The changes of earnings persistence in the pre- and post-Sarbanes-Oxley periods

Hele tekst

(1)MSc Accountancy & Control, Faculty of Economic and Business, University of Amsterdam. The changes of earnings persistence in the pre- and post-Sarbanes-Oxley periods. Student: Hyang Jung Kim Student Number: 10741631 Supervisor: Dr. W.H.P. Janssen Education: MSc Accountancy and Control, Accountancy track.

(2) Statement of Originality This document is written by student Hyang Jung Kim who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.. 2.

(3) Abstract This study describes how earnings persistence has changed in the post-SOX period. First, I expect that management and auditors will become more conservative because SOX increases their legal responsibilities. I addition, I expect that management will promote more real earnings management than before because SOX makes accrual-based earnings management more costly. The empirical research shows that SOX increases earnings persistence in spite of the increased responsibilities. Big N auditors are negatively related to earnings persistence in the preSOX period. However, the effect of Big N auditors is no more negative in the post-SOX period. Therefore, earnings persistence is higher when the firm’s auditor is a Big N auditor in the post-SOX period. Moreover, The R-squared value of the model is improved from 20% to 84% in the post-SOX period. This implies that investors can more reliably estimate the firm’s future performance based on the current information in the post-SOX period. I propose 2 explanations for the research findings. First, auditors acquire better understanding of clients through internal control system over financial reporting. Therefore, they can reduce their doubt about the financial statements which are prepared by management. The other possibility is that the difference between Big N auditors and non-Big N auditors becomes smaller in the post-SOX period. The knowledge of Arthur Andersen might be spread into non-Big N auditors through the former employees. In addition, the decreasing market share of Big N auditors might cause the indifference between Big N auditors and non-Big N auditors.. Key words Earnings persistence, Sarbanes-Oxley (SOX) act, Conservatism, Conservative accounting, Conservatism, Auditor conservatism, accruals, accrual-based earnings management, real earnings management, Big N auditors, and non-Big N auditors 3.

(4) Table of Contents 1. Introduction......................................................................................................................................... 5 1.1. Background............................................................................................................................... 5 1.2. Research questions.................................................................................................................... 8 1.3. Motivation ................................................................................................................................ 8 2. Literature reviews and Hypotheses development ............................................................................... 9 2.1. The adoption of Sarbanes-Oxley Act........................................................................................ 9 2.2. The conservative accounting in the post-SOX period ............................................................ 10 2.2.1. The conservative accounting ........................................................................................... 10 2.2.2. The auditor conservatism ................................................................................................ 12 2.3. Earnings .................................................................................................................................. 14 2.3.1. Earnings, accrual component and cash flow component................................................. 14 2.3.2. The quality of earnings and earnings persistence............................................................ 14 2.4. Earnings management............................................................................................................. 15 2.4.1. The incentives of earnings management ......................................................................... 15 2.4.2. The relationship between accrual-based and real earnings management ........................ 17 2.5. Hypotheses development ........................................................................................................ 19 3. Research design ................................................................................................................................ 20 4. Evidences .......................................................................................................................................... 21 4.1. Samples................................................................................................................................... 21 4.2. Results .................................................................................................................................... 24 5. Conclusion ........................................................................................................................................ 29 References ............................................................................................................................................. 31. 4.

(5) 1. Introduction 1.1. Background Early in 2000, there were numerous audit scandals such as Enron, WorldCom and Xerox. A number of prominent companies turned out houses of cards which were built on the fraudulent financial reporting. The continuous audit scandals devastated investors’ confidence in financial reporting and accounting professionals. The damaged reputation even caused the collapse of Arthur Andersen. Thereafter, the Sarbanes-Oxley (SOX) act was created to bring back investors’ confidence in financial reporting and accounting professionals in 2002. Regulators decided to impose more regulations and responsibilities of management and auditors than before. It seems that regulators consider the decision usefulness of accounting information can be improved by strict rules and requirements. Therefore, SOX consists of various requirements for both management and auditors. For example, SOX 302 and 906 require CEO and CFO’s certifications and responsibilities for financial reporting. The requirements increase the legal responsibilities of management. In case of the error or fraudulent financial reporting, management can be even put to jail sentences. In addition, SOX 404 requires auditors to express their opinion on the effectiveness of internal control system over financial reporting. This makes auditors more cautious about the errors or frauds in financial reporting than before because auditors are more exposed to the risks. In addition, auditors become more cautious to protect themselves from the litigations risks. The SOX requirements raise the legal responsibilities for both management and auditors. This is why the conservative accounting has increased in the post-SOX period (Lobo and Zhou, 2006). Even though SOX is created to enhance the quality of financial reporting, the argument about the costs and the benefits is not settled yet. There have been fierce debates regarding the advantages and the disadvantages of SOX. According to Glassman (2005), SOX generates huge incremental costs such as the installment cost of internal control system over financial reporting, the increased audit fee for the internal audit and the financial audit, the bureaucracy cost and the slow decision making processes. Glassman (2005) criticizes that the benefits of SOX does not outweigh the costs. However, there are lots of studies which contend that the benefits of SOX exceed the related costs. According to previous studies, SOX creates cautious atmosphere on accrual-based earnings management and restores investors’ confidence (Coates, 2007; Cohen et al., 2008). However, the benefits of SOX are difficult to be calculated into monetary value because the benefits are basically from the prevented errors or fraudulent financial reporting. Therefore, lots of researches focus on accrual component to verify the benefits of SOX such as the level of the total (discretionary) accruals (Ashbaugh-Skaif et al., 2008; Chambers et al., 2011), the level of accrual-based earnings management (Cohen et al, 2008) and 5.

(6) the quality of accruals (Chambers et al., 2011) The conservative accounting is one of the most important interests of academia, regulators, and accounting professionals. Therefore, a number of prior studies try to define and verify the advantages and the disadvantages of the conservative accounting. For example, Basu (1997) defines the conservative accountings (the conservatism) as ‘the more timely recognition in earnings of bad news regarding future cash flows than good news’ (p. 33). In the accounting professional world (Lu and Sapra, 2009), the conservative accounting means that auditors will require downward adjustments when they are in doubts about the financial statements prepared by management. The conservative accounting protects auditors from the litigation risks because auditors are more likely to be sued when the firm’s asset and earnings are overstated than understated (St. Pierre and Anderson, 1984; Lys and Watts, 1994, Chung et al., 2003). The requirements of SOX increase the legal liabilities for both management and auditors and make them prefer the conservative accounting in the post-SOX period. In addition, Watts (2003) contends that the outcomes of the conservative accounting are more desirable to all the relative parties such as investors, management, auditors and regulators. Furthermore, auditors show asymmetric monitoring efforts between income-increasing accruals and income-decreasing accruals in financial reporting (Kim et al., 2003; Chung et al., 2003) While management has both income-increasing and income-decreasing incentives according to the private purpose (Healy, 1995), auditors do not have the distinctive income-increasing intention comparing to management. Auditors suffer higher litigation risks when earnings and net assets are overstated than understated (St. Pierre and Anderson, 1984; Lys and Watts, 1994, Chung et al., 2003), therefore auditors more monitor income-increasing accruals carefully than income-decreasing accruals. In general, auditors are divided into Big N auditors and non-Big N auditors. Big 5 auditors became Big 4 auditors by the collapse of Arthur Andersen in 2002. The deep pocket theory explains why Big N auditors are regarded as more conservative auditors than non-Big N auditors. Big N auditors bear higher litigation risks and reputation risks than non-Big N auditors, because complainants may think that Big N auditors have greater monetary resources to pay in case of litigations. Previous researches suggest the decline of accrual-based earnings management due to the SOX adoption. Cohen et al. (2008) contend that accrual-based earnings management has significantly decreased in the post-SOX period. Management thinks that accrual-based earnings management becomes more costly in the post-SOX period. Management switches the way of earnings management from accrual-based earnings management to real earnings management, since management wants to 6.

(7) avoid the risks from SOX. Moreover, Chi et al. (2011) contend that firms engage more real earnings management when accrual-based earnings management is restricted by high quality auditors. They assume that higher quality auditors are city-level industry experts or Big N auditors, because those auditors have better understanding about the clients. These researches suggest that accrual-based earnings management and real earnings management are substitutes. Therefore, management wants to use more real earnings management in the post-SOX period. Accrual-based earnings management influences the timing of recognition in the financial statements. The overall effect of accrual-based earnings management would be smaller than real earnings management because accrual-based earnings management do not affect the actual cash flow. However, real earnings management can cause greater long-term costs to stakeholders because real earnings management affects the actual cash flow (Chi et al., 2011). However, the effect of real earnings management on earnings persistence has been rarely investigated until now in the academia. One possibility is that real earnings management is closely related to the business decisions, therefore it is hard to distinguish real earnings management from the business activities. In this reason, the effect of real earnings management on earnings persistence is not clearly discovered. This study provides the opportunity to see the impact of SOX on earnings persistence. According to the finding of Sloan (1996), investors show great dependence on earnings number for their economic decisions. Investors seem to estimate the future of firm’s performance based on the current earnings information. Sloan suggests that investors often face difficulties to understand accrual component in financial reporting. If earnings persistence increases in the post-SOX period, it means the improved decision usefulness of earnings to investors. Therefore, this study will provide the evidence how the decision usefulness of earnings has changed in the post-SOX period. Furthermore, the findings will contribute the debate of SOX usefulness in terms of the cost/benefit approach. However, this study has some limitations. First, this study is designed to see the effects of SOX and Big N auditors on earnings persistence for all the population. The study results do not verify the effect of real earnings management directly. Therefore, this calls future studies with regard to the impact of real earnings management. Moreover, the effects of SOX and Big N auditors might be different from the group which has a strong intention of earnings management. The other limitation is that the study deals with the aggregated effect on earnings persistence. Cash flow component and accrual component show differences in many ways. The effect of SOX and Big N auditors would be different to each component. Therefore, it would be interesting to verify the direct effect of each component on earnings persistence. 7.

(8) 1.2. Research questions This study is to verify how the decision usefulness of earnings to investors has changed due to the adoption of SOX. If earnings have a stronger predictability of the future performance in the post-SOX period, the decision usefulness of earnings for investors will increase. Therefore, the research questions are the changes of earnings persistence in the pre and the post-SOX period.. 1.3. Motivation SOX was created to restore investors’ confidence in financial reporting and to enhance the decision usefulness of accounting information. However, the debate about the costs and the benefits of SOX is still going on. Due to the huge direct and indirect costs of SOX compliance, related parties show the concern about whether the benefits of SOX outweigh the costs. Various researches have dealt with this debate until now. Prior researches pay attention how accounting quality has been changed in the post-SOX period when they are looking for the benefits of SOX. Generally, accounting quality is measured by the level of total (discretionary) accruals (Ashbaugh-Skaif et al., 2008; Chambers et al., 2011). The researchers assume that the lower level of accruals indicates higher audit quality, since management’s opportunistic behaviors are effectively restricted. They contend that the lower level of earnings management in the financial statements generates the higher decision usefulness of financial reporting. While academia, regulators and auditors show great interests in the changes of accrual component, investors’ understanding of accrual information is questionable. It seems that investors do not fully understand accrual component in the financial statements. Instead, they heavily rely on earnings number when they expect the future of firm’s performance (Sloan, 1996). Therefore, investors can reliably estimate the future performance based on the current earnings, if earnings persistence is high. Despite the importance of earnings persistence, there are few studies which focus on the direct relationship between SOX and earnings persistence. This research will provide the direct empirical evidence on how SOX affects the decision usefulness of earnings for investors in terms of earnings persistence. In addition, this study allows us to verify the benefits of SOX from the investors’ position. This will be a more intuitive indicator of the positive/negative effect of SOX for investors.. 8.

(9) 2. Literature reviews and Hypotheses development In the following sections, I will explain the background of the Sarbanes-Oxley Act and the conservative accounting. Then I will explain the role of earnings and earnings components. Lastly, I will describe the changes of earnings management and the relationship between accrual-based earnings management and real earnings management. After the careful explanations of the related ideas, I will develop the hypotheses regarding the research questions, the changes of earnings persistence in the pre and the post-SOX period.. 2.1. The adoption of Sarbanes-Oxley Act Investors make their economic decisions with the information from various sources. Financial reporting is one of the most important sources for investors. Therefore, management and auditors are required to provide high quality accounting information to investors. However, the confidence in accounting information and accounting professionals is severely damaged through the numerous accounting scandals early in 2000. Regulators decided to impose heavier responsibilities of management and auditors on financial reporting after the scandals. Regulators assume that strict rules and regulations of financial reporting will help to rebuild the investors’ confidence. This is why SOX consists of various requirements. According to the requirements, management must implement internal control system over financial reporting. Furthermore, management is responsible for the effectiveness of internal control system. Auditors are also required to express their opinion on the effectiveness of internal control system over financial reporting in addition to financial reporting. If management and auditors fail to meet the requirements of SOX properly, they could get financial penalties and/or jail sentences in the post-SOX period. The requirements have significantly increased management and auditors’ risks from the error or fraudulent financial reporting. The requirements of SOX generate various compliance costs. Not only the direct cost of implementing and maintaining internal control system over financial reporting, firms need to bear other indirect costs such as the increased audit fee and the slow decision making process (Glassman, 2005). Due to the huge costs of SOX, the related parties show concern about whether the benefits of SOX outweigh the costs. The debate about the costs and the benefits of SOX is still going on. Even though negative critics of the costs, it is generally believed that SOX generates more positive aspects than negative ones. For example, Dorado (2005) contends that SOX is playing an important role in 9.

(10) restricting frauds in financial reporting. Pollock (2005) argues that the costs of SOX are lower than the costs of the accounting scandals. However, the benefits of SOX are hardly calculated into monetary value because the benefits are from preventing the error or fraudulent financial reporting. Therefore, positive effects of SOX are often measured by the level of accrual-based earnings management and accrual quality. For example, Cohen et al. (2008) contend that SOX has decreased accrual-based earnings management in the post-SOX period. Chambers et al. (2011) argue that accrual quality and accrual persistence have been increased. Prior SOX researches show great interests in accrual component more than cash flow component in the financial statements. Cash flow component has concrete criteria such as cash payment and receipt. Therefore, cash flow component does not need management’s discretion for the recognition. However, management needs to estimate accrual component based on the current information. This gives a room for earnings management which is called accrual-based earnings management. In this reason, prior SOX researches pay attention to how accrual component has changed and how accrual-based earnings management has been restricted in the post-SOX period. It will be explained further in the ‘Earnings’ section.. 2.2. The conservative accounting in the post-SOX period 2.2.1. The conservative accounting Regulators decided to impose heavier responsibilities to restore investors’ confidence in financial reporting and accounting professionals after the accounting scandals. The increased responsibilities promote the conservative accounting in various ways because the conservative accounting protects management and auditors from the litigation risks. The conservative accounting is important in financial reporting. Basu (1997) explains the conservative accounting or the conservatism as ‘the more timely recognition in earnings of bad news regarding future cash flows than good news’ (p. 33). This asymmetric recognition of losses and gains implies the understatement of nets assets and earnings through accrual component. The study shows that negative unexpected returns indicate a greater increase in the timeliness of earnings over cash flows than positive unexpected returns. Basu points out that losses are recognized in earnings through accruals before the realization, while gains are deferred until the realization. Unrealized losses reduce the current earnings but do not decrease the current cash flows. On the other hand, unrealized gains increase neither the current earnings nor the current cash flows. Thus, negative earnings’ change will show more extreme reversals than positive earnings’ change. In addition, he also addresses that the 10.

(11) conservatism has increased over time. The increase of the conservatism is directly linked to the increase of auditors' legal responsibilities e.g. the adoption of class suit. Feltham & Ohlson (1995) investigate the conservative accounting in terms of accountingbased valuation. They distinguish financial asset and operating asset based on the different attributes of valuation. For example, a book value of the financial asset is equal to the market value and it is easy to observe the value from the market directly. Whereas, a book value of operating asset is not the same to the market value. It is difficult to observe the value from the market. Therefore, Feltham & Ohlson characterize the conservative accounting as an expectation that the book value of operating asset is reported less than the market value in the long run. Watts (2003) focuses on the benefits of the conservative accounting. He argues that the conservative accounting is relevant to all the stakeholders such as investors, management and auditors. The conservative accounting increases the timeliness and the verifiability of accounting information. Therefore, debt holders can protect themselves from the investors’ risky investment behaviors, as the conservative accounting guarantees the lower boundary of net assets in case of liquidation. In addition, investors can use the conservative accounting to restrict management’s opportunistic behaviors for the private gains such as bonus compensation and stock granting, since the conservative accounting reduces the room for earnings management. With regard to the litigation cost, management and auditors are more likely to be sued when earnings and net assets are overstated than understated. Consequently, both management and auditors can lower the risks of litigation by the conservative accounting. Therefore, Watts argues that the current trend of mark-to market approach does not enhance the value of stakeholders. Therefore he contends that the applying of the conservative accounting should be expanded. The conservative accounting is doing a great role in the financial reporting regardless of the SOX adoption. However, SOX surely reinforces the conservative accounting practice. SOX consists of various requirements of management and auditors. For example, SOX 302 and 906 require CEO and CFO’s certification on internal control system over financial reporting. The requirements have increased the legal responsibilities of management. Management may have imprisonment as well as financial penalties for the error or fraudulent financial reporting. SOX 404 requires that auditors have to express their opinion on the effectiveness of internal control system over financial reporting as well as the opinion on the financial statements. Those SOX requirements increase the legal responsibilities of management and auditors in the post-SOX period. Along the increased legal responsibilities, management and auditors face greater risks from financial reporting. Therefore, increased responsibilities make management and auditors more conservative in the post-SOX period. This might 11.

(12) implicate that the role of the conservative accounting becomes greater in the post-SOX period.. 2.2.2. The auditor conservatism The financial statements are management's primary duty. Therefore, management is responsible for adopting accounting policies and establishing and maintaining internal control system over financial reporting. On the other hand, the responsibility of auditors is to express their opinions on the effectiveness of internal control system over financial reporting and the appropriateness of the financial statements (American Institute of Certified Public Accountants, 1972). As mentioned in the introduction, management and auditors show different incentives with regard to earnings. Management wants to increase or decrease earnings depending on its private purpose. Healy (1995) explains earnings management based on the possibility of bonus receipt. Whereas, auditors do not have distinctive intentions to increase earnings, since generally audit fees are not dependent on earnings number itself. In addition, auditors suffer higher risks when earnings and net assets are overstated than understated (St. Pierre and Anderson, 1984; Lys and Watts, 1994, Chung et al., 2003). Therefore, auditors more carefully monitor income-increasing accruals than income-decreasing accruals to protect themselves. This is called the auditor conservatism (Chung et al., 2003) The collapse of Arthur Andersen was a big shock to auditors. Auditors realize that just one accounting scandal could make them out of business, even if they are one of Big N auditors. The risks from financial reporting make auditors prefer conservative (income-decreasing) accounting practices (DeFond and Jiambalvo, 1993; Watts, 2003). Auditors show asymmetric monitoring efforts between income-increasing accruals and income-decreasing accruals. Auditors more carefully monitor incomeincreasing accruals, since auditors face higher litigation risks when earnings and net assets are overstated than understated (St. Pierre and Anderson, 1984; Lys and Watts, 1994, Chung et al., 2003). Heninger (2001) investigates the association between auditor litigations and abnormal accruals. He finds that the possibility of auditor litigations is positively related to income-increasing discretionary accruals in reported earnings, since the high level of income-increasing discretionary accruals is a manifestation of earnings management. This finding implies that auditors will be exposed to higher litigation risks if auditors fail to deter upward earnings management. Management evaluates accruals before the realization. Therefore, there is always an argument whether the recognition of accruals is appropriate or not. Francis et al., (1999) indicate that auditors face more uncertainty from high-accruals firms, since auditors cannot objectively verify the 12.

(13) appropriateness of accrual estimations. Therefore, auditors control their threshold to issue the modified audit reports to protect themselves from the distressed assets and/or going concern problems. The study shows that high-accruals firms are more likely to have the modified opinion from their auditors. Besides, income-increasing accruals are more likely to cause the reporting conservatism than income-decreasing accruals. Becker et al. (1998) show that Big 6 (now Big 4) auditors report the lower level of discretionary accruals than non-big 6. According to their study, the significant low level of discretionary accruals implies that Big 6 auditors efficiently constrain earnings management. The lower level of discretionary accruals means the less room for earnings management. In this regard, the level of discretionary accruals is often used to measure audit quality and auditor quality as well. Kim et al. (2003) extend the study of Becker et al. (1998). They distinguish incomeincreasing discretionary accruals and income-decreasing discretionary accruals. The study indicates that Big 6 (now Big 4) auditors are more effective to screen income-increasing discretionary accruals than non-Big 6 auditors. However, Big 6 auditors are less effective to screen income-decreasing discretionary accruals than non-Big 6. The authors explain that Big 6 auditors are more conservative than non-Big 6, therefore Big 6 auditors show asymmetric monitoring efforts to protect themselves from the litigation risks and/or reputation risks. According to the findings, Big 6 auditors are likely to accept income-decreasing accruals even if those accruals are intended to shift earnings to the future period. It is generally believed that Big N auditors are more conservative than non-Big N auditors (Kim et al., 2003; Chung et al., 2003). The deep pocket theory is often used to explain why Big N auditors are more conservative. ‘Deep pocket’ means extensive financial wealth or resources. If someone is seen as a deep pocket, he or she is likely to be a defendant in a law suit simply because he or she has more money to pay for complaints. Likewise, Big N auditors face higher litigation risks because they are perceived as ‘deep pockets’ who have more resources to pay settlements in case of litigation (Arthur Andersen et al. 1992). In addition, the damaged reputation of Big N auditors is much bigger than expected. As we can see the case of Enron, the collapse of Arthur Andersen was started from one office in Houston. The fault of Houston office brought about the whole company down. Therefore, a lot of previous studies assume that Big N auditors are more conservative auditors than non-Big N auditors. In conclusion, the adoption of SOX affects accrual component. First of all, the amount of (discretionary) accruals has been reduced in the post-SOX period. Due to the increased legal responsibilities, auditors prefer less room for accrual-based earnings management. Therefore, auditors 13.

(14) exert greater effort to reduce the level of (discretionary) accruals. Furthermore, there are asymmetric changes between income-increasing and income-decreasing accruals. Even though the absolute level of total accruals is reduced, income-decreasing accruals are less reduced comparing to incomeincreasing accruals.. 2.3. Earnings 2.3.1. Earnings, accrual component and cash flow component Current accounting system adopts accrual basis, not cash basis. Therefore, earnings consist of accrual component and cash flow component. Sloan (1996) investigates earnings’ components and their attributes. His study shows the different earnings persistence of each component. Due to the reversing property of accruals, earnings performance attributable to accrual component exhibits the lower persistence than cash flow component. He explains that the lower earnings persistence of accrual component is due to the greater subjectivity of estimation in recognition. He also suggests that the lower earnings persistence of accrual component is due to its reversing property. In addition, when accrual component is bigger, its reversal is more extreme. Scott et al. (2005) extend the research of Sloan (1996). They show that less reliable accruals attribute to lower earnings persistence. The lower earning persistence leads significant security mispricing since investors do not fully understand accrual component in financial reporting. Both studies note the limited understanding of investors. Investors make earnings expectations based on the current earnings. Although accrual component shows greater subjectivity in the recognition and reversal in the future, investors do not fully reflect them into stock prices. Investors often face difficulties to fully understand the higher persistence of cash flow component and the lower persistence of accrual component. On the other hand, hedge funds show the sophisticate understanding of accrual component in the financial statements. They show different investment behaviors from investors. For example, Hedge funds sell the stocks which have higher incomeincreasing accruals and buy stocks which have lower income-increasing accruals.. 2.3.2. The quality of earnings and earnings persistence Before Ball and Brown’s research (1967), people believed that earnings number did not provide any critical information to investors because earnings were regarded as an aggregated number with mixed information. Their empirical research shows that earnings and stock returns are significantly related. 14.

(15) This suggests that information in earnings is important for investors when they make economic decisions e.g. buying and selling stocks. Thereafter, a lot of researches pay attention to the role of earnings and earnings quality (Dichev et al., 2013; Dechow, 1994; Sloan, 1996). Many studies try to define and measure earnings quality. Prior researches suggest the possible attributes of earnings quality such as sustainable earnings with actual cash flow (Dichev et al., 2013), persistent earnings (Dichev and Tang, 2009), the conservative accounting (Basu, 1997), various forms of bench mark beating (Burgstahler and Dichev, 1997), the low level of abnormal accruals (Jones, 1991) and the extent to which accruals map into actual cash flows (Dechow and Dichev, 2002). Earnings persistence is an important attribute of earnings quality because investors make future expectation based on the current earnings (Sloan, 1996). Dichev et al. (2013) also argue the importance of earnings persistence, since it enables investors to reliably estimate firm’s valuation and future performance based on the current earnings. According to Dichev et al., CFOs and standard setters believe that high quality earnings are sustainable and repeatable, consistent with reporting choices, supported by actual cash flows, absent of one-time items and estimated by long-term basis. Among them, sustainable and repeatable earnings can be defined as earnings persistence which is useful for firm’s valuation and predicting future earnings. The purpose of SOX is to provide more relevant and reliable accounting information for investors. Therefore, earnings quality is an important concern of SOX. In other words, if SOX is successful to improve the decision usefulness of earnings for investors, there must be a positive change in earnings persistence.. 2.4. Earnings management 2.4.1. The incentives of earnings management Earnings management is the accounting practice which management intentionally chooses to bias accounting information for the private purpose. Management has various incentives to manage earnings. Healy (1985) studies how management chooses accounting practice to increase the bonus compensation when the bonus is determined by earnings-based performance. Opposed to general belief that management wants to increase earnings only, management wants to increase or decrease earnings according to the situation. For example, management increases earnings when management can increase its bonus in the current period. If earnings are lower than the bonus threshold or higher 15.

(16) than the bonus cap in the bonus scheme, management tries to transfer the current earnings to the next period through income-decreasing discretionary accruals. Transferred earnings increase the possibility of meeting or beating the bonus threshold in the next period. Healy’s study displays that management can choose accounting practices to increase or decrease earnings based on its self-interest. Burgstahler and Dichev (1997) provide the evidence of earnings management to avoid earnings decreases and losses. The study finds that unusually low rates of small decreases of earnings and small losses in the test period. On the other hand, there are unusually high rates of small increases of earnings and small positive gains in the test period. The authors explain the unusual phenomenon based on the stakeholders’ use of information-processing heuristics and the prospect theory (Kahneman and Tversky, 1979). According to the use of information-processing heuristics, stakeholders habitually make decisions based on heuristic cut-offs at zero change or the previous level of earnings. Therefore, management engages earnings management to decrease the transaction cost with its stakeholders. The prospect theory assumes that decision-makers judge gains and losses from a reference point. For example, decision-makers may evaluate the current earnings based on what extent to earnings increase from the prior earnings. The absolute level of earnings does not matter much in this case. Therefore, management tries to meet the prospect through earnings management. DeFond and Jiambalvo (1993) investigate the disagreement between management and auditors. Securities and Exchange Commission (SEC) requires the disclosures of auditor-client disagreement before the auditor change to make an alarm for investors. Management often changes its auditor before issuing the qualified opinion or the negative opinion. Thereafter, management tries to find a new auditor who will give the unqualified opinion. It is called ‘opinion shopping’. For example, an auditor may not agree the accounting policies which the management adopts not to violate debtcovenants. Then the auditor does not want to issue the unqualified opinion based on the disagreement. In this case, the management simply changes the current auditor and finds a new one for a clean opinion. Investors might think the financial statements are prepared appropriately. . This is why SEC requires the disclosure to protect investors from false impressions. This study suggests the conflicting reporting incentives between management and auditors. In conclusion, prior researches suggest that management has various incentives to manage earnings upward and downward as well. There is no absolute criterion for earnings management. Even high-earnings firms want to manage earnings if they expect the increase of earnings from a reference point. In addition, there are reporting conflicts between management and auditors. Auditors do not always agree with management’s accounting choices. This implicates the role of independent auditors against management in financial reporting. 16.

(17) 2.4.2. The relationship between accrual-based and real earnings management There have been numerous attempts to define earnings quality and to measure it objectively. Earnings quality is often measured by the level of earnings management. Prior researches suggest that highquality earnings mean the low level of earnings management. If the financial statements are less biased by the management’s opportunistic behaviors, the decision usefulness of earnings for investors would be higher. In other words, poorer-quality earnings do not reflect the firm’s true value, therefore, earnings will give false impressions to investors about the firm’s value (Lo, 2008). Generally, earnings management means accrual-based earnings management. However, there is one more earnings management called real earnings management. Accrual-based earnings management means the accounting choices that management intentionally makes to bias accounting information for its private gain. It only deals with the timing of recognition in the financial statements. In contrast, real earnings management means the actual business decisions that management intentionally makes to influence accounting information. For example, management wants to cut R&D investments to reduce expenses and increase earnings, or management wants to offer big price discounts or lenient credit terms to boost up sales. Even though real earnings management may adversely effect on the firm’s value in the long run (Roychowdhury, 2006; Cohen et al., 2008), it is difficult to distinguish from strategic business decisions. For example, management might estimate that the R&D investments are not profitable, therefore management does want to spend money any more. This is why real earnings management is hard to distinguish. In this reason, real earnings management is less investigated than accrual-based earnings management in the academia. Cohen et al. (2008) study the change of accrual-based earnings management and real earnings management in the pre and the post-SOX period. They investigate how the bench mark earnings are achieved through accrual-based earnings management and real earnings management. Their study shows the steady increase of accrual-based earnings management from 1987 to 2002. After the adoption of SOX in 2002, accrual-based earnings management decreases significantly. At the same time, the level of real earnings management steadily increases comparing to accrual-based earnings management. This switch from accrual-based earnings management to real earnings management indicates that they are substitutes. In conclusion, the SOX requirements make accrualbased earnings management more costly than before. Therefore, management prefers real earnings management in the post-SOX period in spite of the possible adverse effects on firm’s real value. Prior researches contend that accrual-based earnings management is effectively restricted by high quality auditors (Becker et al. 1998, Balsam et al. 2003). Chi et al. (2011) investigate the alternative relationship between accrual-based earnings management and real earnings management. 17.

(18) They examine the effect of high quality auditors on the choices of earnings management. According to the study, high quality auditors are defined as city-level industry experts and Big N auditors. In other words, they investigate the choices of earnings management if auditors have better understanding and knowledge about the client. Their study suggests that the high quality auditors more effectively restrict accrual-based earnings management than non-Big N auditors. However, high quality auditors make the firms more engage real earnings management when the firms have a strong intention to manage earnings. Along with SOX, auditors are more concerned with accrual-based earnings management. However, this may promote more real earnings management. Chambers et al. (2011) study how accrual quality has been changed in the pre and the postSOX period. They assume accrual quality as persistence of total accruals with short-term, long-term and financial accruals. While other researches pay attention to discretionary accruals, this study deals with the broad concept of accruals. They assume that accrual persistence is directly linked to earnings quality. Comparing to the pre-SOX period, accrual persistence has improved significantly in the postSOX period. In addition, the Big N auditors who are less independent from their client show the significant improvement of accrual persistence. Prior studies suggest that SOX generates 2 important changes of earnings persistence. First, SOX reduces the magnitude of total (discretionary) accruals because management and auditors become more conservative. Additionally, management prefers real earnings management more because accrual-based earnings management becomes more costly in the post-SOX period. The lower level of accruals is positively related to earnings persistence, since accrual component shows the lower earnings persistence than cash flow component (Sloan, 1996). Second, management uses more real earnings management in the post-SOX period. However, the effect of real earnings management on earnings persistence is not clear. Even though it is generally believed that real earnings management may cause adverse effects on the long-term firm value, this is rarely investigated in academia. It is both possible that real earnings management negatively or positively affect earnings persistence. For example, real earnings management would make earnings more volatile, if it disturbs the firm’s business activities temporally. On the other hand, earnings persistence would increase, if real earnings management destroys the firm value fundamentally. In conclusion, accrual-based earnings management and real earnings management are substitutes. In the post-SOX period, firms switch from accrual-based earnings management to real earnings management. The effect of low (discretionary) accruals is positively related to earnings persistence, however, the effect of increased real earnings management is not clearly discovered yet.. 18.

(19) 2.5. Hypotheses development SOX was created to restore investors’ confidence in financial reporting and accounting professionals. Prior researches suggest that SOX brings about a few significant changes of financial reporting, e.g., the decrease of accrual-based earnings management, the lower level of total (discretionary) accruals, the increase of accrual quality and the expansion of the conservative accounting practice. Due to these reasons, they argue that the decision usefulness of accounting information has improved in the postSOX period. Even though SOX tries to improve the decision usefulness of accounting information for investors, it is still controversial whether SOX is successful or not. Here, I pay attention to some arguments which may cause various effects on the decision usefulness of earnings, especially earnings persistence. According to Cohen et al. (2008), firms switch earnings management from accrual-based earnings management to real earnings management. Aggressive engagement in real earnings management could make firm’s value temporally volatile or fundamentally destroyed. In addition, the conservative accounting could decrease the level of total (discretionary) accruals. However, excessive conservative accounting practices could make earnings more volatile because of the preference of income-decreasing accruals which show more extreme reversals in the future (Kim et al., 2003; Sloan, 1996). Therefore, I predict 3 factors which cause different effects on earnings persistence based on the literature reviews. 1) The smaller amount of accruals will have a positive effect on earnings persistence since accrual component shows the lower earnings persistence than cash flow component. 2) The increased auditor conservatism is likely to cut income-increasing accruals and accept incomedecreasing accruals. This will have a negative effect on earnings persistence because incomedecreasing accruals show more extreme reversals than income-increasing accruals. In addition, Big N auditors will show the stronger preference to income-decreasing accruals. Therefore, the negative effect of auditor conservatism will be stronger in case of Big N auditors. 3) Real earnings management could make earnings more volatile as it disturbs firm’s business strategic decisions. At the same time, it could make earnings downward fundamentally. Therefore, I assume the aggregated effect of real earnings management not clear. The factors have different or unclear effects on earnings persistence in the post-SOX period. Therefore, I state the null hypothesis that there is no difference in earnings persistence between the pre and the post-SOX period. H1: Earnings persistence does not differ in the pre and the post-SOX period. 19.

(20) Big N auditors are regarded as more conservative auditors than non-Big Auditors because they face greater legal and reputation risks from financial reporting. At the same time, management will use more real earnings management when its auditor is a Big N auditor. Even though the effect of real earnings management is not clear, I predict that earnings persistence will decrease when the firm’s auditor is a Big N auditor. H2: Earnings persistence is lower when the firm’s auditor is a Big N auditor.. Lastly, I will test the aggregated effect of the SOX adoption and the role of Big N auditors on earnings persistence. The aggregated effect is not obvious now, but I expect the aggregated effect on earnings persistence will be stronger than Hypothesis 1 H3: The change in earnings persistence is stronger when the firm’s auditor is a Big N auditor in the post-SOX period.. 3. Research design H1:. Et+1= α0 + α1Et + α2SOX + α3Et*SOX + α4BigN + α5Loss_dummy + α6 Profitability + α7Total_Assets + α8ind∑Industry dummiest + α9year∑Year dummiest + εt+1. H2:. Et+1= β 0 + β1Et + β2BigN + β3Et*BigN + β4SOX + β5Loss_dummy + β6 Profitability + β7Total_Assets + β8ind∑Industry dummiest + β9year∑Year dummiest + εt+1. H2:. Et+1= γ0 + γ1Et + γ3SOX + γ4BIG + γ5Et*SOX + γ6Et* BigN + γ7SOX* BigN + γ8Et*SOX* BigN + γ9Loss_dummy + γ10Profitability + γ11Total_Assets + γ12ind∑Industry dummiest + γ 13year∑Year. dummiest + εt+1. Apart from the 2 independent variables, SOX and Big N auditors, I set a few control variables to control other effects on earnings persistence. I explain why these control variables are implemented in the models. Loss occurrence I expect that auditors become more conservative when firms record loss in earnings. In loss condition, the firms might be penalized for failing to meet the bench mark earning, e.g. zero earnings. Therefore, the firms have a strong intention of earnings management (Burgstahler and Dichev, 1997). Auditors 20.

(21) are aware of the risks of earnings management when they encounter negative earnings. Therefore, I make an assumption that the management’s intension of earnings management will be stronger when they are in loss condition. At the same time, auditors will be more cautious to earnings management. Profitability Similar to loss occurrence, firms may not have a strong intention of earnings management when the firms are profitable. In addition, auditors will be less conservative to profitable firms because auditors estimate the audit risk of profitable firms low. Therefore, I set the profitability as a control variable, too. Total Assets The level of total assets indicates the size of the firm. This is a common variable to control the size effect. In addition, likely to Loss occurrence and profitability, Total Asset also affects the auditors’ audit risk estimation. Fiscal year The economic situation during the test period was very volatile. The important changes were the Dotcom bubble (called also the internet bubble or the dot-com boom) and the collapse of the bubble. Roughly the Dot-com bubble covers from 1997 to 2000 and the collapse of bubble occurs from 1999 to 2001. Therefore, the whole pre SOX period is affected by the bubble and the collapse. Therefore, I set the fiscal in the model year to control the year effect. Industry classification Economic environment differs from industry to industry. For example, Dot-com bubble and its collapse differently affect each industry. Therefore, I set the industry classification to control the industry effect.. 4. Evidences 4.1. Samples Currently, Big N auditors are the 4 largest accounting firms in the world. Big N auditors were previously known as Big 8, Big 6 and Big 5. With the merge between Cooper & Lybrand and Price Waterhouse, Big 6 became Big 5 in 1998. In 2002, Big 5 became Big 4 because Enron scandal brought about the collapse of Arthur Andersen. Now, Big 4 are Ernest & Young, PwC, Deloitte and 21.

(22) KPMG. Therefore, Big N auditors mean Big 5 in the pre-SOX period and Big 4 in the post-SOX period. Except Arthur Andersen, the consistency of Big N auditors from 1998 to 2005 is the same. To compare the same period of the pre and the post SOX, I choose 4 years before SOX and 4 years after SOX. This setting will allow the balanced weight of the changes of earnings persistence. Therefore, the test period is from 1998 to 2001 for the pre-SOX period and 2002 to 2005 for the postSOX period.. Table 1 Sample Selection First Sample. 131,796. No earnings data. 25,038. No Assets data. 14. No auditor data. 7,577. Excluded industry classifications. 23,267. Fiscal year for 1996, 1997 ,2006 and 2007. 16,307. No next earnings data. 5,301. No previous assets data. 4,149. No Profitability data The observation of the final sample. 226 49,917. To verify the changes of earnings persistence, I perform a database research with North America companies from Compustat. Fundamental annual data are used to collect for the variables of the model. The first sample period is from 1996 January to 2007 June, because I need to consider the fiscal year, the next period of earnings and the previous assets information. First, I collect 131,796 data, and then I drop the data if they have no information of earnings, assets and auditor. After that, I exclude the firms which are regulated industries (SIC from 4000 to 4999), financial institutions (SIC from 6000 to 6999) and unclassified firms (SIC 9999). These firms are either heavily regulated by government or institutions. Therefore, the behaviors may differ from other firms in the sample. After that, I delete the data from the fiscal year of 1996, 1997, 2006 and 2007. Then I delete the data without earnings information of the next period and assets information of the previous period as well. Lastly, I delete also the 226 data which do not have profitability information. Those data have zero values of the pervious assets, because the zero values produce the null values of Profitability. In the end, I acquire 49,917 observations in the final sample. Table 1 describes the sample selection process. 22.

(23) Table 2 describes the important variables for the whole test period, the pre-SOX period and the post-SOX period. Table 2 indicates that the portion of Big N auditors decreases in the post-SOX period. It also shows that almost half of the observations record negative earnings.. Table 2 Sample description Whole Period Observation. Pre SOX Period. Post SOX period. 49,917. 26,142. 23,775. 69.1666. 43.66369. 97.20855. Min. -85,162.21. -85,162.21. -13,274.04. Max. 36,130. 22,071. 36,130. 84.51882. 37.40142. 136.3272. Min. -85,162.21. -85,162.21. -12,613. Max. 39,500. 17,720. 39,500. Big N auditors. 36,879. 20,560. 16,319. Non Big N auditors. 13,038. 5,582. 7,456. 1,878.367. 1,532.261. 2,258.93. Min. 0. 0. 0. Max. 750,507. 495,023. 750,507. 23,708. 12,584. 11,124. Earnings t. Earnings t+1. Asset. Mean. Mean. Mean. Loss occurrence. Table 3 Auditor composition for each fiscal year Pre SOX. Big N auditors. Non-Big N auditors. Total. Post SOX. 1998. 1999. 2000. 2001. 2002. 1,218. 1,322. 1,477. 1,565. 1,709. 1,765. 1,922. 19%. 20%. 22%. 24%. 28%. 29%. 33%. 5,257. 1,477. 5,110. 4,863. 4,458. 4,267. 3,935. 81%. 80%. 78%. 76%. 72%. 71%. 67%. 64%. 6,475. 6,652. 6,587. 6,428. 6,167. 6,032. 5,857. 5,719. 26,142. 2003. 2004. 2005. Total. 2,060 13,038 36%. 26%. 3,659 36,879. 23,775. 74% 49,917. 23.

(24) Table 3 describes how the final sample consists of auditors for each fiscal year. According to Table 2, the portion of Big N auditors decreases in the post-SOX period. Table 3 shows the details of the decreasing trend of Big N auditors. The portion of Big N auditors is continuously decreasing over the test period. Moreover the rate of decrease is accelerated after the SOX adoption. This result may suggest that the clients of Arthur Andersen are spread into not only Big 4 auditors but also non-Big 4 auditors after the collapse.. 4.2. Results For Hypothesis 1, I test the effect of SOX on earnings persistence. Table 4 describes the coefficient value, standard error, T value, P value and the 95% confidence intervals. The R-squared value of the model is 0.5247. This means that 52% of the next earnings can be explained by this model. According to the result, the effect of SOX on earnings persistence is significantly positive. The coefficient value of E_SOX is 0.593833 and the P value is less than 0.05. However, Big N auditors and Profitability are not significant control variables in the model. In conclusion, SOX is positively related to earnings persistence. Therefore, the null hypothesis is rejected. Table 4 The effect of SOX on Earnings persistence Earnings t+1 Earnings t. Coef.. Std. Err. T. P> ǀ t ǀ. 95% Conf. Interval. 0.3652697. 0.0047492. 76.91. 0.000. 0.3559612. 0.3745782. SOX. 53.67547. 11.75819. 4.56. 0.000. 30.62929. 76.72165. E_SOX. 0.593833. 0.0071758. 82.75. 0.000. 0.5797683. 0.6078976. BigN. 7.326206. 6.770764. 1.08. 0.279. -5.944574. 20.59698. Asset t. 0.0110695. 0.0002783. 39.78. 0.000. 0.10524. 0.0116149. Fiscal Year. -10.29302. 2.58794. -3.98. 0.000. -15.36541. -5.22063. Industry. -0.5989803. 0.1253853. -4.78. 0.000. -0.844737. -0.3532235. Profitability. -0.0015452. 0.023724. -0.07. 0.948. -0.489446. 0.0449542. Loss dummy. -27.77314. 5.950221. -4.67. 0.000. -3943564. -16.11064. 20620.03. 5175.02. 3.98. 0.000. 10476.93. 30763.13. Con. The finding suggests that SOX improves the decision usefulness of earnings for investors. According to Sloan (1996), investors heavily rely on earnings information to predict firm’s future 24.

(25) performance. If SOX improves earnings persistence, investors can have better expectations of firm’s future performance based on the current earnings. This finding contributes to the debate of the costs and the benefits of SOX, because the decision usefulness of earnings for investors improves in terms of earnings persistence. Until now, many SOX researches have focused on accrual component in the financial statements. They investigate total (discretionary) accruals, accrual persistence and accrualbased earnings management while researching for the benefits of SOX. However, the improved quality of accrual component is not directly linked to the decision usefulness of earnings for investors, since investors show the limited understanding of accrual component when they expect firm’s future performance. Contrary to accrual component, earnings persistence is a more intuitive criterion for investors. The increased predictability of firm’s future performance is directly increasing the benefits of investors. Therefore, the finding directly supports the benefits of SOX in an investors’ position.. Table 5 The effect of Big N auditors on Earnings persistence Earnings t+1 Earnings t. Coef.. Std. Err. T. P> ǀ t ǀ. 95% Conf. Interval. 0.6477542. -0.0512466. 12.64. 0.000. 0.5473103. 0.7481982. 14.26334. 7.226021. 1.97. 0.048. 0.1002578. 28.42643. -0.0657493. 0.0512536. -1.28. 0.200. -0.1662071. 0.0347084. SOX. 62.94251. 12.53824. 5.02. 0.000. 38.36742. 87.51759. Asset t. 0.174567. 0.0002852. 61.21. 0.000. 0.0168977. 0.0180157. Fiscal Year. -2.05535. 2.75777. -0.75. 0.456. -7.460611. 3.349911. -0.637139. 0.1337087. -4.76. 0.000. -0.8987845. -0.3746433. Profitability. -0.0028675. 0.0252991. -0.11. 0.000. -0.0524541. 0.0467192. Loss dummy. -36.53098. 6.352819. -5.75. 0.910. -48.98258. -24.07938. 4129.985. 5514.617. 0.75. 0.454. -6678.728. 14938.7. BigN E_BigN. Industry. Con. For Hypothesis 2, I expect that Big N auditors are negatively related to earnings persistence. Table 5 suggests that there is no significant effect of Big N auditors on earnings persistence because the P value of the model is bigger than 0.05. Therefore, Hypothesis 2 is rejected. This means that Big N auditors do not have a significant effect on earnings persistence through the test period which covers the pre and the post-SOX period.. 25.

(26) I propose 2 possibilities which may cause this result. The first one is the decrease of the Big N auditors’ market share. According to Table 2 and 3, the portion of Big N auditors is decreasing through the test period. The decreasing rate is accelerated after 2002. It is possible that the change of Big N auditors’ market share effects on earnings persistence differently in the pre and the post-SOX period. It is also possible that non-Big N auditors act similar to Big N auditors after the collapse of Arthur Andersen. For example, non-Big N auditors could change their audit risk strategy after the accounting scandals. At the same time, the knowledge of Arthur Andersen might be spread into nonBig N auditors by the previous Arthur Andersen employees. Therefore, I divide the sample into the pre-SOX period and the post-SOX period. Table 6 describes the big differences between the pre-SOX period and the post-SOX period. The first one is the explanatory power of the models. The R-squared value is 0.1954 in the pre-SOX period. It means that 80% of the next earnings cannot be explained by this model. Therefore, investors should consider other factors to predict the next earnings. However, the R-squared value in the post-SOX period is 0.8356. This means that 84% of the next earnings can be explained by the variables in the model. This implies that investors can make better future estimations with the current information in the post-SOX period. The predictability of the model is greatly improved. In addition, the effect of Big N auditors is significantly negative in the pre-SOX period. This means that the next earnings are less predictable when the firm’s auditor is a Big N auditor. Previously described, Big N auditors are more conservative than non-Big N auditors because they face greater litigation risks and reputation risks. Contrary to the pre-SOX period, the effect of Big N auditors is not significant any more in the post-SOX period, because P value is 0.574. Big N auditors do not affect earnings persistence negatively anymore. This may suggest that Big N auditors become less conservative. In conclusion, Hypothesis 2 is partially accepted. The earnings persistence decreases in the pre-SOX period when the firm’s auditor is a Big N auditor. However, the effect of Big N auditors is not significant in the post-SOX period. This means that there is no significant difference between Big N auditors and non-Big N auditors on earnings persistence in the post-SOX period.. 26.

(27) Table 6 The effect of Big N auditors in the pre and the post-SOX period Panel A: The effect of Big N auditors in the pre-SOX period Earnings t+1 Earnings t. Coef.. Std. Err. T. P> ǀ t ǀ. 95% Conf. Interval. 0.5581787. 0.0756208. 7.38. 0.000. 0.4099578. 0.7063996. -16.6341. 12.26392. -1.36. 0.175. -40.67205. 7.403854. -0.2226728. 0.0756842. -2.94. 0.003. -0.3710179. -0.0743277. Asset t. 0.0174372. 0.0005453. 31.98. 0.000. 0.0163684. 0.0185061. Fiscal Year. -13.03347. 4.470372. -2.92. 0.004. -21.79564. -4.271292. Industry. -0.4744827. 0.2170396. -2.19. 0.029. -0.8998922. -0.0490733. Profitability. -0.0029884. 0.036356. -0.08. 0.934. -0.0742481. 0.0682713. Loss dummy. -68.60307. 10.17414. -6.74. 0.000. -88.54495. -48.66119. 26123.73. 8938.608. 2.92. 0.003. 8603.568. 43643.89. BigN E_BigN. Con. Panel B: The effect of Big N auditors in the post-SOX period Earnings t+1 Earnings t. Coef.. Std. Err. T. P> ǀ t ǀ. 95% Conf. Interval. 0.9876043. 0.0482203. 20.48. 0.000. 0.8930895. 1.082119. 38.18493. 5.760733. 6.63. 0.000. 26.89395. 49.47633. 0.0270502. 0.0481721. 0.56. 0.574. -0.0673702. 0.0124705. 0.006567. 0.0002215. 29.65. 0.000. 0.0061329. 0.0070011. -5.797328. 2.308196. -2.51. 0.012. -10.32154. -1.273117. -0.5682601. 0.1109434. -5.12. 0.000. -0.7857163. -0.350804. Profitability. 0.0034031. 0.02495. 0.14. 0.892. -0.0455004. 0.0523066. Loss dummy. 32.12598. 5.409279. 5.94. 0.000. 21.52344. 42.72851. 11621. 4625.151. 2.51. 0.012. 2555.411. 20686.59. BigN E_BigN Asset t Fiscal Year Industry. Con. I propose a few explanations for the differences between pre-SOX period and post-SOX period. First, I pay attention to real earnings management. Cohen et al. (2008) suggest that management prefers accrual-based earnings management in the pre-SOX period. It is believed that accrual-based earnings management costs less than real earnings management because accrual-based earnings management doesn’t disturb firm’s real business activities. Therefore, Big N auditors are more cautious about accrual-based earnings management in the pre-SOX period. In the post-SOX 27.

(28) period, firms change their preference from accrual-based earnings management to real earnings management because accrual-based earnings management turns to be more costly because of the increased legal responsibilities on management. Besides, auditors regard real earnings management as less risky than accrual-based earnings management. Auditors are not responsible for constraining real earnings management because it is out of the audit scope. Therefore, increased real earnings management would make auditors less worried in terms of audit risk. Another explanation is the conservative accounting. Auditors may reduce the absolute level of accruals in the post-SOX period (Chambers et al. 2011) since they want less room for earnings management. This would positively affect earnings persistence. Furthermore, Big N auditors may reduce their concern about the error or fraudulent financial reporting in the post-SOX period. According to Lu and Sapra (2009), the conservative accounting means that auditors will require downward adjustments when they are in doubts with their client’s financial statements. Management must develop internal control system over financial reporting. In addition, auditors must express the opinion on the effectiveness of internal control system as well as the financial statements. In the beginning of SOX, many accounting firms involved the development of internal control system over financial reporting. In addition, auditors include internal control system into their audit scope. Therefore, auditors can acquire better understanding about the client. Consequently, auditors are able to reduce their doubts about the financial statements prepared by management in the post-SOX period. In addition, audit fees are increased in the post-SOX period (Glassman, 2005) because SOX requires more assurance from auditors. Accordingly, auditors can expand the audit scope to internal control system of the client. Moreover, the increase audit fees allow auditors use more audit resources e.g. use of specialists and audit time in the post-SOX period. Moreover, non-Big N auditors may act similar to Big N auditors in the post-SOX period. Previous employees of Arthur Andersen still remain in the accounting industry after the collapse. The knowledge of Arthur Andersen can be spread into non-Big N auditors by the former employees. At the same time, non-Big N auditors would be critically aware of the risks of the error of fraudulent reporting after the collapse and may change their audit risk strategy. Lastly, the regulatory body and financial sector started to worry the oligopoly in the accounting industry after the collapse of Arthur Andersen. Big 4 auditors mean the less competitive accounting industry comparing to Big 5 auditors. For example, big publicly trading companies are hard to choose non-Big N auditors because of reputation. Those companies generally choose their 28.

(29) auditor out of Big N auditor. If another Big 4 auditor is out of business, it will severally reduce the competence of the accounting industry (Bernard, 2008). Therefore, SEC tries to develop non-Big N auditors in the industry. The decreasing portion of Big N auditors is in line with SEC’s concern. In conclusion, Big N auditors are no more negatively related to earnings persistence in the post-SOX period. Moreover, the explanatory power of the model is greatly improved from 20% to 84% in the post-SOX period. The earnings persistence model better explains the next earnings with the current earnings, assets, industry classification, fiscal year and loss occurrence. Therefore, investors can make a reliable expectation for the future earnings based on the current information. In other words, the decision usefulness of earnings for investors is greatly improved in the post-SOX period.. Table 7 The effect of SOX on Earnings persistence Earnings t+1 Earnings t. Coef.. Std. Err. T. P> ǀ t ǀ. 95% Conf. Interval. 0.6402115. 0.0606012. 10.56. 0.000. 0.5214325. 0.7589906. 43.36794. 15.33613. 2.83. 0.005. 13.30895. 73.42692. E_SOX. 0.2378457. 0.098887. 2.41. 0.016. 0.044026. 0.4316654. BigN. 0.5028833. 9.785082. 0.05. 0.959. -18.67599. 19.68176. -0.2762552. 0.0607206. -4.55. 0.000. -0.3952683. -0.1572421. 14.06599. 13.25421. 1.06. 0.289. -11.91241. 40.04438. E_SOX_BigN. 0.3577573. 0.0991145. 3.61. 0.000. 0.1634917. 0.5520229. Asset t. 0.0110384. 0.0002784. 39.65. 0.000. 0.0104928. 0.0115841. Fiscal Year. -10.23236. 2.587522. -3.95. 0.000. -15.30393. -5.160789. Industry. -0.5974645. 0.1253611. -4.77. 0.000. -0.8431737. -0.3517554. Profitability. -0.0016807. 0.0237195. -0.07. 0.944. -0.0481713. 0.0448099. Loss dummy. -26.87836. 5.9619. -4.51. 0.000. -38.56376. -15.19297. 20503.45. 5174.149. 3.96. 0.000. 10362.06. 30644.84. SOX. E_BigN SOX_BigN. Con. 5. Conclusion Numbers of accounting scandals destroyed investors’ confidence in financial statements and accounting professionals. SOX was created to restore investors’ confidence in financial reporting. However, there are many critics about the costs of SOX such as the implementation cost, the 29.

(30) increased audit fee, the slow decision making and the bureaucracy cost (Glassman, 2005). Related parties show concern about whether the benefits of SOX outweigh the costs or not. The debate is still going on. This study directly focuses on earnings persistence. It is a more intuitive barometer of the benefits of investors since investors seem to use earnings information heavily to make their economic decisions (Sloan, 1996). They show the limited understanding of accrual component such as the lower earning persistence. Therefore, earnings persistence indicates how the decision usefulness of investors has changed in the post-SOX period. According to the findings, 1) Earnings persistence increases significantly in the post-SOX period. Therefore, investors can make better economic decisions regarding the future performance based on the current earnings information. 2) Big N auditors have a negative effect on earnings persistence in the pre-SOX period, however, it is not significant any more in the post-SOX period. There are possible explanations, such as the decrease of accrual-based earnings management, the auditor conservatism, the better understanding about the client, the increased audit scope and the decrease in Big N auditors’ market share in the accounting industry. Moreover, the explanatory power of earnings persistence model is greatly improved in the post-SOX period, 3) The positive effect on earnings persistence is stronger in case of Big N auditors in the post-SOX period. In spite of the costs of SOX, the decision usefulness of earnings for investors is greatly improved in terms of earnings persistence. The findings show that SOX restores investors’ confidence in the financial statements and accounting professionals. This study also contributes to the debate of the costs and the benefits of SOX. Previous studies pay little attention to earnings persistence regarding SOX. Considering the reliance of investors in earnings number, earnings persistence is a direct indicator to see the benefits of SOX. This study shows the empirical evidence of improved earnings persistence due to SOX adoption. However, this study has some limitations. First, this study is designed to see the effects of SOX and Big N auditors on earnings persistence for all the population. The study results do not verify the effect of real earnings management directly. Therefore, this calls future studies with regard to the impact of real earnings management. Moreover, the effects of SOX and Big N auditors might be different from the group which has a strong intention of earnings management. The other limitation is that the study deals with the aggregated effect on earnings persistence. Cash flow component and accrual component show differences in many ways. The effect of SOX and Big N auditors would be different to each component. Therefore, it would be interesting to verify the direct effect of each component on earnings persistence. 30.

(31) References American Institute of Certified Public Accountants (1972). Statement of Auditing Standard No 1. Section 110 Responsibilities and Functions of the Independent Auditor: 1,593-1,595 Arthur Andersen, Coopers & Lybrand, Deloitte & Touche, Ernst & Young, KPMG Peat Marwick, and Price Waterhouse. (1992). The liability crisis in the United States: Impact on the accounting profession. Journal of Accountancy (November): 19-23. Ascher, B. (2008). The Audit Industry: World's Weakest Oligopoly? American Antitrust Institute (AAI). AAI Working Paper No. 08-03 Ashbaugh-Skaife, H., D.W. Collins, W.R. Kinney Jr., and R. LaFond. (2008). The Effect of SOX Internal Control Deficiencies and Their Remediation on Accrual Quality. The Accounting Review 83(1): 217-250 Balsam, S., J. Krishnan, and J. Young. 2003. Auditor industry specialization and earnings quality. Auditing: A Journal of Practice & Theory 22 (2): 71–97. Basu, S. (1997). The conservatism principle and the asymmetric timeliness of earnings. Journal of Accounting & Economics 24 (1): 3-37 Becker, C., M. DeFond, J. Jiambalvo, and K. R. Subramanyam. 1998. The effect of audit quality on earnings management. Contemporary Accounting Research 15 (1): 1–24. Carcello, J.V., and C. Li. (2013). Costs and Benefits of Requiring an Engagement Partner Signature: Recent Experience in the United Kingdom. The Accounting Review 88(5): 1511-1546. Chambers, D., and Payne, J. (2011). Audit quality and accrual persistence: evidence from the pre- and post-Sarbanes-Oxley periods. Managerial Auditing Journal. Vol.26 (5): 437-456 Chen, L.H., Folsom, D.M., Paek, W., and Sami, H. (2014). Accounting Conservatism, Earnings Persistence, and Pricing Multiples on Earnings, Accounting Horizons American Accounting Association Vol. 28, No. 2: 233–260 Chi, W., L.L. Lisic and M. Pevzner. (2011). Is Enhanced Audit Quality Associated with Greater Real Earnings Management. Accounting Horizons American Accounting Association Vol. 25, No.2: 315– 335 Chung, R., Firth. M, & Kim, JB. (2003). Auditor conservatism and reported earnings. Accounting and Business Research 33 (1): 19–32 31.

(32) Coates IV, J.C. (2007). The Goals and Promise of the Sarbanes–Oxley Act. Journal of Economic Perspectives Vol. 21, No.1Winter: 91–116 Cohen, D., Dye, A., and Lys, T. (2008). Real and accrual-based earnings management in the pre- and post-Sarbanes-Oxley periods. The Accounting Review 83: 757-787 Dichev, I. D., Graham, J. R. Harvey, C. R. & Rajgopal, S. (2013). Earnings Quality: Evidence from the Field. Journal of Accounting and Economics 56: 1-33 Kim, J.B., Chung, R., and Firth, M. (2003). Auditor Conservatism, Asymmetric Monitoring, and Earnings Management. Contemporary Accounting Research 20 (2): 323-359 Lo, K (2008). Earnings management and earnings quality. Journal of Accounting and Economics 45: 350-357 Lobo, G.J. and J. Zhou (2006). Did Conservatism in Financial Reporting Increase after the SarbanesOxley Act? Initial Evidence. Accounting Horizons, 20 (1): 57 – 73. Lys, T., & Watts, R. L. (1994). Lawsuits against auditors. Journal of Accounting Research 32 (Supplement): 65-93 Lu, T., and Sapra, H. (2009). Auditor Conservatism and Investment Efficiency. The Accounting Review 84(6): 1933-1958 Roychowdhury, S. 2006. Earnings management through real activities manipulation. Journal of Accounting and Economics 42 (3): 335–370. Scott, A., Sloan, R.G. Soliman, M.T. & Tuna, I. (2005). Accrual reliability, earnings persistence and stock prices. Journal of Accounting and Economics 39:437-485 Sloan, R. G. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings? Accounting Review 71 (3): 289-315 Watts, R. L. (2003). Conservatism in Accounting Part I: Explanations and Implications. Accounting Horizons 17: 207–221.. 32.

(33)

No results found