. An active, informed and independent board monitors and prevents, or at least discourages managers, from enriching themselves (Baber et al. 2012). It also provides a better representation and protection of the owners’ interest.
This leads to the research question of the thesis: could a highly independent board of directors reduce managers’ attempts to engage in real earnings management when the company is experiencing a financial distress situation?
Prior studies demonstrated that firms which experience a financial distress situation could create repulsion among customers in planning to do business with them (Opler and Titman 1994). In order to keep and/or attract customers, managers should create a positive image of the firm. By accelerating the timing of sales through increased price discounts or more lenient credit terms, as a mean of real earnings management, they could influence customers’ decisions about
7
reconsidering doing business with the company. Another reason for applying real activities manipulation is to attract capital. Experiencing financial distress means that firms are not able to honor their financial obligations with their creditors. Increasing earnings and creating a good image of the company could attract financial resources from investors.
My study was conducted in the U.S. economic context. The sample consists of 261 non-financial companies, part of three industries: information technology, pharmaceuticals and industrials (the industrial companies being active in areas such as construction and engineering, electrical equipment and machinery). I chose these industries because they provide relevant information for real earnings management measures, especially for research and development expenses. The data was collected from three online databases (Compustat, Risk Metrics and Audit Analytics) for the period 2008-2012.
I tested the hypotheses combining the statistical models developed by Dechow et al. (1998) – for real earnings management (abnormal level of cash-flow from operations, abnormal level of production costs and abnormal level of discretionary expenses), Altman (1968) – for financial distress metrics (Z score model) and Baber et al. (2012) – for board of directors’ metrics (composite index score).
The results do not fully support the hypotheses. Regarding the connection between board independence and real earnings management, I found that a highly independent board could reduce management manipulation of earnings when real earnings management is indicated by the abnormal level of production costs. The same finding is observed when the proxy for financial distress is introduced. Another result shows that real earnings management and financial distress metrics are positively connected. The statement applies for all three measures of real earnings management.
From academic considerations, this research contributes to prior literature in the study area of real earnings management. It is a practice of earnings manipulation that has been increasing in the past decade, after major accounting scandals, which marked the accounting world. The literature for earnings management contains plenty of studies of accrual-based earnings management, while real activities manipulation is less studied. The innovation of this thesis comes from the examination of three concepts that are interconnected: real earnings management, board of directors’ independence and financial distress. These three concepts were not studied together before.
8
The remainder of my thesis is organized as follows. Chapter 2 presents an overview of prior literature regarding real earnings management, board of directors’ independence and financial distress. All three concepts are connected resulting in the hypotheses establishment. Chapter 3 explains the sample selection procedure and the statistical models used for testing the hypotheses. Chapter 4 presents the results for the relationship between real earnings management, board independence and financial distress, while Chapter 5 shows the conclusions of the thesis.
9
2. Literature review and hypotheses development
This chapter is an overview of relevant prior literature regarding the influence of a highly independent board of directors in diminishing managers’ attempts to engage in real activities manipulation of earnings, while their companies go through a financial distress situation.
Section 2.1 introduces some economic theories that are relevant for my study, like agency theory, positive accounting theory, stakeholder theory and legitimacy theory.
Section 2.2 presents the concept of real earnings management and the three ways employed by managers to manipulate earnings.
Section 2.3 explains the role of the board of directors in reducing real earnings management, while Section 2.4 investigates managers’ tendency to apply real activities manipulation of earnings in firms that experience financial distress.
2.1. Theoretical background
2.1.1. Agency Theory
I consider agency theory a relevant starting point in explaining and understanding the core of all the concepts that are presented here. Agency theory describes the principal-agent relationship that exists between shareholders (principals) and managers (agents) in a company.
Managers are designated by the company’s owners to run the business, while representing their best interests. Hence, a first problem, addressed by the agency theory, arises. It refers to the conflict of interests between owners and managers because of the misalignment that appears among their objectives (Jensen and Meckling 1976).
The second problem addressed by the theory refers to the different attitudes that principals and agents have towards risk, regarding risk tolerance and the manner of approaching risky situations.
In order to reduce this conflict and align the interests of managers with those of shareholders, Shapiro (2005) suggests the use of compensation schemes. Performance bonuses incentivize managers to reach predetermined targets that create benefits for company’s owners. As for now, both parties achieve their goals. Managers receive high payments for their performance, while shareholders obtain high dividends because the business is profitable.
10
Due to the fact that managers’ interest is to maximize their own wealth, they could be tempted, in case of an unfavorable situation like financial distress, to manipulate earnings in order to meet the performance targets, just to gain the “promised prize”.
2.1.2. Positive Accounting Theory
Positive accounting theory (PAT) explains and predicts why managers apply specific accounting practices. Managers choose accounting methods that show a true representation of the company’s performance. Within this perspective, the selection process might be subjective, because managers could choose the accounting practices that serve best their own interest. Watts and Zimmerman (1978) support this perspective, asserting that individuals act to maximize their own utility. In doing so, they are resourceful and innovative.
Dechow (1994) found that earnings are a better measurement of firm’s performance, on a short-term horizon. Earnings are used as a summary measure of firm’s performance by a wide range of users. Because of the fact that earnings are considered an important topic for many stakeholders, this gives managers incentives to manipulate accounting numbers by applying real earnings management, in order to boost earnings. In this way they could attract capital in the company, by emphasizing the profitability of the firm and they could also reach certain performance thresholds in order to collect their bonuses and therefore increase their own wealth. 2.1.3. Stakeholder and Legitimacy Theory
Prior studies (Donaldson and Preston 1995, Guthrie and Parker 1989, Wilmshurst and Frost 2000) have shown that stakeholder and legitimacy theory use the assumption that organizations are part of a broader social system in which they impact and are impacted by the actions of other groups.
Legitimacy theory identifies society as one group, while stakeholder theory divides society into different stakeholder groups. Besides distinguishing different types of influence that stakeholder groups have over the organization, stakeholder theory also takes into account stakeholder power. This power puts the company under pressure, constraining it to meet stakeholders’ expectations.
Legitimacy theory puts pressure on corporate management to react to community expectations. Stakeholders within a community establish which activities are acceptable and
11
companies, “as members of the community, are expected to carry out their activities within the boundaries of what is deemed acceptable by that community” (Wilmshurst and Frost 2000, pp. 11).
Having to accomplish many expectations of different types of stakeholders might create conflicts of interests among them. Also, prioritizing the expectations considering the power of stakeholders leads to the necessity of obtaining and maintaining a certain image of the company. Because markets are in a state of continuous change, which creates an uncertain business environment, and due to the need of having a permanent community acceptance, corporate management might employ certain inappropriate activities (managing earnings) to maintain the good image created.
2.2. Explaining real earnings management
Earnings management refers to techniques of producing financial reports that might show a positive and favorable picture of a company’s business activities and financial position. Earnings management demonstrates flexibility when companies have to recognize revenues and expenses by taking advantage of how accounting rules are applied.
Earnings management is a topic examined by many researchers in their studies. They provide well-grounded evidence that executives engage in earnings management (Healy 1985, Guidry et al. 1999, Defond and Jiambalvo 1994, Kasznik 1999). As Roychowdhury presented in his paper (Roychowdhury 2006), there are two means of managing earnings. The first one is realized by manipulation of accruals with no direct cash-flow effects, while the second one incentivizes managers to manipulate real activities during a fiscal year in order to meet certain earnings targets. Real activities manipulation affects cash-flows.
Cohen and Zarowin (2010) focus their research on real earnings management. They define it as actions employed by managers that deviate from the normal business practices, undertaken with the primary objective of misleading particular stakeholders to believe that earnings thresholds have been achieved in the normal course of operations.
According to Roychowdhury (2006) there are three ways of engaging in real earnings management:
(a) boosting sales through accelerating their timing and/or generating additional unsustainable sales through increased price discounts or more lenient credit terms;
12
(b) overproducing and thus allocating more overhead to inventory and less to cost of goods sold, which leads to lower cost of goods sold and increased operating margins;
(c) aggressively reducing discretionary expenses, which are defined as the sum of advertising, R&D and SG&A expenses, to improve margins (this reduction is most likely to occur when discretionary expenses do not generate immediate revenues and income).
Prior studies have shown evidence of firms distorting real activities to manage earnings (Zang 2012, Roychowdhury 2006, Gunny 2005). Furthermore, Graham et al. (2005) found that most earnings management is realized via real actions as opposed to accounting manipulation. Managers are willing to perform real economic actions such as decreasing advertising, R&D and SG&A expenses or even rejecting positive NPV (net present value) projects, in order to meet the short-term earnings targets: ”80% of the survey participants report that they would decrease discretionary spending on R&D, advertising and maintenance to meet an earnings target, while more than half (55.3%) state that they would delay starting a new project to meet an earnings target, even if such a delay entailed a small sacrifice in value”1
. The preference of shifting from accrual-based earnings management to real earnings management is a consequence of the blemish attached to accounting fraud in the post-Enron and post-Sarbanes-Oxley periods.
Real activities manipulation changes the timing and the structuring of business transactions, which determines these activities to take place during the fiscal year. Zang (2012) states that after the year-end, the outcome of the real earnings management appears and managers can no longer engage in it. When managers distort real business decisions in order to manage earnings, they cannot perfectly control with certainty the amount of the manipulative activities achieved. A relevant example in this case is a pharmaceutical company that reduces R&D expenditures from the current period by stopping or delaying development of a drug. This decision could lead to the shutting down of the research site. The company’s management could make an approximate estimation of the financial impact on R&D expenditure from these decisions, but it does not have perfectly relevant information about this matter. So, after the end of the fiscal year, real activities manipulation could be higher or lower related to the amount that was originally anticipated. This underlines the idea that real earnings management is characterized by uncertainty.
1
Graham, J. R., Harvey, C. R., Rajgopal, S., 2005, “The economic implications of corporate financial reporting”, Journal of Accounting and Economics 40 (2005) 3–73, pp. 32.
13
There is evidence brought by Gunny (2005) that all types of real earnings management have a significant impact on further operating performance. Real earnings management activities are associated with lower return on assets. The author found that, besides discretionary expenses, the other two types of real activities manipulation are associated with a lower level of future cash-flow. Additionally, managing earnings by accelerating the timing of sales, overproduction and decreasing discretionary expenses is associated with lower ROA. Identifying the ways of engaging in real earnings management is considered gradually informative about future earnings and cash flows.
Another important topic related to real earnings management is its relationship with the auditor. Many authors demonstrated in their studies that higher quality auditors reduce the level of accrual earnings management (Becker et al. 1998, Johnson et al. 2002). Chi, Lisic and Pevzner (2011) revealed that this correlation does not work in the same way for real earnings management, meaning that higher audit quality does not decrease real activities manipulation.
Real earnings management, as long as it is properly disclosed in the financial statements, cannot influence auditors’ opinions (Chi et al. 2011). Therefore, managing real activities is less costly for managers, compared to accrual earnings management, because it is less likely to attract control.
Auditors constrain management’s ability to manage accruals. Ewert and Wagenhofer (2005) documented that firms’ preferences for real earnings management increase when their accounting flexibility is reduced. The accounting flexibility is reduced by the auditor who will not tolerate earnings management. So, higher quality auditors are more likely to reduce accrual earnings management (accounting manipulation), encouraging managers to engage in real activities manipulation. Therefore, higher audit quality could be associated with higher levels of real earnings management practices. Chi et al. (2011) found that this shift in managing earnings is potentially more costly to the shareholders on the long run. The authors suggested that mandatory audit firm rotation could potentially reduce real earnings management.
2.3. The role of board of directors in reducing real activities manipulation
Many accounting failures of high-profile U.S. companies were caused by misleading financial reporting, but also by poor corporate governance characteristics (Baber et al. 2012). Regarding corporate governance systems, they are classified as either external or internal
14
corporate governance mechanisms (Cremers and Nair 2005, Brown and Caylor 2005). The external governance mechanism does not support active stakeholder participation in the management process. It comprises the market for corporate control, which has been documented by many researchers to be an effective corporate governance mechanism (Jensen 1993, Shleifer and Vishny 1997).
On the other hand, internal governance mechanisms focus on the relationships between firm insiders (managers, directors, employees). One mechanism of internal governance is the board of directors. Fama and Jensen (1983) highlighted an important function of the board, which is to minimize costs that arise from the separation of ownership and decision control of the modern-day corporation. Even if the board delegates decision rights and decision control functions to the top management, the board has the final control over the top management. Beasley (1996) identifies these controls as the board’s right to approve and monitor important decisions and to choose, dismiss and reward important decision agents. The board of directors is responsible for instituting an appropriate control system within the organization and for monitoring top management’s accordance with this system.
Prior researches documented that the composition of individuals who are part of the board of directors is an important factor for effectively monitoring the management actions (Fama 1980, Fama and Jensen 1983). The effectiveness of the monitoring process is determined by both inside and outside (independent) members of the board of directors.
Inside members have an important role in the board composition, because they have useful specific information about the company’s activities, gathered from internal monitoring of subordinate managers. However, this could lead to a high level of decision discretion of top managers. Williamson found in his study (Williamson 1984) that the advantage of having a full-time status and insider knowledge could affect the objectivity of the board of directors, transforming it into an instrument of management. Thus, the interests of the shareholders would be compromised.
Outside members’ role is to mitigate serious agency problems by acting as a mediator in conflicts that appear between internal managers. Fama (1980) and Fama and Jensen (1983) found that the board’s strength is enhanced by the inclusion of outsiders, being more effective in the monitoring process of management.
15
An increased number of outside members enhance the independence of the board. Board independence is an indicator of the board’s ability to monitor, discipline and influence management. An active, informed and independent board monitors and prevents, or at least discourages managers from enriching themselves (Baber at al. 2012).
Fich and Shivdasani (2007) suggested that outside board members are concerned about their reputation for providing objective and effective financial reporting oversight, as they reach after membership on other boards. Thus, independent directors are more incentivized, compared to insiders, to provide financial reporting supervision. Therefore, firms with independent boards (strong internal governance systems) are less likely to experience misleading financial reporting than firms with manager-dominated boards, as concluded by Baber et al. (2012).
Klein (1998) found that there is a negative correlation between board independence and earnings manipulation. A highly independent board of directors reduces managers’ attempts to engage in earnings management. This statement has also been validated by other researchers. Dechow, Sloan and Sweeney (1996), Byrd and Hickman (1992) confirm that independent directors influence board decisions and are capable of detecting and constraining earnings management practices, in the situation of accruals manipulation. Accounting manipulation through accruals can be easily identified compared to real activities manipulation.
Because my thesis is focused on real activities manipulation, as a manner of earnings management, I aim to analyze if the correlation between board independence and accrual manipulation is applicable for real earnings management too. Therefore, I formulated the first hypothesis:
H1: A high level of board independence will reduce managers’ attempts to engage in real
earnings management.
I want to extend previous research in this area by examining the influence of independent boards, which might lead to a reduction on real earnings management, if expanded.
16
2.4. Real earnings management in firms that experience financial distress
“Corporate distress is a sobering economic reality reflecting the uniqueness of the American way of corporate death”2
. In financial economics literature, financial distress has been portrayed as a costly event that determines companies’ optimal capital structures. Financial distress is seen as a costly process from debt holders and non-financial stakeholders’ point of view (customers, suppliers, employees), because it undermines access to credit and raises costs of stakeholder relationships. These dispositions arise due to the conflicts of interest between borrowers and lenders (Jensen and Meckling 1976, Stulz 1990), between companies and their non-financial stakeholders (Titman 1984) and between shareholders and managers (Gilson and Vetsuypens 1993).
Jensen (1989) and Wruck (1990) found that financial distress could improve firm’s performance and support changes in corporate form that are financed primarily with debt. Financial distress can improve firm value by compelling managers to take difficult value-maximizing decisions, which they would normally avoid.
Altman (1968) determines financial distress using different measures. The variables are classified into five ratio categories: liquidity, profitability, leverage, solvency and activity. They are considered effective indicators and predictors of corporate distress.
Opler and Titman (1994) found that highly leveraged firms lose market share to their less leveraged competitors. This highlights that customers refuse to do business with financially-distressed firms. Taking this argument into consideration, in order to keep or gain new customers, managers should create a positive image of the firm, even though it passes a financial distress situation. A solution is to engage real earnings management. Accelerating the timing of sales through increased price discounts or more lenient credit terms could influence customers’ decisions about whether to do business with a financially-distressed company.
Besides attracting new customers, another reason for firms to apply real activities manipulation is to attract capital. Financial distress indicates a condition when promises made to creditors are broken or are honored with difficulty. If firms cannot be relieved from this situation, it can lead to bankruptcy. Increasing earnings, by means of real earnings management,
2 Altman, E., Hotchkiss, E. 2006. “Corporate Financial Distress and Bankruptcy: Predict and Avoid Bankruptcy,
17
helps to attract funds, which is beneficial for companies seeking to meet their financial obligations.
In this respect, I stated the second hypothesis:
H2: Firms that experience financial distress are more likely to apply real transaction
based earnings management in order to boost earnings.
Prior literature documented that managers of firms experiencing financial distress have strong incentives to engage in earnings manipulation. In this way they manage to send positive signals or at least they reduce the impact of negative signals emanated from financial distress (Sweeney 1994, DeAngelo et al. 1994, Burgstahler and Dichev 1997).
As DeAngelo (1988) and Moyer (1990) documented, managers are motivated to enhance reported earnings so that they keep their jobs and reduce intervention by the firm’s board of directors. Furthermore, managers of financially-distressed companies which are close to a debt covenant violation get incentives to increase income in order to avoid or defer the costs of a violation (Watts and Zimmerman 1978, DeFond and Jiambalvo 1994).
Jaggi and Sun (2012) examined if monitoring earnings management by an independent audit committee differs for financially distressed and non-distressed firms. They found that the monitoring effectiveness of independent audit committee on earnings management is stronger for financially distressed firms. The reason is that managers are motivated to veil the company’s poor performance. This situation obstructs investors’ evaluation of firm’s future performance.
As the audit committee is a subcommittee of the board of directors, I wanted to test this theory. Thus, I formulated the third hypothesis:
H3: A highly independent board of directors constrains managerial behavior to employ
18
3. Research methodology
This chapter presents the research methodology applied in my thesis. Section 3.1 describes the data sample selection.
In the following sections, Section 3.2, Section 3.3 and Section 3.4, I present and discuss the variables and the models used to measure real earnings management, characteristics of board of directors and financial distress.
In Section 3.5 the control variables are described, while Section 3.6 is dedicated to the presentation and depiction of the empirical model.
The last part of the chapter, Section 3.7 is designated to data analysis, describing the procedures used in interpreting the models’ results.
3.1. Sample
My sample consists in 261 US companies for which I collected data for a period of 5 years, between 2008 and 2012, using online databases: Compustat, Audit Analytics and Risk Metrics.
The time frame of my research fits the post-SOX period, after 2002. It is appropriate because this period marks a shift in preferences for earnings management, from accrual-based to real earnings management. Also, the time frame overlaps the financial crisis which highlights an important moment in the global economy that is still affecting us.
The collected data is divided into three categories. The first one regards the financial data from the annual financial statements which I used for determining the real earnings management and financial distress variables. It was obtained using Compustat database. The second one is related to the auditors that audited the annual financial statements of the companies contained in the sample, information that was gathered from Audit Analytics database. The third category includes information about the companies’ board of directors which I used for establishing the board variables. This information was collected from Risk Metrics database.
The companies included in the sample of the research are non-financial firms, part of three industries: information technology, pharmaceuticals and industrials (focused on construction and engineering, electrical equipment and machinery sectors). I chose these three
19
industries because they provide relevant information for real earnings management measures, especially for research and development expenses.
The sample dimension represents the result of merging the data collected from Compustat, Risk Metrics and Audit Analytics, for the period 2008-2012. Table 1 summarizes the sample selection procedure.
Table 1. Sample selection procedure
Initial sample of firm-year observations, US natives, that are classified as Pharmaceuticals, Information Technology and Industrial (construction and engineering, electrical equipment, machinery) in Compustat
9,205
From which firms with incomplete data between 2008 and 2012 7,900
Number of firms tested 261
Number of years tested 5
Final sample (firm-year observations) 1,305
Starting with a number of 2,550 US companies from the three industries, I excluded all the firms that did not have information for the whole 5 year time frame. I made this elimination in order to have a full comparability of the companies. Based on the data collected from Compustat, I reduced the sample because not all the companies had financial information for the entire time frame of the research. Then I took into consideration the data collected from Risk Metrics and I noticed that a number of firms did not have complete information about the board of directors for the whole period between 2008 and 2012.
Overall, the total number of firms used to execute the empirical analysis was reduced to 261, containing 1,305 total observations which are structured as follows: 11.88% pharmaceutical firms, 66.67% information technology firms and 21.45% industrial firms (Table 2).
20
Table 2. Sample distribution by Industry
Industry Firm-years %
Pharmaceuticals 155 11.88%
Information Technology 870 66.67%
Industrial (construction and engineering,
electrical equipment, machinery) 280 21.45%
Total 1,305 100%
3.2. Real earnings management metrics
As presented by Cohen and Zarowin (2010), there are two ways for managing earnings. The first one is done by means of discretionary accruals, accrual-based earnings management, and the second one, which is part of this research, is done by means of real economic decisions, real earnings management.
As shown in prior studies (Roychowdhury 2006, Zang 2012, Gunny 2005), real activities manipulation can be measured considering three metrics: the abnormal level of cash-flow from operations, the abnormal level of production costs and the abnormal level of discretionary expenses. These variables are influenced by three methods of manipulation:
1. Acceleration of the timing of sales through increased price discounts or more lenient
credit terms. – Offering discounts and lenient credit terms to customers helps to increase
the sales volume of companies, but only on a short-term perspective. Once the companies return to old prices, this advantage disappears. Both discounts and lenient credit terms negatively affect the current cash-flows.
2. Reporting of lower cost of goods sold through increased production. – Overproduction is one way that managers can use to increase earnings, which means that firms produce more than is necessary. By increasing the production level, managers can allocate the fixed overhead costs over more products, leading to a decrease in fixed costs per unit. If the reduction in fixed costs per unit is not offset by any increase in marginal cost per unit, total cost per unit declines. Thus, the cost of goods sold is lowered.
3. Decreases in discretionary expenses including advertising, R&D and SG&A expenses. – By reducing these expenses, earnings recorded in the current period increase. Decreasing
21
these expenses also has an impact on the current period cash-flows, by means of increasing them, but, at the same time, they carry the risk of reducing future cash-flows if the company pays in cash for this type of expenses.3
For determining the three variables, abnormal cash-flow from operations, abnormal production costs and abnormal discretionary expenses, I first ran the following regressions to determine the normal levels of cash-flow from operations, production costs and discretionary expenses. The models were developed by Dechow, Kothari and Watts (Dechow et al. 1998).
(1)
Cash-flow from operations is represented by a linear function of sales and change in sales, where:
CFO = cash-flow from operations; SALES = annual sales revenues; Assets = total assets;
ε it = random error term.
(2) Production costs are represented by a linear function of sales, change in sales and change in sales occurred in the previous period, where:
PROD =sum of cost of goods sold (COGS) and change in inventory during a year; Assets = total assets;
SALES = annual sales revenues; ∆SALES = change in sales; ε it = random error term.
(3)
3 Cohen, D. A., Zarowin, P. 2010. “Accrual-based and real earnings management activities around seasoned equity
22
DISX = discretionary expenses; Assets = total assets;
SALES = annual sales revenues; ε it = random error term.
Discretionary expenses are modeled as a function of lagged sales to avoid the situation when the regression can be influenced by management decisions to increase sales in order to report higher earnings in a period, resulting in significantly lower residuals, as in the below model.
To determine the abnormal level of cash-flow from operations, production costs and discretionary expenses, I deducted from the actual cash-flow from operations, production costs, and respectively discretionary expenses, the normal level of the variables - calculated by using the estimated coefficients from (1), (2) and (3).
Firms that manage earnings by increasing them are more likely to present one or all of the following characteristics: unusually low cash flow from operations, and/or unusually low discretionary expenses, and/or unusually high production costs (Cohen et al. 2010).
3.3. Board metrics
As an internal governance measure, board of directors’ metrics is significant from the director independence perspective. The great importance of the board of directors as a corporate governance mechanism is reflected in prior studies (Fama and Jensen 1983). The strength of the board is related to its independence, which is connected with the firm’s performance, as shown by Baysinger and Butler (1985) and Klein (1998).
23
Variables Description
BOD_INDEP The proportion of independent directors on the
board of directors
BSIZE Total number of board members
SEP_CHAIR Dummy variable which takes the values 1 if
the CEO is the board chairman and 0 otherwise I included all these variables as a composite score in an index, B-Index, following the model used by Baber et al. (2012) in their research. I did not use all the variables as the authors do in their model, like AUD_INDEP (fraction of independent directors on the audit committee), COM_INDEP (fraction of independent directors on the compensation committee) and NOM_INDEP (fraction of independent directors on the nominating committee) because my research is only focused on the board of directors measures and not on the whole internal governance measures.
The composite index score, B-Index, increases by one if:
Greater than 2/3 of the board is composed of independent directors; The CEO is not the Board Chairman; or
The board size is less than the median of the distribution for all firms (adjusted for firm size and time).4
Table 3 presents the distribution of B-Index by Fiscal Year.
Table 3. Distribution of B-Index by Fiscal Year B-Index 2008 2009 2010 2011 2012 0 6 4 6 4 4 1 92 99 91 69 69 2 118 111 111 128 129 3 45 47 53 60 59 Total 261 261 261 261 261
Considering this, the model resulted is presented as follows.
24
B-Index = µ1 + µ2BOD_INDEP + µ3BSIZE + µ4SEP_CHAIR 3.4. Financial distress metrics
In order to assess the financial distress experienced by companies, I used the Z-Score Model, as Altman (1968) did in his paper. The purpose of this model is to examine the indicators that present and predict corporate distress.
Z = 0.012X1 + 0.014X2 + 0.033X3 + 0.006X4 +0.999X5 where X1 = working capital/total assets (WC/TA);
X2 = retained earnings/total assets (RE/TA);
X3 = earnings before interest and taxes/total assets (EBIT/TA); X4 = market value equity/book value of total liabilities (MVE/TL); X5 = sales/total assets (S/TA);
Z = overall index.
The indicators represent financial ratios frequently used in studies of corporate problems. They are presented below.
Variables Description
WC/TA Measures the net liquid assets of the firm
relative to the total capitalization
RE/TA Measures the amount of reinvested earnings
and/or losses of a firm over its entire life
EBIT/TA Measures the productivity of the firm’s assets,
independent of any tax or leverage factors
MVE/TL Measures how much the firm’s assets can
decline in value before the liabilities exceed the assets and the firm becomes insolvent
S/TA Illustrates the sales generating ability of the
firm’s assets; measures management capacity to deal with competitive conditions
25
I used the model according to Altman’s clarifications. Variables X1 to X4 are calculated in relative values (percentages) and variable X5 is calculated in an absolute value.
3.5. Control variables
For my empirical model I used five control variables. Firstly, I applied a dummy control variable for firms that have an auditor from one of the Big 4 audit firms (AUDITOR). Chi, Lisic and Pevzner (2011) have shown in their study that, while higher audit quality auditors reduce the level of accrual earnings management, real earnings management is positively associated with higher audit quality. The authors explain the existence of this possibility, when firms have strong incentives to manage earnings.
Secondly, I used LEVERAGE, calculated as total long-term debt divided by total assets, to show the amount of debt used to finance a firm’s assets. A firm with significant more debt than equity is pointed out to be highly leveraged, which also shows potential conflict of interests between stakeholders.
Thirdly, I comprised ROA as a control variable in the model, because it is an indicator of firm’s profitability relative to its assets. As Cohen and Zarowin (2010) used the variable in their research, I calculated the indicator by dividing the net income by total assets.
Fourthly, I used LOGSIZE, which is a measure of firm size. This control variable is calculated as natural log of total assets. As LaFond and Watts (2008) explain in their paper, large companies tend to have greater information asymmetry between their investors. This aspect affects the financial statements. Reducing information asymmetry leads to a decrease in managers’ incentives and their abilities to manipulate earnings.
The last variable is SHARES. I included this in the model because, as Cohen and Zarowin (2010) explained in their paper, the variable controls for the number of shares outstanding. A higher number of shares outstanding requires more earnings management activity to meet a given performance target.
26
Variables Description
AUDITOR Dummy variable which takes the values 1 if a
Big 4 audit firm performs the audit and 0 otherwise
LEVERAGE The amount of debt in the capital structure
ROA Return on assets and it is defined as income
before extraordinary items divided by beginning of period total assets
LOGSIZE Natural log of total assets
SHARES The number of shares outstanding
3.6. Empirical model
I established the following empirical model:
REM = a0 + α1BOD_INDEP + α2BSIZE + α3SEP_CHAIR + α4WC/TA + α5RE/TA + α6EBIT/TA + α7MVE/TL + α8S/TA + α9AUDITOR + α10LEVERAGE + α11ROA + α12LOGSIZE + α13SHARES+ ε it
which leads to
REM = a0 + α1B-Index + α2Z-Index + α3AUDITOR + α4LEVERAGE + α5ROA + α6LOGSIZE + α7SHARES + ε it
REM = one of the three measures for real earnings management;
B-Index = composite score for measuring board characteristics (BOD_INDEP + BSIZE + SEP_CHAIR);
Z-Index = overall index for measuring financial distress (WC/TA + RE/TA + EBIT/TA + MVE/TL + S/TA);
27
LEVERAGE = total long-term debt divided by total assets; ROA = net income divided by total assets;
LOGSIZE = natural log of total assets; SHARES = number of shares outstanding; ε it = random error term.
The empirical model is adapted for each hypothesis in order to test their accuracy. Therefore, the models are presented as follows:
H1: A high level of board independence will reduce managers’ attempts to engage in real earnings management.
REM = a0 + α1B-Index + α2AUDITOR + α3LEVERAGE + α4ROA + α5LOGSIZE + α6SHARES + ε it
H2: Firms that experience financial distress are more likely to apply real transaction based earnings management in order to boost earnings.
REM = a0 + α1Z-Index + α2AUDITOR + α3LEVERAGE + α4ROA + α5LOGSIZE + α6SHARES + ε it
H3: A highly independent board of directors constrains managerial behavior to employ real earnings management when the firms are in financial distress.
REM = a0 + α1B-Index + α2Z-Index + α3AUDITOR + α4LEVERAGE + α5ROA + α6LOGSIZE + α7SHARES + ε it
3.7. Data analysis
The purpose of this section is to explain the procedure used for the interpretations of the empirical results.
After collecting and merging all data gathered from three online databases (Compustat, Risk Metrics and Audit Analytics), I imported it in the statistical software, Stata. I used this program to execute the regression models in order to test the hypotheses.
28
As I mentioned before, for real earnings management measures, I used the abnormal level of cash-flow from operations, abnormal production costs and abnormal discretionary expenses. Utilizing models (1), (2) and (3), I determined the coefficients used to estimate the normal level of the variables. Then, I deducted the normal level of the variables from the actual variables.
Board metrics, B-Index, is a composite score index, comprising of three board characteristics (size, fraction of independent directors and chairman position related to CEO position), while financial distress metrics, Z-Index, was determined by the discriminant function detailed in the previous sections.
The control variables’ values were collected directly from the online databases, without any processing.
For the result interpretation of every regression, I analyzed the values of R2 and the p-values and coefficients of the independent variables. In addition, I examined the descriptive statistics and the univariate analysis (Spearman’s Rank correlation coefficient).
29
4. Empirical results
Section 4.1 is designated to descriptive statistics and univariate analysis.
Section 4.2 presents the empirical results for the three real earnings management regressions. I analyzed the response of cash-flow from operations, production costs and discretionary expenses to sales and change in sales from current and previous periods.
Section 4.3 shows the empirical results for the models that test the hypotheses, examining the response of real earnings management, first to the board of directors’ characteristics, second to financial distress indicators and third to the board of directors and financial distress characteristics.
4.1. Descriptive statistics and univariate analysis
4.1.1. Descriptive statistics
Table 16 presents descriptive statistics on the variables used in models testing the hypotheses. 95% of companies included in the sample are audited by Big4 auditors, 4% are profitable. The average size of the companies represented by LOGSIZE is 3.31 in terms of total assets.
It can be observed that most of the standard deviations are small, suggesting that the values of the observations included in the sample are concentrated around the mean.
30
Table 16. Descriptive statistics
Variable Obs Mean Std. Dev. Min Max
Dependent variables r_cfo 1305 0.029725 0.113281 -1.495635 0.345823 r_prod 1305 -0.080413 0.213382 -0.698057 0.964427 r_disx 1305 0.049413 0.153841 -0.654738 0.787469 Independent variables bindex 1305 1.843678 0.757433 0 3 zindex 1305 4.794035 6.936505 -55.40118 95.08465 Control variables auditor 1305 0.945594 0.226904 0 1 leverage 1305 0.123623 0.143284 0 1.871441 roa 1305 0.038844 0.140952 -1.54071 0.360621 logsize 1305 3.311304 0.648671 1.75923 5.328276 shares 1305 330.2804 963.1918 10.852 9151 Variable definitions: r_cfo = abnormal level of cash-flow from operations;
r_prod = abnormal level of production costs;
r_disx = abnormal level of discretionary expenses;
bindex = composite index for board characteristics (size, fraction of independent directors, Chairman position);
zindex = financial distress measure;
auditor = dummy variable that takes value 1 for Big4 auditors and 0 otherwise;
leverage = amount of debt in capital structure;
roa = profitability indicator calculated by dividing net income to total assets;
logsize = measure of firm size; natural log of total assets;
shares = number of shares outstanding.
4.1.2. Univariate analysis
Spearman’s Rank correlation coefficient was used to test the strength of the relationships between all variables used in each of the models that test the hypotheses. The results are presented in Tables 17, 18 and 19.
Table 17 shows that the abnormal level of cash-flow from operations is significantly negatively correlated (p<0.05) with B-Index and LEVERAGE. This suggests that the board of
31
directors’ characteristics and the level of leverage recorded for companies determine reductions in the abnormal level of cash-flow from operations. Also, significantly negatively correlations are registered between B-Index and Z-Index, AUDITOR, LEVERAGE, ROA, LOGSIZE, SHARES and between Z-Index and AUDITOR, LEVERAGE, LOGSIZE, SHARES. The same correlation appears between LEVERAGE and ROA, SHARES.
Table 17. Correlation between the variables of abnormal cash-flow from operations model
r_cfo bindex zindex auditor leverage roa logsize shares
r_cfo 1 bindex -0.0169* 1 zindex 0.2244 -0.0096* 1 auditor 0.0509 -0.0674* -0.0447* 1 leverage -0.1013* -0.0796* -0.2776* 0.0993 1 roa 0.4244 -0.0733* 0.249 0.0796 -0.1161* 1 logsize 0.1546 -0.3297* -0.123* 0.2607 0.2045 0.2084 1 shares 0.1965 -0.1733* -0.0299* 0.0729 -0.0038* 0.1091 0.5693 1 * denotes a 5% significance level
Table 18 shows that the abnormal level of production costs is significantly positively correlated (p<0.05) with B-Index and significantly negatively correlated with Z-Index, AUDITOR, LEVERAGE, ROA, LOGSIZE, SHARES. The positive correlations indicate an increase in the abnormal level of production costs and the negative correlations indicate decreases recorded for the variable. Board metrics (B-Index) is significantly negatively correlated with financial distress metrics (Z-Index), AUDITOR, LEVERAGE, ROA, LOGSIZE and SHARES. Z-Index is significantly negatively correlated with AUDITOR, LEVERAGE, LOGSIZE and SHARES. LEVERAGE has the same correlation with ROA and SHARES.
32
Table 18. Correlation between the variables of abnormal production costs model
r_prod bindex zindex auditor leverage roa logsize shares
r_prod 1 bindex 0.0361* 1 zindex -0.1035* -0.0096* 1 auditor -0.1134* -0.0674* -0.0447* 1 leverage -0.074* -0.0796* -0.2776* 0.0993 1 roa -0.2578* -0.0733* 0.249 0.0796 -0.1161* 1 logsize -0.207* -0.3297* -0.123* 0.2607 0.2045 0.2084 1 shares -0.2234* -0.1733* -0.0299* 0.0729 -0.0038* 0.1091 0.5693 1 * denotes a 5% significance level
Table 19 shows that the abnormal level of discretionary expenses is significantly positively correlated (p<0.05) with B-Index and ROA and significantly negatively correlated with Z-Index. This means that the characteristics of board of directors do not reduce the abnormal level of discretionary expenses including advertising, R&D and SG&A expenses, while financial distress metrics induce a decrease for the variable.
B-Index is significantly negatively correlated with the other independent variable (Z-Index) and all the control variables. Financial distress metrics are significantly negatively correlated with AUDITOR, LEVERAGE, LOGSIZE and SHARES. LEVERAGE has the same correlation with ROA and SHARES.
Table 19. Correlation between the variables of abnormal discretionary expenses model
r_disx bindex zindex auditor leverage roa logsize shares
r_disx 1 bindex 0.0186* 1 zindex -0.0669* -0.0096* 1 auditor 0.1316 -0.0674* -0.0447* 1 leverage 0.13 -0.0796* -0.2776* 0.0993 1 roa 0.0354* -0.0733* 0.249 0.0796 -0.1161* 1 logsize 0.1901 -0.3297* -0.123* 0.2607 0.2045 0.2084 1 shares 0.1752 -0.1733* -0.0299* 0.0729 -0.0038* 0.1091 0.5693 1 * denotes a 5% significance level
33
4.2. Real earnings management models
This section examines the response of real earnings management to the influence of sales and changes in sales from current and previous periods.
As I mentioned in the previous chapter, in order to calculate the abnormal level of cash-flow from operations, production costs and discretionary expenses, first I determined the coefficients used to estimate the normal level of the variables, based on the models developed by Dechow (1998). Then, I deducted the normal level of the variables from the actual variables. The results are presented in the following Table 4, Table 5 and Table 6.
Table 4. Results for abnormal level of cash-flow from operations regression
Panel A
Scaled
CFO Coef. Std. Err. t P>t [95% Conf. Interval] Scaled Intercept -1.921181 1.841347 -1.04 0.297 -5.533514 1.691151 Scaled Sales 0.0897831 0.0034695 25.88 0.631 0.0829767 0.0965894 Scaled Change in Sales 0.0568242 0.0177092 3.21 0.001 0.0220825 0.915658 Panel B Number of obs = 1305 F( 3, 1302) = 348.92 Prob > F = 0.0000 R-squared = 0.4457 Adj R-squared = 0.4444 Root MSE = 0.11721
For the first regression, that presents the abnormal level of cash-flow from operations, R2 has the value of 0.4457. The significance of this indicator varies from 0 (insignificant) to 1 (significant). The value is established approximately at the middle of the interval, as presented in
34
Panel B, which means that the independent variables of the regression do not have a significant influence on the dependent variable.
According to model (1) presented in the research methodology chapter, sales and change in sales do not significantly influence real earnings management. This means that accelerating the timing of sales through increased price discounts or more lenient credit terms, affecting the current cash-flows, do not have a major boost on earnings in the current period.
The p-value (P>t) for each independent variable tests if the changes in the dependent variable’s value are related to changes in the independent variables. The lower the p-value is (compared to 0.05), the more significant the result is. Analyzing the p-values for the three independent variables presented in Panel A, it is shown that only change in sales influence the abnormal level of cash-flow from operations.
By examining the coefficients of each independent variable (Panel A), it can be observed the positive or negative effect on the dependent variable and how much it increases or decreases because of these influences. Sales and change in sales lead to an insignificant increase in abnormal cash-flow from operations.
Overall, my conclusion for this model is that even though the sales and changes in sales recorded in the current period have a positive impact on the abnormal cash-flow from operations by increasing it, the influence is too small to be significant. In addition, only one variable, change in sales, is related to increased cash-flows. Therefore, managers that accelerate the timing of sales by offering discounts or more lenient credit terms to customers do not increase earnings in a short-term horizon.
35
Table 5. Results for abnormal production costs regression
Panel A Scaled
PROD Coef. Std. Err. t P>t [95% Conf. Interval] Scaled Intercept -47.79275 3.592962 -13.3 0.000 -54.84138 -40.74412 Scaled Sales 0.7564276 0.0071339 106.03 0.000 0.7424324 0.770423 Scaled Change in Sales -0.2087954 0.0346695 -6.02 0.000 -0.27681 -0.140781 Scaled Previous Change in Sales -0.0449534 0.0340723 -1.32 0.187 -0.111796 0.021889
For the second regression, which presents the abnormal level of production costs, R2 has the value of 0.9250, as presented in Panel B, which places it near the maxim value of the significance interval [0, 1]. This means that the independent variables significantly influence the dependent variable.
According to model (2), represented in the research methodology chapter, sales and change in sales from the current period, along with change in sales from previous period, significantly influence the abnormal level of production costs.
The analyze of p-values from Panel A shows that the first three independent variables, that record value 0, represent influences for the dependent variable, while the last independent
Panel B Number of obs = 1305 F( 4, 1301) = 4008.6 Prob > F = 0.0000 R-squared = 0.9250 Adj R-squared = 0.9247 Root MSE = 0.2283
36
variable has no influence due to its value of 0.187 related to the threshold of 0.05. Therefore, the abnormal level of production costs is influenced mostly by sales and change in sales from current period and change in sales from previous period.
In addition, most of the independent variables have a negative effect on the dependent variable, besides one which positively influences it.
Overall, the results of the regression show that managers engage in overproduction, in order to increase earnings. This leads to a lower value for the cost of goods sold because they allocate the fixed overhead costs over more units of product.
Table 6. Results for abnormal discretionary expenses regression
Panel A
Scaled
DISX Coef. Std. Err. t P>t [95% Conf. Interval] Scaled Intercept 53.98785 2.53554 21.29 0.000 49.01368 58.96203 Scaled Previous Sales 0.1435986 0.00472 30.43 0.000 0.134341 0.152856 Panel B Number of obs = 1305 F( 2, 1303) = 1283.81 Prob > F = 0.0000 R-squared = 0.6634 Adj R-squared = 0.6628 Root MSE = 0.16165
For the third regression, that presents the abnormal level of discretionary expense, R2 has the value of 0.6628 (Panel B), which places it above the middle in the significance interval. Along with the p-values (Panel A) that record values below 0.05, it shows that the independent variables influence the values of the dependent variable. Also, the positive coefficients show that the independent variables are associated with increases in the dependent variable.
According to model (3), represented in the previous chapter, sales recorded in previous period influence the level of abnormal discretionary expenses. This means that by decreasing
37
expenses including advertising, R&D and SG&A, earnings recorded in the current period will increase.
In accordance with the analysis of the statistical data from the three models that describe the abnormal level of cash-flow from operations, production costs and discretionary expenses, it can be concluded that managers are more likely to engage in real earnings management by producing more than it is necessary in order to increase earnings. They do this because the cost of goods sold is reduced by allocating the fixed overhead costs over more units of product. Also, another preference for real activities manipulation is to decrease the discretionary expenses (advertising expenses, R&D, SG&A) in order to increase current period earnings. The regressions’ results provide evidence that managers are less likely to accelerate the timing of sales by offering increased price discounts or more lenient credit terms to customers.
4.3. The influence of board of directors and financial distress characteristics on real earnings management
Real earnings management is a topic that presents interest for many researchers. This type of earnings manipulation is based on economic decisions taken by managers, who try to reach performance targets by boosting earnings.
As it can be seen in the previous parts of the thesis, real earnings management is determined using three metrics: abnormal level of cash-flow from operations, abnormal level of production costs and abnormal level of discretionary expenses. It represents the dependent variable for each model used for testing the three hypotheses.
In the following part, I will examine the results of the regressions and establish if the hypotheses are supported or rejected.
4.3.1. Results of the association between real earnings management and board of directors metrics
The first hypothesis states that a high level of board independence will reduce managers’ attempts to engage in real earnings management, by mean of:
accelerating the timing of sales through increased price discounts or more lenient terms; reporting of lower cost of goods sold through increased production; or
38
In addition, I want to examine how the results are influenced by the firm’s auditor, if it is a Big 4 company, and by the values of leverage, ROA and shares. Also, another control is the firm size.
Table 7. Abnormal level of cash-flow from operations and board metrics
Panel A
r_cfo Coef. Std. Err. t P>t [95% Conf. Interval] bindex 0.005489 0.0039168 1.40 0.161 -0.002195 0.013173 auditor 0.0075398 0.0128553 0.59 0.558 -0.0176796 0.0327592 leverage -0.0415382 0.0204999 -2.03 0.043 -0.0817548 -0.0013217 roa 0.3235756 0.0206176 15.69 0.000 0.2831282 0.364023 logsize -0.0000359 0.0059342 -0.01 0.995 -0.0116775 0.0116056 shares 0.0000186 3.59E-06 5.17 0.000 0.0000115 0.0000256 _cons -0.0009686 0.0224627 -0.04 0.966 -0.0450358 0.0430987 Panel B Number of obs = 1305 F( 6, 1298) = 56.58 Prob > F = 0.0000 R-squared = 0.2073 Adj R-squared = 0.2037 Root MSE = 0.10109
Table 7 presents the results of the regression, where real earnings management is expressed by abnormal level of cash-flow from operations. B-Index is the variable that measures the board of directors’ characteristics. It is a composite score index for board size, fraction of independent directors and chairman position.
As examined in this model, the value of R2 of 0.2073 shows that the independent variables do not significantly influence the dependent variable. Even though the fraction of independent directors is very high, this does not affect the managers’ attempts to manipulate earnings through sales decisions (offering increased price discounts or more lenient credit terms to customers).
AUDITOR is a dummy variable that takes value 1 when the company’s auditor is Big 4 and 0 otherwise. 95% of the financial statements of the companies included in the sample were
