Corporate Governance and Earnings Management in the Netherlands Doctoral Thesis

Doctoral Thesis

Corporate Governance and Earnings Management

in the Netherlands

Jel Codes: G30, C25

Key words: corporate governance, earnings management

Date: 7 August, 2007 M. Leuvenink

1198440

martijnleuvenink@gmail.com University of Groningen, Holland

Preface

In front is my doctoral thesis on corporate governance and earnings management in the Netherlands. The process of writing a doctoral thesis is by many students described as the cliché of an interactive process of ups and downs. Although it may be a cliché, I must admit it is true, however it stroke me as a positive experience. While writing this preface it occurs to me that my period of being a student almost has come to an end. All the things I have learned during these years at the RuG, and in particular the period while writing my thesis, I hope to be using again in my professional life.

I would like to thank mr. T.A. Marra for his supervision, advise and sharp recommendations during the forthcoming of this thesis. I would also like to thank mr. C.A. Huijgen for checking and reviewing my thesis. Based on the recommendations of both supervisors I have a clear view on my thesis.

I would like to thank my parents for their years of support. Furthermore, I would like to thank my girlfriend and my friends for all their support, pleasure and experiences during my life as a student. I would like to wish the reader of this thesis the same amount of pleasure that it has given me while writing.

Executive Summary

In recent years we have witnessed a number of accounting scandals in the United States and in Europe, like the fall of Enron and the debacle concerning Ahold. These actions resulted in a decline of confidence in the supervisory board and the board of directors of companies. In the Netherlands this made way for the Committee on Corporate Governance, which formulated the Dutch Corporate Governance code, also called the Tabaksblat Code.

Under Dutch GAAP and since 2005 under IFRS, managers have a certain degree of freedom with respect to the recognition of receivables and payables. Accrual accounting allows managers to adjust cash flows in order to better reflect the performance of the firm. When managers manipulate earnings it can be qualified as earnings management.

The purpose of this study is to explore the relationship between internal governance mechanisms and earnings management by firms in the Netherlands. In order to explore this relation, I first calculated the discretionary short-term working capital accruals for every listed firm in the Netherlands. Secondly, I established a link between these accruals and several internal governance variables, by means of regression analysis.

With respect to the estimation of the discretionary short-term working capital accruals, the aggregate accrual models by Dechow and Dichev (2002), McNichols (2002) and Cheng et al. (2005) were adapted. The models are adapted since only one year, 2004, is available to test the relation between earnings management and internal governance mechanisms. Since only one year was available for testing the discretionary short-term working capital accruals needed to encompass a short-term orientation. For this end the accrual was corrected for depreciation and amortization items. With respect to the regression analysis of the three models, firm specific, industry specific and pooled regressions were used. The only statistically significant findings were stemming from the pooled regression analysis. The model by McNichols (2002) documented the highest adjusted R2 _{(0.375) and}

is therefore considered as the better model to estimate the discretionary short-term working capital accruals.

statistically insignificant results (F-value 0.476, significance level 0.907). Further analysis made use of the backward function of SPSS, in which all variables are entered into the equation and then sequentially removed. This analysis showed that only the size of the audit committee (ACSIZE) variable was not eliminated. Based on the size of the audit committee (ACSIZE) variable, the model showed an F-value of 3.8310 with a significance level of 0.0607. Other analysis focused on different dependent variables, like the discretionary short-term working capital accruals based on the model by Dechow and Dichev (2002) and Cheng et al. (2005), and on total accruals. The models did not document any statistically significant findings. Based on the insignificance of the models, the null hypothesis is accepted, indicating that all regression coefficients, except the constant, are zero. Since the models displayed no significant findings the hypotheses in Chapter 4 could not be accepted. This indicates that there is no relationship between internal governance mechanisms and earnings management for the Netherlands in the year 2004.

Additional analyses focusing on fewer independent variables decomposed the original model in two models. Model 1 focused on the supervisory board and model 2 focused on the audit committee. Furthermore, only one control variable is considered, the financial leverage (LEV). Model one displayed insignificant coefficients and an insignificant significance level on the F-value. The second model focused on the audit committee, whereas the findings displayed an insignificant model. This model was further decomposed into 2 variants, where the one that focused on the number of meetings of the audit committee (ACMEET) and the financial leverage (LEV) displayed insignificant statistics. The second variant that focused on the size of the audit committee (ACSIZE) and the financial leverage (LEV) also presented a statistically insignificant model (significance level of the F-value of 0.060). The model has an explanatory power of 9.00%. Only the coefficient on the size of the audit committee (ACSIZE) proved significant. The hypothesis that states that earnings management is associated with the size of the audit committee can however not be accepted, because the total model is not significant. The coefficient of the size of the audit committee (ACSIZE) is however positive (0.034), which indicates a positive relation between the size of the audit committee and earnings management. The hypothesis documented no direction on the relation, since there is no consensus on the direction of the relationship in the literature.

2006, p.923). The SEC (1999) mandates that the audit committee consists of a minimum of four directors. Since this study found a positive coefficient on the size of the audit committee it could be concluded that there is to some extent a relation between a minimum and a maximum number of members of the audit committee. A possible recommendation for future research is to investigate this relation more thoroughly.

Table of Contents

Chapter 1 Introduction _________________________________________________- 7 - Chapter 2 Corporate Governance _______________________________________- 10 - 2.1 Corporate Governance in the Netherlands________________________________ - 10 - 2.2 The supervisory board_________________________________________________ - 12 - Chapter 3 Earnings Management Models _________________________________- 14 - 3.1 Aggregate Accruals Models ____________________________________________ - 14 - 3.1.1 Model by Dechow and Dichev (2002) __________________________________________ 15 -3.1.2 Model by McNichols (2002)__________________________________________________ 19 -3.1.3 Model by Cheng, Peng and Thomas (2005) ______________________________________ 22 -3.2 Discretionary and nondiscretionary accruals______________________________ - 24 -

3.2.1 Short term and long term discretionary accruals__________________________________ 24 -3.2.2 Shortterm discretionary accrual models________________________________________ 25 -Chapter 4 Corporate governance and earnings management _________________- 28 -

4.1 Supervisory board ____________________________________________________ - 28 - 4.2 Audit committee______________________________________________________ - 30 - Chapter 5 Data and methodology________________________________________- 33 - 5.1 Earnings Management_________________________________________________ - 33 - 5.1.1 Data sample ______________________________________________________________ 33 -5.1.2 Descriptive statistics ________________________________________________________ 35 -5.1.3 Regression analyses ________________________________________________________ 36 -5.1.4 Shortterm discretionary working capital accruals_________________________________ 38 -5.2 Corporate Governance Variables _______________________________________ - 39 -

Chapter 1 Introduction

In recent years we have witnessed a number of accounting scandals at companies in the United States as well as in Europe. These accounting scandals resulted in a serious decline of confidence in the boards of directors and the boards of supervisors of corporations. In the Netherlands the Ahold debacle and the discussion on the compensation of CEO’s made way for the Committee on Corporate Governance which documented the Dutch Corporate Governance Code, also called the Tabaksblat Code. This committee was launched in March 2003 and took effect on January 1st _{2004. Among other}

things, Dutch companies are obliged by the Dutch Corporate Governance Code to disclose information regarding the composition of the board of directors, the audit committee and the remuneration committee. In this research the focus will be on the internal governance structure, which entails several variables concerning the board of directors and the audit committee.

The separation of ownership and control in corporations led to the agency problem (Berle and Means, 1932; Jensen and Meckling, 1976; Fama and Jensen, 1983a,b). The agency problem stems from the conflicting interests of the principal (financier/owner) and the agent (manager). The manager raises funds from different investors in order to make investments or to cash out his holdings in the firm (Shleifer and Vishny, 1997). The financiers on the other hand need the manager because of his special management capacities to generate returns on their funds. This raises the question of how the financiers, once he or she has invested, can make a competitive rate of return without the deception of receiving a worthless piece of paper back from the manager? The standard solution is a complete contract, between the principal and the agent, which specifies exactly what the manager does in all states of the world, and how the profits are allocated (Shleifer and Vishny, 1997). In practice it is however almost impossible to draw such a contract. The principal and the agent therefore need to allocate the residual control rights, which are defined as the rights to make decisions in circumstances not fully foreseen by the contract (Grossman and Hart, 1986; Hart and Moore, 1990). It is therefore possible to draw a contract that specifies that the manager receives funds from the financiers and that the financiers retain all of the residual control rights. Whenever something happens that is not documented in the contract, the financiers are allowed to decide what to do. However this is not the concept of the contract between the manager and the financiers, since the financiers have specifically hired the manager because of his specific capacities to solve problems like these. Therefore, most of the residual control rights are in the hands of the manager. The contract may specify limits to the discretion specified in the contract, but the fact is that managers do have most of the residual control rights (Shleifer and Vishny, 1997). Corporate governance deals with these limits.

the Dutch Generally Accepted Accounting Principles (GAAP). Under both systems a certain degree of freedom with respect to the recognition of receivables and payables is allowed. Accrual accounting allows managers to adjust cash flows in order to better reflect the performance of the firm. When managers have incentives to manipulate the accrual component of earnings to reach specific earnings targets, the prospect of exploitation increases (Dechow et al., 1996; Teoh et al., 1998b; Holland and Ramsay, 2003). When managers manipulate earnings it can be qualified as earnings management. Healy and Wallen (1999,p 368) define earnings management as:

Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to either mislead some stakeholders about the underlying economic performance of the company of to influence contractual outcomes that depend on reported accounting numbers.

Managers can have several incentives for manipulating the accrual component of earnings. The bonus plan hypothesis states that managers of firms with bonus plans are more likely to choose accounting methods that shift reported earnings from future periods to the current period (Healy, 1985; McNichols and Wilson, 1988; Gaver, 1995; Holthausen et al., 1995). Further, the use of accounting information is widely used in the valuation of stocks, which creates an incentive for managers to manipulate earnings in order to influence short-term stock price performance (Neill et al., 1995; Subramanyam, 1996; Teoh, 1998, 1998a, 1998b; Healy and Wallen, 1999). Also, Watts and Zimmerman (1986) documented the size hypothesis, which states that the larger the firm, the more likely the manager is to choose accounting procedures that defer reported earnings from current to future periods.

Prior research on fraudulent financial reporting and the role of governance mechanisms (Beasley, 1996; Dechow et al 1996; Jiambalvo 1996) found a negative relationship between effective governance mechanisms and financial accounting decisions that are in conflict with Generally Accepted Accounting Principles (GAAP). Current research focuses on the association between corporate governance mechanisms and earnings management. For example, numerous studies find significant relationships between board of director, audit committee characteristics and earnings management (Bédard et al., 2004; Chtourou, 2001; Klein, 2002a; Lin et al., 2006; Xie et al., 2003).

The purpose of this study is to explore the relationship between internal corporate governance mechanism and earnings management by firms in the Netherlands. In order to explore this relation, I first calculated discretionary short-term working capital accruals for every listed firm in the Netherlands. Secondly, I established a link between these accruals and several internal governance variables, by means of regression analysis.

I focused on firms that are listed in the Netherlands. Peasnell et al. (2005) focus on UK data, Davidson et al. (2005) focus on Australian data and Rahman & Ali (2006) focus on Malaysian data. De Jong et al. (2000) focus on the relation between corporate governance and firm value in the Netherlands. This study is therefore the first to investigate the relation between internal governance mechanisms and earnings management in the Netherlands. With respect to the corporate governance data, the Netherlands Board Index 2004 by Spencer Stuart is used. With respect to earnings management I analyze three models, namely the model proposed by Dechow and Dichev (2002), the model by McNichols (2002) and the model by Cheng et al. (2005).

Chapter 2

Corporate Governance

This chapter will elaborate on the Dutch corporate structure and the roadmap to the establishment of the Code Tabaksblat in section 2.1. Section 2.2 will focus in more detail on the supervisory board and the committees of the non-executive directors.

2.1 Corporate Governance in the Netherlands

In 1965 the Verdam committee was established with the task to reform the Dutch company law. The Verdam committee referred to the situation as “no longer acceptable”; inadequate control of management’s activities led to their propensity to misstate the firm’s financial position and to violate the position of shareholder, debtholders and employees (Verdam Committee, 1965, p. 119-125). The committee published their draft version in 1968 which took effect in 1971. The underlying problem concerned a monitoring problem within large companies and a desire to increase management’s accountability to the company’s broader set of “stakeholders” (i.e., investors, employee’s and the general public) (De Jong et al., 2000). The new proposed law addressed these problems by altering the legal form of large companies by means of the creation of the “Structured Regime” (Structuur regeling).

Within Dutch firms, a two-tier board prevails. A two-tier board consists of the top management team/board (“Raad van Bestuur”) and a supervisory board (“Raad van Commissarissen”). There exists a formal separation between the management board and the supervisory board. The supervisory board is through the formal separation independent of the company and therefore responsible for the supervision of the policy that is developed by the management board. The supervisory board is in principle in place to supervise the general course of business of the company and consists therefore generally of outsiders. The management board controls the day-to-day operations and consists of the company’s management team, whereas the chairman of the management board is the most prominent director and Chief Executive Officer (CEO). The CEO is not involved with the supervisory board, which would cause a CEO-duality.

The first variation is the Fully Structured Regime which requires a supervisory board that takes over the following tasks from the shareholders: establishment (and by default the approval of) the annual accounts, election of the management board and election of the supervisory board itself (called co-optation), whereas the supervisory board has authority over major decisions made by the management. Top management (and especially the CEO) in practice substantially influences the composition of the supervisory board (Van der Goot and Van het Kaar, 1997). In the fully structural regime, the supervisory board has potentially substantial power. Firms who are not obliged to adopt the structural regime may choose to voluntary adopt this regime (Postma and Van Ees, 2000). The second variation is the mitigated structural regime and is required if more than 50% of the shares of a Dutch subsidiary are in hands of a foreign company. This structure requires the establishment of a supervisory board, whereas these members elect new members for their board (co-optation). The difference with the fully structural regime is that the approval of the annual accounts remains in the hands of the shareholders. Another difference is the fact that the supervisory board is not allowed to hire or fire members of the management board. The third variation is the common regime and applies to the remaining limited liability companies. These companies are free to voluntarily install a supervisory board, however the board has less power since the shareholders meeting take up the control of the annual statement of accounts and the composition of the management board and supervisory board (Postma and Van Ees, 2000). There is however, also a fourth category called the exempted regime. Firms under this category are exempted from the structural or exempted regime when they are dependent on a firm that is already subject to one of these regimes, when they are part of a foreign company, or when they are a mere service organization for affiliated corporations (Postma and Van Ees, 2000).

The Peters Committee presented in 1997 its final recommendations. The committee focused in its recommendations on the relationships between shareholders, management board and supervisory board in order to arrive at more transparency and accountability of Dutch firms (Postma and Van Ees, 2000). One of the recommendations concerned the reinforcement of the independency of the supervisory board. During the establishment of the code it was agreed upon a one-time monitoring report.

spending only little time and effort on the functioning of the supervisory board and were publishing rather limited amounts of information concerning corporate governance in their annual reports. It appeared that a number of companies were only complying with the code on a broad level, while at a more detailed level the compliance was very limited. These findings formed the basis in 2002 for Peters to recommend a new code.

In March 2003 a committee, under the direction of Morris Tabaksblat, was established to develop a code for listed companies and their shareholders. This code took into account the findings from De Jong and Roosenboom (2002). On December the 9th _{2003, the new code was published and}

took effect by January 1st _{2004. The code is based on the already existing legislation in the}

Netherlands concerning the mandatory application of the two-tier board system and on the case-law on corporate governance. The code is based on the principle, accepted in the Netherlands, that a company is a long-term form of collaboration between the various parties involved (Dutch Corporate Governance Code 2003, p.3, item 3). The most important conditions for this are integrity and transparency of decision-making by the management board, and proper supervision thereof, including accountability for such supervision (Hooghiemstra et al, 2004). The Dutch Corporate Governance Code is based on principles and best practice provisions. All the stakeholders and persons in the company should observe these principles and best practices in relation to each other. The principles can, according to the Dutch Corporate Governance Code (2003), be regarded as views that reflect the latest general views on good corporate governance, which now enjoy wide support. These principles are further elaborated in the specific best practice provisions. These provisions create a set of standards governing the conduct of management board and supervisory board members (also in relation to the external auditor) and shareholders (Dutch Corporate Governance Code 2003, p.4, item 5). They reflect the national and international ‘best practices’ and may be regarded as an elaboration of the general principles of good corporate governance (Dutch Corporate Governance Code 2003, p.4, item 5).

An important aspect of Dutch Corporate Governance Code is the fact that limited liability companies are allowed to deviate from the best practice provisions. Whenever a company deviates from the code on certain points it is obliged to make a statement in their annual reports on why they deviate from the code. This principle is called the ‘comply or explain principle’ and is in concordance with international ruling and codes, and is also embedded in the Dutch Civil Law (BW 2:391.4)

2.2 The supervisory board

(Dutch Corporate Governance Code, principle 1, p8). The Dutch Corporate Governance Code also states clearly the role and procedures of the supervisory board:

“The role of the supervisory board is to supervise the policies of the management board and the general affairs of the company and its affiliated enterprise, as well as to assist the management board by providing advice. In discharging its role, the supervisory board shall be guided by the interests of the company and its affiliated enterprise, and shall take into account the relevant interests of the company's stakeholders. The supervisory board is responsible for the quality of its own performance.” (Dutch Corporate Governance Code, 2003, p. 15)

Because of the supervisory role of the non-executive directors their independency is important. Independence with respect to non-executive directors is defined in the Dutch Corporate Governance Code as: “the composition of the supervisory board shall be such that the members are able to act critically and independently of one another and of the management board and any particular interests” (Dutch Corporate Governance Code, principle III, p. 17). Best practice provision III.2.1 states that all supervisory board members need to be independent, with a maximum of one member being not independent. The independency of supervisory board members is defined in best practice provision III.2.21_.

Principle III.5 of the Dutch Corporate Governance Code states that if the supervisory board consists of more than four members, it is obliged to form an audit committee, a remuneration committee and a selection and appointment committee. The function of these committees is to prepare the decision-making of the supervisory board (Dutch Corporate Governance Code, 2003). Best practice provision III.5.4 focuses on the supervisory tasks of the audit committee. The audit committee is expected to review the financial reporting process, as well as to facilitate a flow of information among the board of directors, the internal and external auditors, and management in order to effectively monitor the financial discretion of management (McMullen and Raghunandan, 1996).

Chapter 3 Earnings Management Models

As indicated in the introduction, there are three approaches for estimating earnings management: specific accruals, aggregate accruals and the distribution of reported earnings. In this study the aggregate accruals model is used. With respect to the estimation of the discretionary and non-discretionary accruals this method shows better results, since the approach that focuses on specific accruals has limitations with respect to the sample size and the generalizability of the results (Beneish, 2001). The approach on the distributions of earnings has as detrimental effect that it does not indicate the extent or nature of the earnings management according to Beneish (2001). Based on these findings, the choice for the aggregate accruals model is reconfirmed based on superiority in estimating discretionary accruals. Recent publications by Francis et al. (2005) and Ecker et al. (2006) focused on the market pricing of accruals quality and a return based representation of earnings quality. Both models investigate the relation between the debt and equity cost of capital and the quality of accruals. Both theories use the model by Dechow and Dichev (2002) and follow the adaptations suggested by McNichols (2002) by including the change in sales and the level of property, plant and equipment.

This chapter elaborates on the three different aggregate accrual models in section 3.1, whereas section 3.2 will discuss the discretionary and non-discretionary accruals. The proposed models in this section on short-term discretionary working capital accruals on the three models are innovative since it is not investigated before in a similar manner.

3.1 Aggregate Accruals Models

(2005) is a combination of the model by Dechow and Dichev (2002) and the Modified-Jones model focusing besides the past, present and future cash flow from operations on the level of property, plant and equipment and the change in revenues minus the change in receivables. The table below displays the different models and their formulas and clearly demonstrates the composition of the different models.

Model Formula Composed of

Jones Model (1991) ∆WC_t =b₀+b₁*∆sales_t +b₂*PPE_t +ε_t

Modified-Jones model ∆WC_t =b₀+b₁*

(

)

Dechow and Dichev

(2002) t t t t t CFO b CFO b CFO b b WC ε + + + + = ∆ + − 1 3 2 1 1 0 * * * McNichols (2002) t t t t t t t sales b PPE b CFO b CFO b CFO b b WC ε + ∆ + + + + + = ∆ − + * * * * * 5 4 1 3 2 1 1

0 Dechow and Dichev

(2002) + Jones (1991) Cheng et al. (2005)

(

)

0 Dechow and Dichev

(2002) + Modified-Jones model

Beside the focus on the specific variables that relate to the working capital accruals or total accruals is the article by Dechow and Sloan (1991) that documents an industry model, which relaxes the assumption that nondiscretionary accruals are constant over time. The industry model assumes that variation in the determinants of nondiscretionary accruals is common under firms in the same industry. The industry model uses cross sectional analysis which is later used by many others (Dechow et al., 1995; Dechow et al., 1996; Bartov et al., 2001; Klein, 2002a). However, due to data constraints this study will not focus on the industry model, which will be discussed in more detail in chapter 5.

3.1.1 Model by Dechow and Dichev (2002)

The model by Dechow and Dichev (2002) is based on the principle of accrual accounting, which allows companies to adjust for cash flow timing problems. These timing problems stem from the fact that the timing of the economic achievements and sacrifices of companies often differs from the timing of the related cash flows (Dechow and Dichev (2002). Dechow and Dichev (2002) state that earnings equal cash flows plus accruals and that cash flows for any period t (CFt) can be categorized into three

groups:

CFtt-1 Cash collection or payment of amounts accrued at t-1 (net)

CFtt Current cash flow (net)

Therefore the total cash flow in period t is: 1 1 + − _{+ +} = t t t t t t t CF CF CF CF (Eq. 1)

Dechow and Dichev (2002) stipulate that the timing problem concerning cash flows leads to the creation of an opening and a closing accrual when the recognition of a cash flow is adjusted. An opening accrual is established when either revenue or expense is recognized before the cash is received or paid, or when cash is received or paid before it is recognized in earnings, whereas a closing accrual is established when the other element of this pair had occurred and reverses the accrual portion of the original entry (Dechow and Dichev, 2002). Possible errors arise when a cash flow occurs after the corresponding revenues and expenses are recognized in earnings. Since the cash flow realizations differ from their accrual estimates, it is necessary to include estimation error terms in the accruals. With respect to the deferred cash flows that occur before the recognition, these accruals do not contain a measurement error.

Name Accrual Amount

Opening accrual for future collections and payments ACFt+1/tO CFt+1t + t+1t

Closing accrual for future collections and payments ACF/t-1C -CFtt-1 – tt-1

Opening accrual that defers the recognition of cash flows ACF/t+1O -CFtt+1

Closing accrual that defers the recognition of cash flows ACFt-1/tC CFt-1t

The table above states the notations for the opening and closing accruals, for future and deferred accruals and the estimation errors.

The notations in the table above together with equation 1 lead to the following equation: ) ( ) ( 1 1 1 1 1 1 1 1 t t t t t t t t t t t t t t t t t t t CF CF CF CF CF CF CF E − + − − − + + − + − − − + + + + = ε ε (Eq. 2) Where: Et = Earnings;

CFts= cash flow from operations realized at time t and recognized at time s;

ts = estimation error associated with accruals recognized at time t and cash flows realized at time s.

This can be rearranged as:

CFts= cash flow from operations realized at time t and recognized at time s;

ts = estimation error associated with accruals recognized at time t and cash flows realized at time s.

Equation three is the centre of the publication of Dechow and Dichev (2002), since they present earnings as a function of the past, present and future cash flows plus an adjustment for estimation errors and their correction. However, since the past, present and future cash flows are used in estimating the earnings, the timing problem of the accruals is alleviated because not the cash flow from a single year is considered but the cash flows of the years surrounding the year at time t. This improvement comes at the cost of using estimates which can be of detrimental effect on the quality of earnings, because of the inclusion of errors in estimates and their corrections.

Dechow and Dichev (2002) focus in their publication on the origination and reversal of working capital accruals. They specify the following expression for accruals at time t after rearranging the accrual components of equation 2:

1 1 1 1 1 1 ( + − ) + + − − − + + + − = t t t t t t t t t t t t t CF CF CF CF A ε ε (Eq. 4) Where:

At = current accrual recognized at time t.

CFts= cash flow from operations realized at time t and recognized at time s;

ts = estimation error associated with accruals recognized at time t and cash flows realized at time s.

This equation shows that accruals are temporary adjustments that delay or anticipate the recognition of realized cash flows plus an estimation error term. The error term determines to what extent the accruals map into the realization of cash flows. These error terms can be used as a measure of earnings and accrual quality. Equation four is rearranged by Dechow and Dichev (2002) into the following equation in order to empirically test it:

t t t

t

t b b CFO b CFO b CFO

WC = + + + +ε

∆ 0 1* −1 2* 3* +1 (Eq. 5)

Where:

WC = change in working capital at time t; CFOt= cash flow from operations at time t.

of rearranging equation four into equation five can be explained by showing the cash flow components as below: ) ( ) ( ) ( ) ( 1 1 1 1 1 1 2 t 1 t 1 t 1 t t t 2 t 1 t 1 t 1 t CF CF CF CF CF + + + + + + − + − − − − − − + + = + + = + + = t t t t t t t t t t t CF CFO CF CF CFO CF CFO

In the equations above the components in bold represent the measurement error components. With respect to CFOt-1 the measurement error is composed of the cash flow received and recognized in

earnings in the same period (CFOt-1t-1), plus cash payments/collections for transactions recognized in

earnings at t-2. The measurement error for CFOt consists of the cash flows that are received and

recognized in the same period. With respect to CFOt+1 the measurement error is composed of the cash

flows that are received and recognized in the same period (CFOt+1t+1) and deferred cash flows that are

to be recognized in the following period (CFOt+1t+2). Therefore as can be seen from the above

mentioned equations the measurement error for CFOt is smaller than for CFOt-1 and CFOt+1.

The cash flow from operations is estimated following Dechow et al. (1995), Leuz et al. (2003) and Wysocki (2006).

(

) (

)

CFOt = Cash flow from operations at time t;

Earningst = Net income before extraordinary items and discontinued operations at time t;

CAt= Change in current assets at time t;

CLt= Change in current liabilities at time t;

Casht= Change in cash and cash equivalents at time t;

STDt= Change in short term debt (debt included in current liabilities) at time t;

Dept= Depreciation and amortization expense at time t;

At= Total assets at time t.

Dechow and Dichev (2002) define the change in working capital as:

t t t t t t AR Inventory AP TP OtherAssets WC =∆ +∆ −∆ −∆ +∆ ∆ (Eq. 7) Where:

Inventory = Change in inventory at time t; AP = Change in accounts payable at time t; TP = Change in taxes payable at time t; Other Assets = Change in other assets (net) at time t;

All the variables mentioned above are scaled by average total assets in order to reduce heteroscedasity. Based on equations 6 and 7 it is possible to estimate the working capital accruals based on a regression analysis on equation 4. By performing a regression analysis on equation 4, it is possible to estimate the residuals which indicate the working capital accrual components.

Dechow and Dichev (2002) show in Appendix B a simulation in order to examine to what extent the R2 _{is adjusted through the estimation errors. In a situation of perfect information, where all credit sales}

are collected, all deferred revenues are recognized the following period, and there are no write-downs of inventory, there is no estimation error, since there is perfect knowledge of the related cash flows (Dechow and Dichev, 2002). In this case there is no estimation error and the coefficients are then: b1=1, b2=-1 and b3=1, with an R2 of 1,0, implying an explanation power of 100%. Appendix 2 will

present a numerical example on the signs of the coefficients.

When however the cash flow from operations needs to be estimated based on equation 6, we need to include the estimation errors. In that case we expect that the coefficients on CFOt-1 and CFOt+1

are larger than for CFOt forcing b1 and b3 more towards zero than coefficient b2. This stems from the

fact that, as depicted on the previous page, the current cash flow from operations (CFOt) has only one

component in bold, whereas the past and future cash flow from operations have two components in bold. Therefore the expected coefficients by using equation five are:

0 < b1 < 1;

-1 < b2 < 0;

0 < b3 < 1.

Whereas the R2 _{will have less explaining power, it will be biased towards zero. In order to control for}

this measurement error the regressions will be performed firm specific, industry specific and as a pooled regression. This procedure will lead to an R2 _{in line with previous research (Dechow and}

Dichev, 2002; McNichols, 2002; Cheng and Thomas, 2005; Cheng et al., 2005; Wysocki, 2006).

3.1.2 Model by McNichols (2002)

past and present cash flows. Whereas they claim that the extent to which working capital accruals map into these cash flows measures their accrual quality. The accruals are estimates of future cash flow realizations or cash flow realizations from earlier periods, which are recognized in later periods. Dechow and Dichev (2002) measure the extent to which accruals map into cash flow realizations in contemporaneous and adjacent time periods (McNichols, 2002). Since accrual quality is beyond the scope of this research the focus will be on the limitations and innovations of Dechow and Dichev (2002) and the modified model proposed by McNichols (2002).

McNichols (2002) reviews the model proposed by Dechow and Dichev (2002) and raises three issues with respect to the empirical implementation of the earnings quality measure. First, McNichols (2002) mentions that the standard deviation of the residual reflects the absolute variation in the residual, rather than the relative variation in relation to the variation in accruals. Second, McNichols (2002) elaborates on the problems concerning the estimation error on the cash flow from operations, which Dechow and Dichev (2002) also discuss in their Appendix B. Third, McNichols (2002) elaborates on the financial data regarding firms that engage in mergers and acquisitions or divestures. These transactions cause the financial data (e.g. operating cash flow) to be biased because of a change in the underlying entity. The problem with mergers and acquisitions and extreme growth is also mentioned by Collins and Hribar (2001). Collins and Hribar (2001) recommend eliminating the most extreme 1 percent of cash from operations, earnings and changes in working capital.

McNichols (2002) proposes a model in which she combines the models of Jones (1991) and Dechow and Dichev (2002) to achieve a higher explanatory power. Jones (1991) focuses on the discretionary component of total accruals and uses time-series models to estimate nondiscretionary accruals. Dechow and Dichev (2002) do not disentangle accruals in a discretionary and nondiscretionary component, they asses accruals as a whole. By combining the approaches of Jones (1991) and Dechow and Dichev (2002) it is possible to gauge the measurement error associated with the discretionary accruals of Jones (1991) and the measure of accrual quality of Dechow and Dichev (2002). The Jones (1991) model can be written as:

t t t t b b Sales b PPE WC = + ∆ + +ε ∆ 0 1* 2* (Eq. 8) Where:

WCt = the change in working capital at time t;

Salest = change in sales at time t;

PPEt = level op property, plant and equipment at time t.

Where all variables are scaled by average total assets in order to reduce heteroscedasticity.

firm. This change in sales is an objective measure of the operations of a firm because it is beyond the influence of the managers, though they are not completely exogenous (Jones, 1991)2_{. The absolute}

level of property, plant and equipment is used to control for the relative amount of the total accruals that is related to the nondiscretionary depreciation expense. The absolute amount of property, plant and equipment is used because the absolute amount of depreciation expenses is used in the working capital accruals.

The Jones (1991) model assumes that accruals react to changes in current sales, but do not react to lagged and future sales. Bernard and Stober (1989) and Dechow et al. (1998) document that accruals do not fully adjust to a contemporaneous sales shock, but rather adjust over succeeding periods. On the other hand is it likely that future sales influence the estimates of managers, which will be reflected in accruals. McNichols (2001) documents that long-term earnings growth forecasts by analysts have significant explanatory power for residuals estimated using the Jones model, suggesting that earnings growth is a significant correlated omitted variable in the model (McNichols, 2002). The findings by Dechow and Dichev (2002) show that past, present and future operating cash flows can increase the explanatory power of the Jones model since these are correlated with the firms’ economic environment.

Combining the models of Dechow and Dichev (2002) and Jones (1991) leads to the proposed model of McNichols (2002), which is presented below.

t t t t t t t PPE b Sales b CFO b CFO b CFO b b WC ε + + ∆ + + + + = ∆ _{− +} * * * * * 5 4 1 3 2 1 1 0 _{(Eq. 9)} Where:

WCt = change in working capital;

CFOt = cash flow from operations at time t;

Salest = change in sales at time t;

PPEt = level of property, plant and equipment at time t.

McNichols (2002) tests the models of Dechow and Dichev (2002), Jones (1991) and the new proposed model by McNichols (2002). The findings with respect to the model by Dechow and Dichev (2002) are in line with the findings by Dechow and Dichev (2002). McNichols finds a significant positive correlation between the accruals and prior and past year cash flow from operations, whereas she finds a significant negative correlation between the accruals and the current year cash flow from operations. McNichols (2002) finds an R2 _{of 0.20. With respect to the model of Jones (1991), McNichols (2002)}

finds a significant positive correlation between the accruals and the change in sales, whereas she finds

2 _{Reported earnings may be affected to some extent by managers’ attempts to decrease reported earnings (Jones,}

a significant negative correlation with respect to the level of property, plant and equipment. These findings are consistent with the findings of Jones (1991). McNichols documents an R2 _{of 0.07 with}

respect to the Jones model. With respect to the model by McNichols (2002), she finds a significant positive correlation between accruals and the past and future cash flows from operations and a significant positive correlation between accruals and the change in current year sales. She documents a significant negative relation between accruals and the current year cash flow from operations and between accruals and the level of property, plant and equipment. These findings are all in line with the findings from the original models by Dechow and Dichev (2002) and Jones (1991). The only remarkable finding is the increase in the R2_{, which was 0.20 for Dechow and Dichev (2002) and 0.07}

for the Jones (1991) model. The R2 _{of the model by McNichols amounts 0.30, which indicates the}

superiority of her model over both other models.

3.1.3 Model by Cheng, Peng and Thomas (2005)

Cheng, Peng and Thomas (2005) estimate abnormal accruals using 22 models. These models are all variations on the Jones (1991) model and the model by Dechow and Dichev (2002). Cheng et al. (2005) measure the relation between current abnormal accruals and future abnormal return in order to identify an approach in modeling abnormal accruals in the context of investor mispricing (Cheng et al., 2005). Cheng et al. (2005) also consider the findings from Kothari et al. (2005) which suggest controlling for a firm’s return on assets (ROA). Since a best model approach is not distinguished in the current literature Cheng et al. (2005) perform a large scale comparison. Cheng et al. (2005) examine three dichotomous choices with respect to measuring accruals and two dichotomous choices with respect to the regression model. Cheng et al. (2005) identify three measurement types for accruals:

Use of total accruals versus working capital accruals

Use of accruals from the statement of cash flows versus estimates from the changes in the balance

sheet items

Scaling of the variables by average total assets or by lagged total assets.

With respect to the regression models Cheng et al. (2005) identify two types of regression models:

Estimation by firm versus cross-sectional within industry Inclusion versus exclusion of an intercept in the model.

Since the above mentioned choices are dichotomous there are 25 _{= 32 possible accrual}

1. The original model by Dechow and Dichev (2002)

2. The proposed model by Cheng et al. (2005), see equation 10 below. 3. The original model by McNichols (2002).

Within each of the 32 accrual measurement/regression approach combinations, abnormal accrual models are ranked for each of the seven test measures (table 5, Cheng et al., 2005).

The model that Cheng et al. (2005) propose is a combination of the model by Dechow and Dichev (2002) and the Modified Jones Model by Dechow et al. (1995). Interpreted from another perspective, the proposed model is an adjustment to the model by McNichols (2002). Equation 10 presents the model proposed by Cheng et al. (2005).

(

)

WC = change in working capital accruals at time t; CFOt = cash flow from operations at time t;

Salest = change in sales at time t;

Rect = change in receivables at time t;

PPEt = level of property, plant and equipment at time t.

All variables are scaled by average total assets in order to reduce heteroscedasticity.

Dechow et al. (1995) present the Modified Jones Model. The model is a small amendment to the original Jones (1991) model. The amendment encompasses the correction of the change in sales for the change in receivables. Jones (1991) controls for gross property, plant and equipment and the change in revenues estimating discretionary accruals because of changing economic circumstances. Jones (1991) measures the total accruals as the change in current assets, the change in cash, change in current liabilities, and depreciation and amortization3_{. Since items like the change in accounts receivable,}

inventory and accounts payable reflect changes in working capital accounts, they are subject to changes in revenues. Revenues are used to control for the economic environment of the firm because they are an objective measure of the firms’ operations before managers’ manipulations, but they are not completely exogenous (Jones, 1991). The Modified Jones model as described by Dechow et al. (1995) corrects the change in revenues for the change in receivables. The reasoning behind this is that it is easier to manage earnings by means of the recognition of revenue on credit sales than it is through the recognition of revenues through cash sales. The Modified Jones Model assumes that all changes in credit sales in the event period result from earnings management.

3.2 Discretionary and nondiscretionary accruals

Total accruals can be divided in discretionary and nondiscretionary accruals. Nondiscretionary (or normal) accruals are based on the concept of accrual accounting and are therefore a perfectly normal phenomenon. The nondiscretionary accruals are liable to the economic circumstances of the firm. Therefore the nondiscretionary accruals are, in principle, free from manipulation by managers. The fact that accruals reverse over time is also known to managers. It is however far more difficult to conceal the manipulation of short-term accruals than it is to conceal the manipulation of long-term accruals. This stems from the fact that the market expects these short-term accruals to reverse within one accounting period, according to Whelan (2004). Long-term accruals reverse further into the future, which provides managers a longer time period to conceal the manipulation and to gradually reverse the manipulation (Whelan, 2004). Therefore discretionary accruals may possibly be the result of opportunistic behavior by managers, which leads to a lower level of earnings quality. Since the discretionary component of accruals is liable to manager’s discretion they are used as an indicator of earnings management (Jones, 1991; DeFond and Jiambalvo, 1994; Dechow et al., 1995; Teoh et al., 1998a, 1998b; Bartov, 2001).

3.2.1 Short term and long term discretionary accruals

Discretionary accruals can be decomposed into short-term and long term discretionary accruals. Short term accruals affect working capital accounts and changes to current assets and current liabilities. Long-term accruals include depreciation, future tax benefits, employee entitlements, assets revaluation, and adjustments to the fair value of financial instruments (Whelan, 2004). Guay et al. (1996) find that the market relies more on nondiscretionary accruals than on discretionary accruals, whereas Guay et al. (1996) were not able to distinguish between the opportunistic and performance related use of discretionary accruals. Guay and Sidhu (2001) find that although both short-term and long-term accruals provide incremental information to the market, the short-term accruals are more value relevant than the long-term accruals. The incremental explanatory power of the long-term accruals is impeded by the timeliness problems and measurement error in the indirect method of computing cash flows and accruals. However, Cotter (1996) and Guay and Sidhu (1996) document that the explanatory power of the long-term accruals increases as the return interval increases. The discretionary nature of short-term and long-term accruals has until 2004 (Whelan) not been addressed in the literature.

documents that only the Jones (1991) model and the model by Kang and Savaramakrishnan (1995) perform moderately well in estimating discretionary and nondiscretionary accruals.

The models of Jones (1991), DeFond and Jiambalvo (1994) and Dechow et al. (1995) include two variables (the change in sales and the level of property, plant and equipment) in order to identify the drivers of nondiscretionary accruals based on a firm’s economic circumstances (Whelan, 2004). The change in sales can be regarded as a short-term driver of discretionary accruals, whereas the level of property, plant and equipment can be regarded as a long-term driver of discretionary accruals. Since the focus is on only two components, it is possible that other potential drivers of short-term or long-term components are overlooked. As short-long-term and long-long-term accruals have the potential to mitigate different timing and matching problems, distinguishing between these components is an important consideration in any investigation of the role of discretionary accruals as an earnings management tool (Whelan, 2004). Therefore, there is a need for a model that further decomposes total accruals into short-term and log-term discretionary accruals according to Whelan (2004).

The focus in my study is on the models by Dechow and Dichev (2002), McNichols (2002) and Cheng et al. (2005). These models are based on aggregate accruals, but none of these models makes a distinction between discretionary and non-discretionary models, and between short-term and long-term discretionary accruals. The contribution of this study lies in the fact that three new models, based on the models by Dechow and Dichev (2002), McNichols (2002) and Cheng et al. (2005), are developed in order to estimate the short-term discretionary working capital accruals. The estimation of short-term discretionary working capital accruals is innovative, since it has not been subject of investigation before.

3.2.2 Short-term discretionary accrual models

The model by Dechow and Dichev (2002) proposes a model on working capital accruals based on past, present and future cash flows from operations. However, they do not differentiate between a discretionary and nondiscretionary component. Wysocki (2006) indicates that the model has limited liability to distinguish between nondiscretionary and discretionary accruals, or capture a firm’s accounting quality, because of the negative contemporaneous correlation between accruals and cash flows. Since the focus is on short term discretionary accruals it is necessary to eliminate components that are associated with the long-term accruals. Whelan (2004) proposes a model in order to estimate long-term discretionary accruals based on the Jones (1991) model, which uses the following key drivers of long-term accruals:

Non-current provisions.

Based on these findings the working capital accrual will be corrected for depreciation and amortization. Depreciation can be regarded as an estimate for the level of property, plant and equipment. The item amortization can be regarded as an estimate for the level of intangibles. By subtracting the depreciation and amortization from the working capital accrual most of the items concerning the long-term accruals are eliminated.

With respect to the short-term discretionary accruals, Whelan (2004) identified the change in revenues as a short-term component based on the Jones (1991) model. Since the models by Dechow and Dichev (2002), McNichols (2002) and Cheng et al. (2005) use past, present and future operating cash flows in order to estimate working capital accruals, these components are considered to be short-term oriented. The reasoning behind this is that operating cash flows have a short-term orientation. The cash flow from investing activities and the cash flow from financing activities encompass a long-term orientation. Since the operational cash flows have a short-term orientation they are taken into account in estimating short-term discretionary accruals. The original model by Dechow and Dichev (2002) can be considered as a model on short-term accruals. This can be explained by the fact that Dechow and Dichev (2002) focused on working capital accruals, which can be labeled as having a short-term orientation.

The model of McNichols (2002), which also considers the change in sales and the level of property, plant and equipment, is adjusted. In this model the level of property, plant and equipment is eliminated because of its long-term orientation. The change in sales is considered to be short-term oriented because changes in sales automatically lead to changes in the working capital because of the changes in inventory, accounts payable, accounts receivable e.g. The model by Cheng et al. (2005) is also adjusted in line with the model by McNichols (2002) by eliminating the level of property, plant and equipment. Cheng et al. (2005) also consider the change in sales minus the change in receivables. This component can be considered as short-term oriented, because the changes in sales minus the changes in receivables are the cash flows that accompany the changes in sales. By correcting for the change in receivables it corrects for the automatic change in receivables, which makes the term more cash oriented. Based on the above stated reasoning the following three models are presented:

Dechow and Dichev (2002):

t t t

t

t b b CFO b CFO b CFO

Cheng et al. (2005):

(

)

ε

ExWCAt = Expected working capital accrual at time t.

In order to arrive at the working capital accruals the coefficients for b0, b1, b2, b3 and b4 are used from

the regression using respectively equation 5, 9 and 10 for the models by Dechow and Dichev (2002), McNichols (2002) and Cheng et al. (2005). By substituting these coefficients into the corresponding equation (11, 12 and 13) the output represents the expected working capital accruals. Because the short-term working capital accrual is needed the depreciation and amortization charges need to be subtracted. By subtracting the depreciation and amortization, all the components that have a long-term orientation are eliminated.

j i j i k j i ExWCA Depr ExWCA ST_ _{, ,} = _, − _. (Eq. 14) Where:

ST_ExWCAi,j,k = expected short-term working capital accrual at time i, for company j, and

using model k;

ExWCAi,j = Expected working capital accrual at time I, for company j;

Depri,j = Depreciation and amortization at time I, for company j.

Since the forecast is assumed to be representing the nondiscretionary component of the short-term accruals, the difference between this estimation and actual short-term accrual represents the discretionary component of the short-term accruals (Whelan, 2004). This leads to the following equation: j i j i j i WC ST ExWCA WCA St D_ _ _, =∆ _, − _ _, (Eq. 15) Where:

D_St_WCAi,j = discretionary short-term working capital accrual at time i and for company j;

WCi,j = change in working capital at time i, and for company j;

ST_ExWCAi,j = expected short-term working capital accrual at time i and for company j.

Chapter 4

Corporate governance and earnings management

This chapter elaborates on the relationship between corporate governance and earnings management. Both topics have been discussed separately in the previous two chapters. The strength of corporate governance mechanisms and the practice of earnings management are related. Strong internal governance mechanisms involve a balance between with an appropriate level of monitoring and corporate performance (Cadburry, 1997).

With respect to earnings management the internal governance structure of the company is highly relevant, because an adequate internal structure can reduce earnings management to a minimum level. The internal governance structure of a firm consists of the functions and processes established to oversee and influence the actions of the firm’s management (Davidson et al., 2005). The role of these mechanisms in relation to financial reporting is to ensure compliance with mandated reporting requirements and to maintain the credibility of a firm’s financial statements (Dechow et al., 1995). This study focuses on the role of the supervisory board and the audit committee in constraining earnings management. This chapter will discuss these mechanisms in relation to earnings management. In section 4.1, I will elaborate on the supervisory board and in section 4.2 on the audit committee.

4.1 Supervisory board

presence of an independent nomination committee (Klein, 2002a). Since the Netherlands is mainly characterized by a two-tier structure, the roles of the chairman and the CEO are separated. The role of independent directors on the board is executed in the Netherlands by the members of the supervisory board.

Non-executive directors have incentives to develop a reputation as experts in decision control and monitoring (Fama and Jensen, 1983a,b). In a two-tier system the supervisory board should be independent of the company and is responsible for the supervision of the policy that is developed by the management board. The independency of the supervisory board follows from the independency of its non-executive directors. A non-executive director, who is entirely independent from management, is expected to offer shareholders the greatest protection in monitoring management (Baysinger and Butler, 1985). It is therefore the case that the members of the supervisory board, the non-executive directors, are independent. I hypothesize a negative association between supervisory board independence and earnings management.

H1: Earnings management is negatively associated with the independency of the

supervisory board.

Another indicator is the size of the board of directors. In the literature there is no consensus on the direction of the relationship between board size and its effectiveness. The larger the board, the less likely it is to function effectively. On the other hand a large board provides more expertise and better environmental links (Dalton et al., 1999). Dalton et al. (1999) documented a positive and significant relation between board size and financial performance. Yermack (1996) demonstrated that smaller boards (less then ten directors) are better performers. Eisenberg et al. (1998) also argued that smaller boards are associated with better financial performance. Beasley (1996) found a positive relationship between board size and the likelihood of financial statement fraud, whereas Abbott et al. (2000) found no relationship between the two (Chtourou et al., 2001). Kao and Chen (2004) documented that the larger the board size, the higher the extent of earnings management. Rahman and Ali (2006) argued that there is a significant negative relationship between discretionary accruals and the size of boards. Xie et al. (2003) and Peasnell et al. (2001) found that having a larger board is associated with less earnings management.

Previous research is predominantly based on one-tier structures. Since listed firms in the Netherlands are mainly characterized by a two-tier structure, the findings are to some extent difficult to comparable with the previous literature. This study will focus on the monitoring role and capacity of the supervisory board. The monitoring capabilities are expected to cohere with the size of the supervisory board in relation to the size of the management board. Since previous research is dispersed on the sign of the relation, I do not hypothesize a specific relation between the relative size of the supervisory board and earnings management.

The number of meetings of the supervisory board is indicative of the quality of their monitoring activities. Vafaes (1999) argued that when boards met more often, the companies show improved financial performance. Supervisory boards that meet more often have more time available to discuss items like earnings management. Therefore, I hypothesize that the number of meetings of non-executive directors is negatively related to earnings management.

H3: Earnings management is negatively associated with the number of meetings of the

non-executive directors.

The longer a non-executive director sits on the supervisory board, the more knowledge he has about the company and its executive directors. This allows the non-executive director to develop his monitoring competencies. The non-executive director, therefore, becomes more capable of overseeing the firm’s financial reporting process effectively (Chtourou et al., 2001). Kosnik (1987) argued that as the average tenure of non-executive directors increases, the company is more likely to resist hostile takeover bids. Beasley (1996) and Chtourou et al. (2004) documented that the likelihood of financial accounting fraud is a decreasing function of the average tenure of non-executive. On the other hand, it is intuitively appealing that when the non-executive director sits longer on the supervisory board for a longer period, he will become more involved with the company. Confronted with an in depth knowledge of the company it is possible to become too much acquainted. Based on these conflicting statements on the direction of the relation, I hypothesize that the average tenure of non-executive directors is related to earnings management.

H4: Earnings management is associated with the average tenure of non-executive

directors.

4.2 Audit committee

According to Davidson et al. (2005), the effectiveness of the audit committee relates to its independence, size and activeness of the audit committee. Another aspect of the audit committee that is examined by Chtourou et al. (2001), Xie et al. (2003) and Bédard et al. (2004), deals with the competence of the directors, more specifically the financial competencies of the directors. Since data on the profession of the directors is not available, I am not able to investigate this aspect concerning the professional background of the non-executive directors.