Is the Wta based on facts or fiction?

Transcription

1 Is the Wta based on facts or fiction? Rens Adam August 15, 2007

2 Is the Wta based on facts or fiction? Master thesis Department Accountancy, Faculty of Economics and Business Administration, Tilburg University Name: Rens Adam ANR: Company: KPMG Beemdstraat 1, Eindhoven Date: August 15, 2007 Supervisors: Yuping Jia (UvT) Michael Regouw (KPMG)

3 Preface In front of you is my master thesis that I wrote during the period April 2007 till August This master thesis completes my master in Accounting at Tilburg University. Its aim is to investigate whether the Wet toezicht accountantsorganisaties (Wta) can be justified by empirical evidence. The Wta is a new Dutch law that made audit regulation more severe in response to recent accounting scandals and new regulation in the US and EU. Since it is often suggested that major accounting scandals imply that audit quality has declined, I want to find out whether audit quality really declined over time or that these scandals are just some extreme incidents. Furthermore, in Accounting Theory class it was told that accounting regulation over time is hardly evidence based. Therefore this study uses the Wta to test whether accounting regulation is evidence based by investigating if audit quality has declined over time. I would like to thank my supervisor from Tilburg University, Ms. Jia, for her valuable comments and advice. I want to thank KPMG Eindhoven for facilitating me during the period I wrote my thesis, especially Michael Regouw, who supervised my thesis at KPMG. Furthermore, I want to thank my colleagues Ali, Arnoud, Folkert, Niek, and Robert for their help and the nice time. Last but not least, I am grateful to my parents for providing me with the opportunity to go to university, my fraternity for the nice student-time, and my girlfriend Caroline for always supporting me. Rens Adam, Tilburg, August 15, 2007

4 Table of Contents Preface... Abstract 1. Introduction.. 2. Literature & hypothesis 2.1 Background Properties of the Wta 2.3 Hypotheses development. 3. Methodology 3.1 Selection of measure Discretionary accruals model Tests of hypotheses Sample.. 4. Results Data collection. 4.2 Descriptive Statistics 4.3 Test of hypothesis Test of hypothesis Additional tests 5. Conclusions, limitations, and suggestions for further research... References

5 Abstract The Wta is a new Dutch law, which is enacted to improve audit quality. The aim of this study is to find out whether the Wta can be justified by empirical evidence. This is done in two ways. First, this study investigates whether the WTA can be justified ex ante by investigating whether audit quality declined in the Netherlands prior to enactment of the Wta. Second, it is investigated whether the Wta can be justified ex post by investigating whether audit quality increased in the Netherlands after enactment of the Wta. Since audit quality cannot be directly measured, discretionary accruals, which are inversely related to audit quality, are used as a proxy for audit quality. This study finds a negative relation between audit quality and time, whereas no relation between audit quality and the Wta is found. Since these results are robust to different testing methods, I conclude that the Wta can be justified by ex ante evidence, since audit quality is negatively related to time prior to its enactment.

6 1. Introduction This study investigates whether the Wta can be justified by empirical evidence. The Wta is a new law in the Netherlands which is enacted to increase audit quality in response to the recent accounting scandals. However it is not clear whether these accounting scandals are some extreme events or that average audit quality has actually declined. Furthermore, Buijink (2006) indicates that it is hard to argue that financial reporting regulation around the world is evidence based. Therefore this study uses the Wta to test whether accounting regulation is evidence based by investigating whether audit quality has declined prior to its enactment. Furthermore this study provides preliminary evidence on the effect of the Wta on audit quality. In my study I fina a negative relationship between audit quality and time in the Netherlands prior to enactment of the Wta. Therefore the Wta can be justified by empirical evidence. However no relation between audit quality and the Wta is found. This implies that the Wta is no contribution to audit regulation in the Netherlands, yet. Since audit quality cannot be directly measured (Francis; 2004) discretionary accruals are used as a proxy. Discretionary accruals are estimated by applying the modified-jones model (Dechow et al.; 1995) The sample used consists of Dutch firms listed on the Amsterdam stock exchange between 1993 and This study contributes to the existing literature by investigating audit quality over time and the effect of more severe regulation on audit quality in a different institutional setting than the US. The social relevance of this study is whether the Wta is a contribution to the existing Dutch accounting regulation. This study continues as follows. In section two, I discuss the coming about of the Wta, the properties of the Wta, and at last I develop my hypotheses. Section three discusses the variables and methods used to test my hypotheses. The results of these tests are discussed in section four. Finally, I provide conclusions and recommendations for further research in section five.

7 2. Literature and hypotheses In this section I first discuss why the Wta is enacted by the Dutch government. Secondly, I summarize the contents of the Wta. Lastly, I develop my hypotheses, which are based on the findings of previous literature. 2.1 Background Nowadays large firms are characterized by the separation of ownership and control. The management (agent) interests are different from those of the shareholders (principals). The management wants to maximize its own wealth as much as possible, whereas the shareholders want to maximize the wealth of the firm as much as possible. Management can increase its own wealth by withdrawing wealth from the firm, and thus reducing the wealth of the shareholders. In the literature this is referred to as the agency problem (Jensen and Meckling; 1976, Fama and Jensen; 1983). Management communicates the financial position of the firm with its shareholders through the financial statements. Since information asymmetries exist between management and owners, the owners do not know whether the financial statements faithfully represent the financial position of the firm. Therefore management has opportunities to manage a firm s earnings for its own goals or some other purpose, without notice of the shareholders. The audit of the financial statements by an independent accountant mitigates the risk of earnings being managed (Deckers and Van Kollenburg; 2002). Since the accountant has inside information and is independent, the audit is a signal to the shareholders that the financial statements can be trusted (Chow; 1982). Furthermore, Watts & Zimmerman (1983) suggest that the audit existed early in the development of firms (1200) to reduce agency costs. Thus the audit decreases management opportunity to manage earnings and therefore increases the quality of the financial statements. However, in the beginning of this decade, accounting firms proved not being able to remove all earnings management from the corporate books. In this period it became public that some major firms managed their earnings upward. The most widely cited accounting scandal is Enron. Enron applied mark-to-market accounting to overstate its earnings and used special purpose entities to understate debt and overstate earnings and equity. At the end of 2001, Enron had to reduce its past earnings by $613 million, increase liabilities by $628 million, and decrease equity with $1.2 billion (Healy and Palepu; 2003). After Enron more accounting scandals

8 followed. Amongst others, WorldCom overstated cash flow by booking $3.8 billion in operating expenses as capital expenses and gave founder Bernard Ebbers $400 million in off-the-books loans. Qwest inflated revenue using network capacity swaps and improper accounting for long-term deals (Forbes). In Europe there were accounting scandals too. In order to get extra bonuses, the management of Ahold subsidiary US Food Services intentionally booked higher promotional allowances to show higher profits. Parmalat borrowed money from multinational banks and justified those loans by inflating its revenues through fictitious sales to retailers. It would then cook its books some more to make the debt vanish, by transferring it to shell companies based in offshore tax havens (Gumbel; 2004). Legislative authorities reacted immediately. In July 2002, only three quarters after the first accounting scandal, the US government enacted the Sarbanes-Oxley act (SOX). SOX establishes the Public Company Accounting Oversight Board (PCAOB), which is an independent supervisory authority. The PCAOB supervises the audit and has the right to set accounting standards (Vergoossen and Wallage; 2004). Furthermore, SOX prohibits auditors from performing certain nonaudit services for their audit clients, and imposes greater criminal penalties for corporate fraud. Section 404 of SOX requires that management assess internal controls and that auditors report on their clients internal controls (Zhang; 2005). All firms that are listed on a US stock exchange must comply with SOX. Therefore, the legislative power of SOX is extraterritorial (Vergoossen and Wallage; 2004). However, the PCAOB announced to lean on the activities of recognized foreign counterparts. In May 2006, the European parliament revised the eighth directive. The revised eighth directive is a reaction on the recent accounting scandals as well as the extraterritorial power of SOX (Bolkestein; 2004). It demands an accounting organization to be independent of its audit clients. Thus, it prohibits performing nonaudit services to audit clients, as SOX does. Furthermore, each EU member state should establish an independent supervisory authority that supervises the audits of the accounting organizations (Official Journal of the European Union; 2006). Then these European counterparts of the PCAOB can supervise audits of European firms that are listed on an US stock exchange. In summary, the Wta is established because of three reasons. First, the recent accounting scandals demand new regulation to guarantee the quality of the audit. So the audit of the financial statements can be retained as a signal to the shareholders that

9 the financial statements can be trusted. Second, the extraterritorial power of SOX forces the Netherlands to have its own independent supervisory authority if it does not want its cross-listed organizations to be supervised by the PCAOB. At last, since the Netherlands are a member state of the European Union, they must follow the directives set by the European Union. Thus the Wta is the implementation of the revised eighth directive. This summary is graphically represented in figure 1. Figure 1: Causes of the Wta 2.2 Properties of the Wta 1 The Wta makes a distinction between four different domains: A. Legal audits. In addition to the audit of the financial statements of firms, which is the most important legal audit, the Wta also classifies 30 other audits as legal, such as the audits of municipalities, provinces, hospitals, educational institutions, and cooperative housing associations. B. The issuance of a qualified opinion demanded by law, such as legal (de)mergers, or subsidy applications. C. Review orders and voluntary audits. D. Other orders such as financial statements compilation, due diligence investigations, announcements with prospectuses, forensic tasks, valuation, tax advices, and other advisory services. 1 All information in this section is based on the references with a.

10 The scope of the Wta is the legal audit of domain A. Accounting firms that perform the legal audit are therefore under the supervision of the AFM, the Financial Market Authority, which is the external supervisory authority established by the Wta. The professional bodies maintain the supervision on domain B, C, and D and maintain the right to issue accounting standards. The accounting organization needs a license of the AFM to perform the legal audits of domain A. As a license holder, the accounting organization is responsible for compliance with the Wta. This does not only relate to compliance with the rules in the Wta concerning the accounting organization itself, but also to the compliance of the external accountants it employs. So the accounting organization is the basis for the supervision by the AFM. The concept of the external accountant is introduced by the Wta. External accountants are natural persons who are employed by an accounting organization and are responsible for executing the legal audits. The accounting organization needs to register its external accountants with the AFM. The Wta sets requirements to the external accountants concerning craftsmanship, independence, objectivity, and integrity. Violation of theses rules can lead to charges to the external accountant and the accounting organization where the external accountant is employed. Furthermore, external accountants are only allowed to perform a firm s audit for seven subsequent years. Therefore, external accountants have to sign the financial statements with their own name instead of the accounting organization s name. Since the accounting organization is the basis for the supervision by the AFM, the functioning of the daily board becomes of interest. The Wta sets requirements concerning the reliability of the members of the board. The functioning of the board is dependent on the quality assurance with respect to the administrative organization and internal control (AO/IC) of the accounting organization. Therefore, the Wta demands controlled and integer management. This means amongst others: Control of business operations and operational risks. Prevention of a conflict of interest. Prevention of involvements of the accounting organization, external accountants and employees with (legal) offences harmful to the confidence in the accounting organization or the financial markets. Prevention of relations with customers harmful to the confidence in the accounting organization or the financial markets

11 Avoidance of the independence of an external accountant or another employee becoming untrustworthy during a legal audit. For the legal audit and the set up of the audit working papers, standard procedures are required by the Wta. The audit working paper need to incorporate amongst others: A recording of the assessment of independence of the accounting organization. A recording of reported frauds. A recording whether the accounting organization has enough skilled employees to fulfill the audit. A recording of the assessment of the integrity of the customer. A recording of professional consults in the audit working paper. In summary, the Wta establishes an independent external supervisory authority, the AFM. Accounting organization who perform legal audits need to register itself and its external accountants with the AFM. The AFM will supervise the AO/IC, audit working papers, and the reliability of board members of an accounting organization. Furthermore, the AFM will examine whether external accountants meet the requirements set by the Wta. The AFM is authorized to impose sanctions if accounting organizations do not comply with the rules. The AFM can impose a fine with a maximum amount of Euro 900,000 per separate violation, a designation order to follow a certain line of conduct, and as a last resort a violation can lead to the withdrawal of an accounting organization s license. 2.3 Hypotheses development As it is the aim of the Wta to increase the quality of the audit, one can infer that the quality of the audit was too low prior to the Wta. According to Francis (2004), the number of outright audit failures is much less than one percent of all audits done annually. So audit failures with material consequences occur relatively infrequent. Gul and Krishnan (2002) compared audit quality between the period and the period. They find that audit quality, measured by discretionary accruals, has declined. Cohen et al. (2004) find that earnings management increased dramatically in poorly performing industries during the period of the major accounting scandals. Since the audit has to prevent earnings from being managed, it can be induced that audit quality has declined during the period of the major

12 accounting scandals in the US. Jain et al. (2003) find evidence that the accounting scandals widened the bid-ask spreads and lowered the depths. 2 Stocks with less reliable information have higher adverse selection costs for market makers, whom they recover through higher bid-ask spreads. Since the audit should improve the reliability of the information in the financial statements, it can be induced that audit quality has declined. Although, outright audit failures are limited, different proxies for audit quality find that audit quality has declined recently in the US. Though the institutional settings are different in the Netherlands, most public firms are audited by Big Four auditors. These Big Four auditors are multinationals that use the same mechanical audit procedures in every country where they are located (Healy and Palepu; 2003). Therefore, I expect that audit quality has also declined in the Netherlands prior to the enactment of the Wta. So my first hypothesis will be: Hypothesis 1: Audit quality has declined in the Netherlands prior to the enactment of the Wta. If hypothesis one is accepted, the Wta is empirically justified ex ante. By establishing new quality demands for accounting organizations, and an independent external supervisory authority, the Wta makes the audit regulation in the Netherlands more severe. Maijoor and Vanstraelen (2006) find that companies in countries with a more flexible audit regulation engage in more earnings management than companies in countries with a strict audit regulation. This is in conformity with Leuz et al. (2003), who find that companies in countries with stronger investor right and more legal enforcement, engage in less earnings management. Lobo and Zhou (2006) investigate the change in managerial discretion over financial reporting following SOX. They find a significant reduction in managerial discretion in the post-sox period relative to the pre-sox period. Since managerial discretion gives managers the freedom to manage earnings, it is usually associated with low audit quality in the literature. Cohen et al. (2004) find that earnings management decreases after SOX. Furthermore, Jain et al (2003) find that after SOX, the bid-ask spread declined and the depth in the market increased. Vergoossen and Wallage (2004) compare Sox and the Wta. They find that the policies in Sox and the 2 Depth is the ability of a security to absorb buy and sell orders without the stock price dramatically moving on either direction, which implies that trading volume increases with depth.

13 Wta are very similar. However, the policies of SOX are more severe and under the Wta, professional bodies are still allowed to set their own standards. In summary, the Wta made audit regulation more severe in the Netherlands, what is associated with less earning management and thus higher audit quality. Furthermore, SOX and the Wta have roughly the same policies and all evidence on SOX indicates that SOX is associated with higher audit quality. Therefore I expect that the audit quality has also increased in the Netherlands after the enactment of the Wta. Therefore, my second hypothesis will be: Hypothesis 2: Audit quality has increased in the Netherlands after the enactment of the Wta. If hypothesis two is accepted, the Wta is empirically justified ex post. However, there is a restriction on the results of the test of hypothesis two. The Wta is effective since October Therefore the annual reports of 2006 are the first financial statements that are audited under the Wta. So there is only one year of data available and hence results have to be interpreted as preliminary evidence of the effect of the Wta.

14 3. Methodology This section discusses the method that will be used to test my hypotheses. First, I explain why I am using discretionary accruals as a measure of audit quality. Second, I discuss which model I am going to use to measure discretionary accruals since there are different models to measure discretionary accruals. Third, I develop two regression models to test the hypotheses. The sample that I intend to use to test my hypotheses will be discussed last. 3.1 Selection of measure Audit quality can be directly measured in the case of outright audit failures only (Francis; 2004). When audit quality is conceptualized as a continuum ranging from very low to very high, audit failures will occur on the lower end of the quality continuum. Since audit failures only measure one side of the continuum, they are not very informative about average audit quality. Another problem of audit failures as a measure of audit quality is that they are not very timely. The recent accounting scandals provide evidence that it can take years before audit failures are recognized. Since the only direct measure of audit quality is not informative about average audit quality and is not a timely measure, a proxy will be used to measure audit quality instead. To overcome timing and matching problems that cause realized cash flows to be noisy measures for performance, the primary role of accruals is to make earnings more informative (Dechow; 1994). According to Schipper and Vincent (2003), the unmanipulated and error-free portion of accruals increases the extent to which accounting earnings faithfully represent income. However, discretionary accruals which capture measurement error in accruals vis-à-vis cash from operations, provide opportunities for management to manipulate earnings.. Becker et al. (1998) and Francis et al. (1999) find that companies with Big Six auditors have significant less discretionary accruals compared to companies with non-big Six auditors. According to Teoh and Wong (1993) Big Eight auditors are associated with higher audit quality than non-big Eight audit firms. Furthermore, Lee et al. (1999) find that fraud firms are associated with high discretionary accruals. Thus low (high) discretionary accruals are associated with high (low) audit quality. Thus, the higher audit quality, the less discretion to manage accruals for the firm s management. Therefore, discretionary accruals will be used as a proxy for audit quality.

15 As measure, discretionary accruals have the advantage of data being widely available, but there are several drawbacks too. Discretionary accruals are a noisy measure of audit quality and potentially performance biased. Furthermore, evidence is necessarily indirect, making it difficult to convincingly rule out alternative explanations for the findings (DeFond and Jiambalvo; 2005). My results might suffer from these drawbacks too. 3.2 Discretionary accruals model Several discretionary accrual models are evaluated by Bartov et al. (2001) on their ability to detect earnings management. They find that the modified-jones model, the cross-sectional Jones model and the cross-sectional modified-jones model are the most successful in identifying an association between audit qualifications and discretionary accruals. Since my sample includes enough observation years to perform a time-series analysis, I will use the modified Jones model to estimate discretionary accruals. The Jones model (Jones; 1991) relaxes the assumption from previous discretionary accruals models that nondiscretionary accruals are assumed to be constant. Her model attempts to control for changes in the economic environment of the company by including the change in revenues, because they are an objective measure of the firms operations before management manipulations. Gross property, plant, and equipment is included to control for the portion of total accruals related to nondiscretionary depreciation expense: TA it /A it-1 = α[1/a it-1 ] + β 1 [ REV it /A it-1 ] + β 2 [PPE it /A it-1 ] + ε it (1) Where: TA it = total accruals in year t for firm i; A it-1 = total assets in year t-1 for firm i; REV it = revenues in year t less revenues in year t-1 for firm i; PPE it = gross property, plant, and equipment in year t for firm i; ε t = error term in year t for firm i. In equation (1), all variables are scaled by lagged assets to reduce heteroscedasticity. Ordinary least squares at firm level is used to obtain estimates for α, β 1, and β 2.

16 The error term represents the amount of discretionary accruals in the modified Jones model (Dechow et al.; 1995). The coefficients α, β 1, and β 2 from the original Jones model are used, to determine the amount of discretionary accruals. The modified Jones model eliminates the conjectured tendency of the Jones model to measure discretionary accruals with error when discretion is exercised over revenues, by adjusting the change in revenues for the change in receivables: Where: TA it /A it-1 = α[1/a it-1 ] + β 1 [( REV it - REC it )/A it-1 ] + β 2 [PPE it /A it-1 ] + ε it (2) REC it = net receivables in year t less net receivables in year t-1 for firm i. The original Jones model implicitly assumes that discretion is not exercised over revenues. The modified version of the Jones model assumes that all changes in credit sales result from earnings management. Despite the availability of accurate accruals data, the majority of previous studies use the indirect balance sheet approach to calculate total accruals data. However, the indirect balance sheet approach can lead to erroneously concluding that earnings management exists when no such activity is present (Hribar and Collins; 2002). Since a portion of the balance sheet working capital accounts relates to the non-operating events, and is erroneously shown as accrual. Therefore, total accruals are calculated as follows: TA t = EBXI t - CFO t (3) Where: EBXI t = earnings before extraordinary items and discontinued operations in year t; CFO t = operating cash flows (from continuing operations) in year t. 3.3 Tests of Hypotheses To test hypothesis one, I have to find out whether the average amount of discretionary accruals changed in the period prior to the enactment of the Wta. First, I will present a per year time-trend of the mean and median discretionary accruals in

17 the period. This will indicate whether there is an increasing trend in discretionary accruals as is presumed by the Wta. The results of the discretionary accruals time-trend do not give conclusive evidence whether audit quality has actually declined prior to the enactment of the Wta since more factors are known to influence discretionary accruals. The following regression analysis will be performed to control for these factors: DAC it /A it-1 = α + β 1 TIME it + β 2 LTA it + β 3 BIG4 it + β 4 SOX it (4) + β 5 ASTD it + ε it Where: DAC it /A it-1 = absolute value of discretionary accruals in year t for firm i scaled by total assets in year t-1 for firm i; TIME it = 1,,13 corresponding to the years in year t for firm i; LTA it = log of total assets in year t for firm i; BIG4 it = takes the value of 1 if the auditor is a Big Four company, takes the value of 0 otherwise in year t for firm i; SOX it = takes the value of 1 from 2002 onwards if firm i is also listed in the US, takes the value of 0 otherwise; ASTD it = measures the accounting standard adopted by firm i in year t, either Dutch GAAP or IFRS. In equation (4), TIME has to be positive at the 95 percent significance level to accept hypothesis one. A positive relationship between TIME and discretionary accruals means that audit quality has decreased over time. Discretionary accruals are measured at their absolute value, since it is not the direction, but the magnitude of discretionary accruals that matters. They are scaled by total assets to provide an uniform distribution. If discretionary accruals are not scaled, discretionary accruals might correlate automatically with firm size. All other variables in the regression equation are control variables. The log of total assets is included to control for firm size. Lang and Lundholm (1993) find that large firms have incentives to disclose financial information more accurately to avoid litigation. Therefore, larger firms will have less discretionary accruals. Furthermore, Gu et al.(2005) suggests that relative to small firms, large

18 firms are more likely to be mature and operate in a steady state. Large firms also tend to be more diversified, with operating volatilities in different business sectors offsetting each other. This implies that large firms have lower operating volatility and therefore lower discretionary accruals. This hypothesized negative relationship between firm size, measured as the log of total assets, and discretionary accruals is found by Balsam et al. (2003), Lee and Mande (2003), and Gu et al. (2004). Therefore, I expect to find a negative relationship between discretionary accruals and firm size too. I control for auditor size, since prior studies (Teoh and Wong; 1993, Becker et al.; 1994, and Francis et al.; 1999) find that auditor size is positively associated with audit quality, and thus negatively associated with discretionary accruals. Therefore, I expect to find a negative relationship between discretionary accruals and companies audited by a Big Four auditor. The last two control variables are specific for the context of this study. Dutch firms that are cross-listed in the US had to comply with SOX in These firms will probably have less discretionary accruals while audit quality did not increase. Since Lobo and Zhou (2006) find that US listed firms applied more conservative accounting after the enactment of SOX. Therefore, I expect a negative relationship between SOX and discretionary accruals for Dutch firms too. I control for accounting standards since most firms use different accounting standard over time which may influence their amount of discretionary accruals. Because the European Union made IFRS accounting rules mandatory for all European listed companies from January 2005, Dutch companies changed their accounting principles within the sample period. Furthermore, there are also some companies who voluntary adopted IFRS prior to Previous studies suggest that the adoption of IFRS is associated with an increase in disclosure and a decrease in the number of accounting method choices (Leuz and Verrecchia; 2000, Ashbaugh and Pincus; 2001). So IFRS tends to be stricter than local GAAP, and therefore I expect discretionary accrual to decrease because of the adoption of IFRS instead of increasing audit quality. Therefore, I expect a positive relationship between Dutch GAAP and discretionary accruals. This study does not control for the level of debt, the variability of cash flow from operations, the growth of the company, and losses, which are found to influence discretionary accruals in previous studies. DeFond and Jiambalvo (1994) find that

19 managers of highly levered companies have incentives to make income increasing discretionary accruals to avoid debt covenants. Firms with more volatile cash flows will use discretionary accruals to smooth income (Gu et al.; 2005). According to Matsumoto (2002), companies with high long-term growth forecasts manage earnings upward to meet the expectations of analysts. Loss-making firms try to manage their earnings upward by making use of discretionary accruals (DeAngelo et al.; 1994, Burgstahler and Dichev; 1997). All factors just discussed measure a part of earnings management. This study s aim is to find whether audit quality has decreased, by not being able to constrain discretionary accruals for earnings management purposes. Consequently, variables which measure aspects of earnings management will not be included. To test hypothesis two, I will compare the mean and median discretionary accruals of the period prior to enactment of the Wta and the period after enactment of the Wta and test whether they are different from each other at the 95 percent significance level. In practice, this means that the mean and median discretionary accruals of the period will be compared with the mean and median discretionary accruals of 2006, since this is the only year of data available after the enactment of the Wta. Thereafter, I will perform a regression analysis with control variable to see whether the behavior of discretionary accruals is due to the enactment of the Wta, or has other causes. The following regression will be used: DAC it /A it-1 = α + β 1 WTA it + β 2 LTA it + β 3 BIG4 it + β 4 SOX it (5) + β 5 ASTD it + ε it Where: WTA it = takes the value of 1 when the observation is after the enactment of the Wta, takes the value of 0 otherwise in year t for firm i. WTA is the variable of interest in this regression. I expect it to be negatively related with discretionary accruals, since it is the aim of the Wta to improve the audit quality. To confirm, hypothesis two, WTA has to be negatively related to discretionary accruals with 95 percent significance. Furthermore, I expect the directions of the control variable to be the same as expected for equation (4).

20 3.4 Sample The sample will contain all Dutch companies listed on the Amsterdam Stock Exchange for the period Firm years for which not all required data is available will be left out of the sample. Although the Wta applies to listed and nonlisted companies, I use listed firms since data of their annual reports of 2006 is already available, as the Amsterdam Stock Exchange demands listed companies to publish their annual reports before April, whereas non-listed firms do not have a deadline. Firms in regulated industries (SIC 4000 to 4900) and firms in the financial sectors (SIC 6000 to 6900) will be excluded from the sample because their special accounting practices make the estimation of their discretionary accruals difficult. Firms using US GAAP as only accounting practice will also be excluded, to increase the homogeneity of the sample. To increase the accuracy of the tests, outliers will be eliminated by removing the 1 st and 99 th percentiles of the sample distribution. To increase the accuracy of the estimate of the discretionary accruals further, firms with less than six firm-year observations are also removed from the sample.

21 4. Results This section discusses the results of the tests developed in section three. In the first subsection, I discuss how I collected the data for the tests. In subsection two, the characteristics of the sample are summed up. In subsection three and four, I discuss the results on the tests for hypothesis one and two, respectively. Alternative tests for the hypotheses are developed and performed in subsection five. 4.1 Data Collection Data for Dutch firms listed on the Amsterdam Stock Exchange in the period is collected using Compustat Global. Since Cash Flow from Operations is not included in the Compustat Global database, it is collected manually using Annualreports.info. This is done manually since Hribar and Collins (2002) suggest that the balance sheet approach to calculate accruals data can lead to erroneously concluding that earnings management exists when no such activity is present. Information about the type of auditor and whether the firm is cross-listed in the US and thus whether the firm is SOX compliant is also collected from Annualreports.info. As can be seen in table 1, the total sample comprises 1,492 firm-year observations after the removal of observations with missing values. Subsequently, 79 observations of firms in regulated industries and 223 observations of firms in the financial sectors are deleted since their special accounting practices make the estimation of their discretionary accruals difficult. Hereafter, 41 observations of firms applying US Gaap are deleted to increase the homogeneity of the sample. To increase the accuracy of the estimate of the discretionary accruals, the first and ninety-ninth Table 1: Sample Size Number of firmyear observations Dutch firms listed on the Amsterdam Stock Exchange 1,492 Firms in regulated industries (SIC 4000 to 4900) -79 Firms in the finanicial sectors (SIC 6000 to 6900) -223 Firms applying US GAAP -41 Outliers -70 Firms with less than six firm-year obervations -43 Sample size used for tests of hypotheses 1,037

22 percentile of every variable necessary to estimate discretionary accruals are deleted. These percentiles comprise 70 firm year observations. At last, 43 observations of firms with less than six firm year observations are removed from the sample. This leads to a sample size of 1,037 firm year observations, which consists of 91 firms that are used to test the hypotheses. 4.2 Descriptive statistics Table 2 reports the descriptive statistics for the dependent and independent variables used in this study. Descriptive statistics are presented per year and in total, since it is the aim of this study to find out whether the magnitude of discretionary accruals has changed over time. Over the total sample period the mean and median absolute discretionary accruals are respectively 5.4% and 3.7% of lagged total assets. This implies that the distribution is somewhat skewed to the right, which is in line with the samples used by Gul and Krishnan (2002), Balsam et al. (2003), and Lee and Mande (2003). The magnitude of the absolute discretionary accruals is however not in line with these Table 2: Descriptive statistics DAC /A t -1 LTA Dutch Year n mean median mean median BIG4 SOX GAAP IFRS % N/A 100% % N/A 100% % N/A 100% % N/A 98.6% 1.4% % N/A 98.7% 1.3% % N/A 97.5% 2.5% % N/A 97.6% 2.4% % N/A 95.1% 4.9% % N/A 92.8% 7.2% % 7.7% 94.9% 5.1% % 7.4% 95.1% 4.9% % 7.3% 96.3% 3.7% % 8.8% 3.7% 96.3% % 8% - 100% total 1, % 7.8% a 82.9% 17.1% a The average percentage of firms that adopted SOX is not based on the total sample period, but on the period, from the moment that SOX was enacted.

23 studies. The mean value of absolute discretionary accruals used in this study is around five percent points smaller than in the studies just referred to. This can be due to different sample periods, or different institutional settings since this study examines Dutch companies instead of US companies. The highest mean (median) absolute discretionary accruals are found in 1997 and 2002 (2003) with a value of 6% (4.6%) of lagged total assets. The lowest mean (median) absolute discretionary accruals are found in 1993 (1993) with a value of 3.8% (3%) of lagged total assets. The table provides evidence that the magnitude of absolute discretionary accruals grew during the first five years of the sample period. The mean (median) log of total assets for the total sample is (12.572). The magnitude of the log of total assets displays an increasing trend over the sample period. Over the whole sample period 93.3% of the firms are audited by a big-four auditor. The percentage of firms audited by a big-four auditor has declined over the sample period from 96% to 86.7%. Since Sox is enacted, 7.8% of the companies included in the sample have to comply with it. Furthermore, there is a clear shift from Dutch GAAP to IFRS, which is in line with EU regulation. Because the EU demands all European listed companies to use IFRS as basis of preparation instead of local GAAP from 2005 on. The 3.7%, which still uses Dutch GAAP as accounting standard in 2005 can be explained by companies that do not have their fiscal year end at December 31. Table 3 presents the correlation coefficients of the dependent and independent variables. The two variables of interest, TIME and WTA, do not have a significant relation with absolute discretionary accruals. This is contrary to my expectations, outlined in hypothesis one and two. Of the control variables, only LTA is related with absolute discretionary accruals as expected. This confirms the suggested relationship between LTA and discretionary accruals by Lang and Lundholm (1993) and Gu et al. (2005). The other control variables, however, do not display any significant relationship with absolute discretionary accruals. This is contrary to the findings of previous studies. The exact negative correlation between Dutch GAAP and IFRS is logically, since only these two accounting standards are applied by the companies under investigation. These companies do either apply Dutch GAAP or IFRS, therefore a perfect negative correlation exists.

24 Table 3: Correlation Coefficients DUTCH DAC /At-1 TIME WTA LTA BIG4 SOX GAAP IFRS DAC /At-1 1 TIME WTA *** 1 LTA *** 0.09 *** ** 1 BIG *** ** *** 1 SOX *** *** *** DUTCH GAAP *** *** *** * *** 1 IFRS ** *** *** * *** 1 *** 1 *, **, and *** denote significance at the 10%, 5%, and 1% levels (two-tailed) respectively.

25 4.3 Test of hypothesis 1 The first hypothesis of this study states that audit quality has declined in the Netherlands prior to the enactment of the Wta. Since audit quality cannot be measured directly, absolute discretionary accruals are used as proxy. Low discretionary accruals proxy for strong audit quality, while high discretionary accruals proxy for poor audit quality. Consequently, discretionary accruals should have grown in the period prior to the enactment of the Wta, to confirm hypothesis one. The bold lines in Figure 2 present the behavior of the mean and median absolute discretionary accruals over the time period In line with table 2, it can be observed that the magnitude of mean absolute discretionary accruals grew in the first five years of the sample period. This behavior corresponds to the findings of Gul and Krishnan (2005), who find higher values of mean and median absolute discretionary accruals in the time period relative to the time period in the US. Furthermore, two peaks of mean absolute discretionary accruals of 6% of lagged total assets can be observed in 1997 and These peaks do exactly match with two major economic events, the Asia crises and the disclosure of major accounting frauds, respectively. Between these two peaks mean absolute discretionary accruals remain constant at 5.7% of lagged total assets. After the second peak, mean

26 absolute discretionary accruals decrease. Since absolute discretionary accruals were the highest between these peaks, it can be argued that audit quality was at its lowest. The behavior of mean absolute discretionary accruals is, however, not representative for the whole sample, since median absolute discretionary accruals do not increase as much and in line with mean absolute discretionary accruals. This implies that the strong increase in mean absolute discretionary accruals is caused by some extreme observations. To get a better grasp on the behavior of absolute discretionary accruals over time, trend lines are added to the graph. Both trend lines have a positive slope which is in line with the expectation stated in hypothesis one. Because this positive trend in absolute discretionary accruals can also be influenced by other factors than audit quality, I will perform a regression analysis, which controls for other influences on discretionary accruals, to determine whether a positive association between time and absolute discretionary accruals really exists. As stated in equation (4), Time should be significantly positive related to absolute discretionary accruals, while controlling for firm size, auditor size, SOX compliance, and accounting standard. These variables are all known to influence the level of discretionary accruals, but do not proxy for earnings management. The results of the regression are presented in table 4. As can be seen in table 4, TIME is positively related to absolute discretionary accruals, as predicted by hypothesis one. This relation is in line with the findings of Gul and Krishnan (2002), Jain et al. (2003), and Cohen et al. (2004), who all suggest that audit quality declined recently. However, the magnitude of this relation is very small; the magnitude of absolute discretionary accruals increases by 0.1% of lagged total assets per year. This small magnitude might be due to the fact that time is not directly related to absolute discretionary accruals, but proxies for underlying events. Furthermore, firm size is negatively related to discretionary accruals. This is in line with the suggested relationship between LTA and discretionary accruals by Lang and Lundholm (1993) and Gu et al. (2005). The other control variables do, however, not relate to absolute discretionary accruals as predicted by previous studies. This implies that large audit firms do not better detect earnings management than small audit firms in the Netherlands. This is opposite to the findings of Francis et al. (1999), who find a negative relation between absolute discretionary accruals and big-six auditors in the US. Furthermore, SOX does not have a positive negative

27 Table 4: Regression of absolute discretionary accruals on time and various control variables DAC it /A it -1 = α + β 1 TIME it + β 2 LTA it + β 3 BIG4 it + β 4 SOX it + β 5 ASTD it + ε it Predicted Independent variables Direction Coefficients t -statistic Intercept N/A 0,114 8,018 *** TIME + 0,001 2,743 *** LTA - -0,007-7,209 *** BIG4-0,008 1,059 SOX - 0,005 0,441 Dutch GAAP + 0,006 1,063 Adjusted R square 0,056 F-Statistic 12,323 *** # of observations 962 *** denotes significance at the 1% level (two-tailed). impact on the amount of absolute discretionary accruals of cross-listed Dutch firms in the US. This is contrary to the findings of Lobo and Zhou (2006), who find that the amount of discretionary accruals for US firms declined after the enactment of SOX in the US. This could be due to the fact that the PCAOB uses its power less abroad than within the US. At last, Dutch GAAP does not produce more absolute discretionary accruals than IFRS does, which is not in line with Leuz and Verracchia (2000) and Asbaugh and Pincus (2001), who suggest that IFRS is more severe than local GAAP. The fact that IFRS is only recently introduced in the Netherlands and accountants are not yet acquainted to it might explain this finding. So the different behavior of the control variables in this study relative to studies carried out in other countries could be due to difference in the Dutch environment. In summary, hypothesis one, which posits that audit quality has declined over time, is accepted since there is significant evidence that discretionary accruals increased during the period First, figure 2 provides a positive trend line over the sample period. Second, equation (4), which controls for other possible explanations of the growth of discretionary accruals over time, still finds a positive relation between absolute discretionary accruals and time.

28 4.4 Test of hypothesis 2 Hypothesis two posits that audit quality has increased after the enactment of the Wta. Once more, absolute discretionary accruals are used as a proxy for audit quality. Therefore, hypothesis 2 will be accepted if there are less absolute discretionary accruals after the enactment of the Wta than prior to enactment of the Wta. Table 5: Discretionary accruals: vs t -statistic mean median In table five, the mean and median values of absolute discretionary accruals of the time period, are compared with the mean and median absolute discretionary accruals of The time period is the time period prior to enactment of the Wta, where 2006 is the only year with available data after enactment of the Wta. As can be seen in table five, both mean and median absolute discretionary accruals are higher for the pre-wta time period, however, both t-statistics are insignificant. Therefore, no difference exists between the amounts of absolute discretionary accruals in the pre-wta period relative to the post-wta period. The fact that there is no difference between both periods does not have to mean that audit quality has not changed after enactment of the Wta. It can also be due to a change in some other variable that influences discretionary accruals. As stated in equation (5), WTA should be significantly negative related to absolute discretionary accruals, while controlling for firm size, auditor size, SOX compliance, and accounting standard. As mentioned before, these variables are all known to influence the level of discretionary accruals, but do not proxy for earnings management. The results of this regression are presented in table 6. As can be seen in table 6, WTA is not related to absolute discretionary accruals. Therefore the Wta is not associated with more severe accounting regulation, as Leuz et al. (2003) and Maijoor and Vanstraelen (2006) find that more severe accounting regulation is associated with low discretionary accruals. Furthermore, Vergoossen and Wallage (2004) find the policies of SOX and Wta very similar, its implications are very different. Whereas Jain et al. (2003), Cohen et al. (2004), and