Does the audit review of quarterly financial
statements impact the quality of earnings?
International evidence.
Name: Tom van Achteren Student number: S2564203 Supervisor: V. A. Porumb Co-assessor: D. A. de Waard Study: MSc Accountancy Word count: 10.653 Date: 19-06-2017
1
Abstract
This thesis examines the relation between quarterly audit reviews and earnings quality, and the moderating effect of investor protection on the relation between quarterly audit reviews and earnings quality. There is no consensus on effects of audit review among regulators and researchers.
Some regulators mandate reviews and others do not. Previous studies successfully show relations between reviews and increase in audit fees, however, no evidence is found on benefits of the audit reviews. Using an international sample across 15 countries over a period of 11 years, I find that firms who purchase reviews have in general lower amounts of discretionary accruals. The results
therefore indicate that the purchase of audit reviews is related to higher earnings quality. On the other hand, I find no evidence on the moderating effect of investor protection on the relation
between audit reviews and earnings quality.
Keywords: review; earnings quality; investor protection; accruals, agency. JEL Classification: K42; M41; M42,
2
Table of content
Abstract 1 Table of content 2 1. Introduction 3 2. Institutional Background 72.1 Origins of Quarterly Financial Statement Reviews 7 2.2 International Differences in the Origin of Law 8
3. Theoretical Background 9
3.1 Quarterly Reviews and Earnings Quality 9
3.2 Investor Protection 11
4. Hypothesis Development 12
5. Methodology 14
5.1 Measures of Earnings Quality 15
5.2 Quarterly Audit Review Variable 16
5.3 Investor Protection 16
5.4 Control Variables 18
5.5 Sample and Procedures 19
6. Results 20
6.1 Descriptive Statistics 20
6.2 Multivariate Regression Analysis 21
7. Robustness 22
8. Discussion and conclusion 24
8.1 Interpreting the Results 24
8.2 Implications 25 8.2.1 Practical implications 25 8.2.2 Theoretical implications 25 8.3 Limitations 25 8.4 Further research 26 References 27 Appendix 1 33 Appendix 2 34 Appendix 3 35 Appendix 4 36 Appendix 5 37
3
1. Introduction
Auditing is historically valued for its ability to provide independent assurance and credibility to accounting information (Defond and Zhang, 2014). Credibility has traditionally been created by “relying on the fact that verification of balance sheet components indirectly verifies reported income.” (Simunic, 1980, p. 172). In order for accounting information to be useful, it has to be relevant and reliable (Scott, 2015). In this paper, I draw on an international sample of 15 countries to investigate if quarterly financial statement reviews by auditors improve the quality of earnings. Additionally, I investigate if the relation between quarterly financial statement reviews and the quality of earnings is moderated by countries’ investor protection.
The perfect case to illustrate the importance of earnings quality, as well as the need for research in whether or not to perform quarterly reviews is the case of Satyam Computers from India. Back in 2009, Satyam consequently overstated its revenues and profits to meet analysts’ expectations. They eventually increased their quarterly revenues by 75% and its profits by 98% (Bhasin, 2013). This lead to an overstated balance sheet by over 1.47 billion dollars. Immediately after the news came out in March 2009, Satyam Computers’ share prices on the NYSE dropped from $29,10 to $1,80 per share, which caused a loss for investors of over $2.8 billion in Satyam (Bhasin, 2013). Although quarterly results are highly inflated (Bhasin, 2013), PwC performed quarterly reviews as a requirement of Satyam’s stock listing at the NYSE. Given the fact that auditors performed quarterly reviews, it is questionable to what extend reviews on quarterly reports restrain firms from engaging in manipulation of earnings. After all, the purpose of reviews is to mitigate the risk of misstatements in reporting. On the other hand, one could ask whether performing the reviews on quarterly financial statements is useful in detecting earnings management at all.
Years before the Satyam Computers scandal, in 2000, the Security and Exchange Commission (SEC) implemented mandatory reviews of quarterly reports by auditors in the United States (US). Quarterly audit reviews are reviews performed every quarter by auditors on a firms’ quarterly financial statements1. From the date quarterly reviews became mandatory, auditors received new duties and responsibilities by performing reviews (Bedard and Courteau, 2015). As opposed to the year-end audit, auditors now also perform reviews on a quarterly basis on a firms’ quarterly financial statements.
Relative to annual reports, managers have considerable more discretion over expense recognition in quarterly financial statements (Brown and Pinello, 2007). This is a result of the
1 Throughout the thesis, I make use of the word “review” to refer to quarterly reviews performed by auditors. Reviews
4 limited involvement of auditors with quarterly financial statements compared the annual statements (Frankel et al., 2002). Reviews of quarterly accounts involve less analytical reviews and other auditing procedures than audits of annual statements (Ettredge et al., 2000). The principle of managers having considerable more discretion over quarterly financial statements, and auditors’ involvement is limited led Brown and Pinello (2007) to question the reliability of quarterly financial statements, and therefore also the value of auditors in reviewing quarterly reports contributing to earnings quality.
An additional motivation for my research is based on claim of Bedard and Courteau (2015) that benefits and costs of reviews have been the topic of discussion for the last decade and empirical evidence on this topic is scarce.
Given the doubts involved regarding the usefulness of auditors’ reviews on quarterly financial statements, I contribute to this debate by examining one of the possible benefits of quarterly financial statement reviews: a higher earnings quality as a result of reviewed quarterly statements. I will thereby differentiate my thesis from Bedard and Courteau (2015) by increasing the amount of firm-year observations included in the sample, as well as taking the international perspective. On the one hand, I add knowledge on our understanding of earnings quality by researching one of its determinants. On the other hand, I also contribute to the understanding of the agency theory, and auditors’ capability to act as monitoring mechanism for firms.
Despite the lack of scientific evidence on the increased earnings quality of quarterly financial statements reviews, there have been performed a couple of studies on the topic of reviews. Ettredge et al. (2000) prove that firms with quarterly reviews perform more accounting adjustments during the first three quarters and less in the fourth quarter. This indicates that firms with quarterly reviews shift some adjustments from the fourth quarter to earlier quarters (Ettredge et al., 2000). Shifting adjustments into earlier quarters also suggests that earnings throughout the year are better predictable as when all adjustments are predominantly made in the fourth quarter (Ettredge et al., 2000). Manry et al. (2003) conclude in their study that firms with timely reviews “increase the likelihood that accounting earnings reflect economic events contemporaneously with returns” (p. 251).
Lastly, Bedard and Courteau (2015) investigate the positive and negative effects of quarterly financial statement reviews. They find evidence for an increase in costs incurred for extra audit services of approximately 18%. However, they do not find significant evidence that these quarterly reviews do lead to a higher earnings quality. The absence of this evidence on improved earnings quality provides an option for further research on the benefits of quarterly reviews given the fact that Bedard and Courteau (2015) are not able to find results. Aside from the fact that evidence is missing, extension of research is needed because studies are only performed while referring to one
5 single country like Canada in the study of Bedard and Courteau (2015), or just to the U.S., where the reviews on quarterly statements are required (Ettredge et al., 2000). Therefore, I draw on the ideas of Bedard and Courteau (2015), but I simultaneously extend their research to an international perspective on the benefits of reviews, whereby I also consider the influence of diverse regulatory environments, which will enhance generalizability.
The international environment does not provide research on the effect of reviews on earnings quality, although it points out the importance of considering country specific effects regarding differences in earnings quality as effect of audits. In general, the adoption of international regulations such as IFRS also enhances the quality of financial reporting (Daske et al., 2008) by achieving greater comparability and transparency. However, Brown (2011) states that although standards like IFRS are international, national patterns exist where they are allowed, implying that also improvement of earnings quality will differ between countries.2 Since national differences in investor protection are proven to be important in explaining variation in auditing as well as earnings quality (Persakis and Iatridis, 2016; Kanagarenam et al., 2011; Brown, 2011) it is important to use an international dataset that accounts for different regulatory environments. Since previous literature is inconclusive on the exact effect of stronger investor protection on earnings quality (Leuz et al., 2003; Gaio and Raposo, 2011), I also contribute to current understanding of the role of investor protection.
I perform my analyses on a dataset which only includes listed firms incorporated in countries that purchase reviews on a voluntary basis over a time span of 11 years (2005-2015). The reason for taking only firms that disclose voluntary is that it enables me to measure the differences between firms with- and without reviews more accurately since all firms operate under more equal conditions. Also, with taking only listed firms, the sample is more homogeneous since stock exchanges issue a certain level of minimal requirements for disclosure where firms should respond to by adopting regulations such as IFRS (Daske et al., 2008). With taking a sample over multiple countries, I can also account for differences in investor protection, and see to what extend this contributes to the expected relation between quarterly reviews and the quality of earnings.
The dataset that I use is predominantly compiled with financial reporting information from the databases DataStream and WorldScope. Besides, I obtain information regarding investor protection in studies performed by La Porta et al. (1998) and Spamann (2010). Lastly, in order to measure the effect of having a review, I hand collect information from annual and quarterly reports to determine if they purchased audit reviews. Subsequently, I measure the absolute discretionary
2 For example, Kanageratnam et al. (2011) find that differences in regulation, monitoring and culture can cause
cross-country differences in earnings quality levels. More specifically, Persakis and Iatridis (2016) imply that auditors will increase their performance quality when they are exposed to higher national levels of investor protection, which will in turn increase the reported earnings quality.
6 amount of total accruals using the Modified Jones Model with improvement by Dechow et al. (1995) as proxy for earnings quality. This model dominates other representative discretionary accrual models (Dechow et al., 1995), providing reason to justify the use of this specific model instead of others.
I perform a pooled ordinary least square regression analysis which I cluster by firm to determine the effect of reviews on earnings quality as well as the moderating effect of investor protection on the relation between reviews and earnings quality. Additionally, based on the fact that I draw on a dataset consisting data of firms over multiple years, I also perform a random effects panel regression to check the robustness of the results.
Regarding the first hypothesis, I expect to find a significant enhanced reported earnings quality for firms that purchase quarterly reviews. The results including robustness check confirm that firms who purchase reviews have significant lower levels of discretionary accruals. Since I use accruals as proxy for earnings quality, the results imply that firms who purchase reviews therefore report higher levels of earnings quality. I hereby manage to find the relation which Bedard and Courteau (2015) failed to establish. Besides, regulators are now provided with a well-grounded argument to mandate quarterly reviews. In addition, I provide managers with useful insights to increase their quality of reported earnings. Literature shows that increasing the earnings quality does result in firms receiving several perks such as decreased cost of capital, cost of debt, and better access to finance (Garcia-Teruel et al., 2014a; Garcia-Teruel et al., 2014b).
Further, I test for the moderating effect of investor protection on the relation between reviews and earnings quality and I expect to find a significant effect on this association. However, although the results cannot confirm my expectation as the results are insignificant, it certainly does not confirm that no moderating relation exists. The results that I found are robust across the different regressions I perform. This is quite remarkable since this indicates that there is no reason to reject the null hypotheses stating that investor protection has no moderating impact. These results are also opposing the studies of Leuz et al. (2003) and Gaio and Raposo (2011), who did manage to find relations regarding the influence of investor protection on earnings quality. Further research has to be done to investigate the moderating role of investor protection on the relation between reviews and earnings quality to establish a definite link.
The remaining of this paper is structured as follows: Institutional background, Theoretical background, Hypothesis development, Methodology section, Results section, Robustness section, and finally the Discussion and conclusion section. The next chapter is the institutional background section, where I will discuss the historical perspective on the reviews and investor protection. In the theoretical background, I provide a brief overview on the theories that help explain the effect of reviews on earnings quality. Based on the theories and by adding new arguments, I clearly state my
7 hypothesis in the hypothesis development section. Further on, in the data section, I provide an overview on the measurements of variables, including controls. I also provide definitions on the variables where needed and motivate why I expect specific relations. Then, in the results section, I provide descriptive statistics on the variables and I also include the results of the pooled ordinary least squares regression analysis. The results section is followed by the robustness section where I perform an additional regression to check the robustness of the results: a random effects panel regression. Lastly, in the discussion and conclusion section, I draw conclusions and implications based on the findings in previous chapters. Furthermore, I also discuss the limitations of my research and provide suggestions for further research.
2. Institutional background
2.1 Origins of Quarterly Financial Statement Reviews
Quarterly financial statement reviews can be distinguished into two separate categories, timely reviews and retrospective reviews. Retrospective reviews are reviews about quarterly financial statements which are performed at the end of the year in conjunction with the year-end audit (Ettredge et al., 2000). Timely reviews are the reviews about quarterly financial statements performed at the end of the last quarter (Ettredge et al., 2000).
The development of quarterly financial statement reviews starts back in 1976, when the Securities and Exchange Commission (SEC) filed a new Accounting Series Release (ASR), No. 177. Prior to ASR no. 177, involvement in interim reporting was based on a completely voluntary relation between auditors and firms (Edmonds, 1983). The release of ASR No. 177 required certain SEC registrants to include footnotes to their annual reports about interim statements which had to be audited retrospectively, but timely reviews were only recommended (Edmonds, 1983).
More recently, end 1990’s, The Securities and Exchange Commission (SEC) attempted to accomplish more reliability and relevance in financial reports. In one of the SEC’s actions, they formed the Blue Ribbon Committee (BRC) which’ objective it was to gain more oversight on the financial reporting process. In turn, the BRC reported several recommendations for improvement (Blue Ribbon Committee, 1999). One recommendation of the BRC is that firms should be required to assign external auditors to review interim statements. The SEC followed the BRC’s specific recommendation in 2000, which caused timely reviews of quarterly reports by auditors to be mandatory in the U.S. (Bedard and Courteau, 2015). From then on, auditors received new duties and responsibilities by performing quarterly reviews. Some countries, like Australia and France, followed the U.S. in making timely quarterly reviews mandatory.
8 However, other jurisdictions do not require quarterly financial statements reviews.3 Bedard and Courteau (2015) state that such diversity in national regulations suggests that there is no consensus that the benefits derived from performing quarterly reviews are exceeding the costs incurred and thus research is required. The voluntary adoption of quarterly reviews is fundamental in this study, so I can assess the differences between firms that opt in for the reviews and the firms that do not.
2.2 International differences in the origin of law
As discussed earlier in the introduction section, I take different settings of investor protection into account as a modifier for the expected relation between the performance of quarterly reviews and the quality of earnings. These settings of investor protection originate from different laws. La Porta et al. (1998) conclude that although no two nations have the exact same set of laws and enforcement, national legal systems could be classified into major families of law as a result of their sufficient similarity in certain critical respects. Specifically, La Porta et al. (1998) distinguish two predominant groups of laws: common law and civil law. Beck et al. (2003) state that the common laws evolved to protect private property owners, whereas civil laws are developed to solidify state power. Laws of many countries are influenced by either the common laws or the civil laws (Glaeser and Shleifer, 2002). Common law, originated in England, has been transplanted to its colonies through conquest and colonization which affect countries as the US, Canada, Australia and even Asian and African countries (Glaeser and Shleifer, 2002). The Civil laws have its origin in the Roman law (La Porta et al., 2008). Although lost in the Dark Ages, the catholic church rediscovered the civil law and caused its implementation in different states, including France (Glaeser and Shleifer, 2002). Under Napoleon, the French civil law was exported by conquest and colonization into many European countries, Latin American countries and parts of Africa and Asia (Glaeser and Shleifer, 2002). La Porta et al. (1998) further specified the civil law countries into three different subgroups: the French civil law countries, the German civil law countries and the Scandinavian
originated laws.
Where German civil laws originated when the German Commercial Code was written after the unification of Germany in the late 1890’s (La Porta et al., 1998). Through doctrine, the German laws had an important influence on several countries in central Europe according to La Porta et al. (1998). The Scandinavian countries are seen as the most divergent subgroup of civil law countries, where La Porta et al. (1998) further state that the Scandinavian civil laws are less derived of Roman laws when compared to French and German civil laws. Although distinct from other regulations, Scandinavian laws are highly similar to each other.
3
9 Differences between investor protection occur among countries, but originate from legal developments under different categories of laws. With respect to this research, La Porta et al. (1998) find that, in general, common law countries face stronger legal protection than civil law countries.
3. Theoretical background
3.1 Quarterly Reviews and Earnings Quality
With respect to this study, by reviews of quarterly financial statements, I only include reviews that are performed by auditors. Reviews performed by the management of firms regarding their quarterly financial statements are therefore not considered in this study.
Although research on quarterly reviews lack, theories that help explain potential relations on earnings quality could be borrowed of related topics such as auditing, oversight and governance. Dechow et al. (2010) define earnings quality as: “Higher quality earnings provide more information about the features of a firm’s financial performance that are relevant to a specific decision made by a specific decision-maker.” (p. 344).
One of the most commonly used theories regarding governance and auditing topics in relation to earnings quality is the agency theory. Jensen and Meckling (1976) define the agency relationship as: “a contract under which one or more persons (principal) engage another person (agent) to perform some service on their behalf which involves delegating some decision making authority to the agent.” (p. 308). Jensen and Meckling (1976) further state that: ”If both parties to the relationship are utility maximizers there is good reason to believe that the agent will not always act in the best interests of the principal.” (p. 308). In order to limit divergent interests, the principal can establish appropriate incentives or/and incur monitoring costs to monitor agents’ activities (Jensen and Meckling, 1976). The agency theory is concerned with resolving problems between agents and principals due to different interests and helps to explain why managers voluntarily engage auditors in testifying the accuracy of accounting reports (Jensen and Meckling, 1976).
Eisenhardt (1989) states that one of the major agency issues concerns shareholders’ inability to verify the management’s performance. In order to improve verifiability, audits are successful in reducing managements’ opportunistic behavior (agency costs) when auditors are independent (Watts and Zimmerman, 1983). The mechanism of audits result in greater verifiability, which imply that shareholders are enabled to measure management’s performance through audits.
10 lack information to review managers’ performance. Ettredge et al. (2000) find that performing reviews on quarterly financial statements result in financial information to become more predictable. This is a result of more adjustments throughout the first three quarters of the year instead of only the last quarter (Ettredge et al., 2000). In a sense, this better predictability could be interpreted as reducing the information risk as earnings will be represented in a way that is more fair and corresponding with actual values. If information risks can be eliminated, firms will inherently report a higher quality of earnings. Concluding, agency theory could help explain the value of quarterly audit reviews, which is expected to reduce information risk and inherently increase earnings quality.
Another way to look at quarterly reviews and its possibility to enhance earnings quality is through its history and nature of existence. Although mandated in some countries, quarterly reviews were only recommended when the first firms applied for them. In other countries, quarterly reviews are not mandatory at all. The pattern of assurance becoming mandatory after a history of voluntary use also existed when the first annual audits were introduced according to Watts and Zimmerman (1983). The agency theory of Jensen and Meckling (1976) helps to explain how monitoring activities function as control mechanism for shareholders in order to reduce information asymmetry. However, Watts and Zimmerman (1983) further investigate the importance of monitoring activities, and find that not only are monitoring activities incurred to limit divergent interest, but do also seem to be crucial to the formation of modern firms. Watts and Zimmerman (1983) argue that auditing activities as mechanism of monitoring existed since the early developments of corporate businesses, long before auditing became mandated by governments, therefore indicating that due to its long lasting survival, it is part of the efficient technology for firms’ organization. The voluntary use of auditors as professionals before it became mandated by governments indicates that the assurance is a market driven effect (Watts and Zimmerman, 1983). This simply means that annual audits must have had a perceived marginal value. The thoughts of Watts and Zimmerman (1983) are reinforced by the results of Becker et al. (1998) and Balsam et al. (2003), who indicate that auditors, as expected, increase earnings quality. The same theory can be applied to quarterly reviews. Quarterly reviews became mandated in 2000 by the SEC, although development started since 1976 with the introduction of ASR no. 177. Even before 1976, quarterly audit reviews were issued on a completely voluntary basis (Edmonds, 1983). Like the early annual reports, it could be stated that there is an underlying value for quarterly reports due to the fact that firms are voluntarily willing to pay for them. This could be explained once again using the agency theory, which declared that managers apply for reviews to reduce information risk.
11 firm, it still is beneficiary in multiple perspectives. Dechow et al. (2010) discussed in their paper the effects on firms for having a higher earnings quality, which is an inherent result of reducing information asymmetry, whereby several benefits are listed. Firms face lower cost of capital, lower cost of debt, and face a lower litigation propensity as a result of higher earnings quality (Garcia-Teruel et al., 2014a; Garcia-(Garcia-Teruel et al., 2014b; Gong et al., 2008). Garcia-(Garcia-Teruel et al. (2014a; 2014b) state that all kind of stakeholders prefer a low information risk, resulting in suppliers giving more trade credit to firms with higher earnings quality, and banks granting access to loans for firms with higher earnings quality. Also, Gong et al. (2008) find a negative correlation between the level of earnings quality and the amount of litigations, which implies that the litigation propensity will increase as firms increase accrual based earnings management.
3.2 Investor Protection
According to Defond and Hung (2004), investor protection can be defined as: “the extent of the laws that protect investors' rights and the strength of the legal institutions that facilitate law enforcement.” (p 269). As we speak of investors, I do not intend to only include the shareholders of the firm, but also creditors are considered as investors, just in line with research of La Porta et al. (1997). According to La Porta et al. (1998) investor protection can be measured by a combination of variables that represent shareholder rights, creditor rights and enforcement (rule of law).
Former research is not conclusive on the specific influence of investor protection on earnings quality. There are two contrasting strings of literature. One debates a positive influence of investor protection on the quality of earnings, and the other string of literature states a negative relation cannot be ruled out either.
La Porta et al. (1997; 1998) argue that investor protection is high under common law countries, whereas low investor protection exists in civil law countries. This can be explained by the findings of Beck et al. (2003), who state that the common laws evolved to protect private property owners. This evolvement under common law countries provide abilities for owners to protect their invested money. Leuz et al. (2003) argue that companies operating under stricter investor protection regimes tend to have less incentives in managing earnings since companies can be easier confronted with disciplinary actions if outsiders detect private benefits for the management, which is consistent with the observations of Beck et al. (2003). When reviews are performed by auditors, chances are that auditors will detect irregularities which can lead to lawsuits of outsiders wanting to protect their investment.
Leuz et al. (2003) also conclude that earnings management decreases as investor protection rises. Furthermore, Leuz et al. (2003) find that the quality of earnings are higher in stronger investor
12 protected regimes, and are reinforced by Francis and Wang (2008). Besides, Francis and Wang (2008) add to the debate that auditors’ performance is higher under strong investor protection regimes because auditors have an incentive to perform better as a result of reputation risk and legal exposure. Van Tendeloo and Vanstraelen (2008) find, in line with the results of Leuz et al. (2003), that auditors increase their audit quality to constrain earnings management when auditors are scrutinized by other parties.
Evidence of Leuz et al. (2003) provide reason to believe that the effect between the performance of quarterly reviews and the improved quality of earnings will be greater when firms are operating under strongly investor protected regimes.
On the other hand, Bedard and Courteau (2015) do not find positive results between the effects of quarterly financial audit reviews and the quality of earnings. This is remarkable since Bedard and Courteau (2015) perform their research in Canada, which is a common law country and therefore should enjoy high investor protection. However, there is reason to believe that the level of investor protection has a negative impact on the relation between the reviews of quarterly statements and the quality of earnings, which could explain the results of Bedard and Courteau. Giofré (2013) for example, find that strong shareholder rights as part of the measure of La Porta et al. (1998) cause firms to take excessive risks, because strong shareholder rights attract more foreign shareholders, and shareholders in general, who are generally prone to risk-taking. Excessive risk taking is an incentive to manage earnings for firms (Dong et al., 2010) which decreases its quality (Lo, 2008). Francis and Wang (2008) also state that higher investor protection does not per se increase the quality of earnings, since other factors are influential on this relation.
Lastly, Leuz et al. (2003) state that as investor protection increases, earnings management decreases. However, Gaio and Raposo (2011) find contrary results. In contrast to Leuz et al. (2003), Gaio and Raposo (2011) state that lower investor protection levels increase the quality of earnings. Firms compensate low investor protection regimes by adopting higher earnings quality standards to enhance, for example, the access to external finance (Gaio and Raposo, 2011). Since the auditors’ reviews on quarterly statements are not mandatory, it is likely to assume that firms demand a higher audit quality in weakly investor protected regimes to assure a higher quality of earnings for investors. Becker et al. (1998) and Balsam et al. (2003) find evidence for the fact that an improved audit quality will lead to a higher quality of earnings. According to this explanation, the level of investor protection has a negative moderating effect on the positive association between the auditors’ review of quarterly financial statements and the quality of earnings.
13 According to the agency theory, managers could act in ways that are not in the best interests of shareholders (Jensen and Meckling, 1976). Given the situation that mangers have considerable more discretion over expense recognition in quarterly financial statements than in annual reports (Brown and Pinello, 2007), managers could be tempted to use this discretion to perform earnings management in order to accomplish their own interests. Managers using this discretion by engaging in earnings management will in turn lower their quality of earnings. Therefore, shareholders will try to reduce this discretion by performing monitoring activities. Shareholders can ask the firm to purchase quarterly financial statement reviews that would reduce managements’ discretion and improve earnings’ predictability (Ettredge et al., 2000). I expect that the auditors contribute in decreasing management discretion resulting in higher earnings quality in the quarterly financial statements.
Additionally, since annual audits were also not mandated when the first firms were audited, it indicates a sign of a market driven effect that audits have a perceived underlying value (Watts and Zimmerman, 1983; Becker et al., 1998; Balsam et al., 2003). The same thought process can be applied to the existence of quarterly reviews, which were not mandated by regulators in the first place, indicting an existence of market driven effect. Therefore, reviews are like annual audits also expected to have a perceived marginal value since firms are willing to pay for them. Using the arguments and the framework provided by the agency theory, I develop the following hypothesis:
H1: The voluntary purchase of a quarterly audit review is associated with increased earnings quality.
Regarding the moderating effect of investor protection on the relation between quarterly audit reviews and the quality of earnings, former research provides mixed evidence. On the one hand, it can be argued that investor protection probably has a negative influence on the relation between quarterly audit reviews and the quality of earnings. Justification for this statement can be found in the results of Gaio and Raposo’s (2011) research. They find that lower investor protection results in higher earnings quality. Firms compensate weak legal environments to gain access to external finance by adopting higher standards (Gaio and Raposo, 2011). In high investor protected regimes there is no need to compensate by adopting higher standards to gain external finance, and therefore reviews are less valued in high protected regimes. In other words, the ability to undertake disciplinary actions against firms in high investor protection regimes will create incentive for firms to comply with high standards without the need for assurance from quarterly reviews. Absence of disciplinary actions against firms that do not comply with reporting standards in weak investor protected regimes provide room for audit reviews to act as substitute.
14 Ultimately, the performance of quarterly reviews can add more value in achieving higher earnings quality in weak investor protected regimes. According to this reasoning, investor protection will have a negative influence on the relation between quarterly audit reviews and the quality of earnings.
As discussed in the theoretical background section, there is also a string of literature which debates a positive moderating role between having a quarterly review and the quality of earnings. One of the reasons why investor protection should increase the relation between quarterly reviews and the earnings quality is that not only firms increase their standards due to possible lawsuits and disciplinary actions, but so do auditors. Auditors in highly investor protected regimes should increase their audit quality due to reputation risk and legal exposure (Francis and Wang, 2008). This means that investor protection could also have a positive effect on the relation between quarterly audit reviews and earnings quality. Both strings of literature concerning the influence of investor protection state that investor protection does influence the quality of earnings. However, a specific direction for the hypothesis cannot be conducted given the contrary results in literature. I therefore formulate the following non-directional hypothesis:
H2: Investor protection impacts the association between the purchase of an audit review and earnings quality.
5. Methodology
In order to test both Hypotheses 1 and 2, I measure the quality of earnings as the discretionary part of total accruals consistent with (Bedard and Courteau, 2015). Accruals are currently defined as the difference between earnings and cash flows in a certain year (Dechow et al., 2010). I use the Modified Jones Model of Dechow et al. (1995) to measure the discretionary part of the accruals. The first step to measure discretionary accruals is to measure the total amount of accruals, which is calculated according equation (1):
(1) 𝑇𝑂𝑇𝐴𝐿 𝐴𝐶𝐶𝑅𝑈𝐴𝐿𝑆 = 𝑇𝐴𝑖𝑡 𝐴𝑖𝑡 −1= 𝛼1 1 𝐴𝑖𝑡 −1 + 𝛼2 △𝑅𝐸𝑉𝑖𝑡− △𝑅𝐸𝐶𝑖𝑡 𝐴𝑖𝑡 −1 + 𝛼3 𝑃𝑃𝐸𝑖𝑡 𝐴𝑖𝑡 −1+ 𝜀𝑖𝑡
Where 𝑅𝐸𝑉𝑖𝑡 stands for revenues of firm i in year t; 𝑅𝐸𝐶𝑖𝑡 hold the accounts receivable for firm i in year t; 𝑃𝑃𝐸𝑖𝑡 denotes the property plant and equipment for firm i in year t; 𝐴𝑖𝑡 −1 represent the lagged total assets for firm i in year t, and 𝜀𝑖𝑡 is the error term in year t for firm i.
Total accruals for firm i in year t (𝑇𝐴𝑖𝑡) are defined as the net income before extraordinary items
15 al., 1995). Cash flows from operations can be measured according to the following formula (2).
(2) 𝐶𝐹𝑂𝑖𝑡 =
𝑁𝐼𝑖𝑡+𝐷𝐴𝐸𝑖𝑡− △𝐶𝐴𝑖𝑡− △𝐶𝐴𝑆𝐻𝑖𝑡 + △𝐶𝐿𝑖𝑡− △𝐶𝑀𝐿𝑇𝐷𝑖𝑡− △𝐼𝑇𝑖𝑡
𝐴𝑖𝑡 −1 .
Where 𝐷𝐴𝐸𝑖𝑡 stands for depreciation and amortization expenses for firm i in year t; 𝐶𝐴𝑖𝑡 hold the current assets of firm i in year t; 𝐶𝐴𝑆𝐻𝑖𝑡 stands for cash and cash equivalents for firm i in year t;
𝐶𝐿𝑖𝑡 represents the current liabilities for firm i in year t; 𝐶𝑀𝐿𝑇𝐷𝑖𝑡 represent the current portion of long term debt for firm i in year t, and 𝐼𝑇𝑖𝑡 are the income taxes payable for firm i in year t. All variables are divided by lagged total assets (𝐴𝑖𝑡 −1).
Note that equation (1) calculates the total amount of accruals, while I use only the discretionary part of total accruals (ACCRUALS) as measurement for earnings quality. The residuals of equation (1) act as the discretionary amount of accruals and can be calculated using the following equation:
(3) 𝐴𝐶𝐶𝑅𝑈𝐴𝐿𝑆 = 𝑇𝐴𝑖𝑡−(𝛼1+ 𝛼2∗ △𝑅𝐸𝑉𝑖𝑡−△𝑅𝐸𝐶𝑖𝑡 + 𝛼3∗𝑃𝑃𝐸𝑖𝑡)
𝐴𝑖𝑡 −1
𝛼1, 𝛼2, and 𝛼3 are estimated regression coefficients based on industry sector (Jones, 1991). This
means that α is the average level of discretionary accruals across all firms of a certain industry sector in the estimation period (2005-2015). Barth et al. (1998) identified 14 different industries which I use to estimate α.
To test Hypothesis 1, I assess if the audit review explains variation in abnormal accruals after controlling for all normal determinants. Further, in order to test the predictions of Hypothesis 2, I use the interaction between investor protection and audit review to measure the moderating effect on the quality of earnings. The respective models are presented in equation (4) and (5).
(4) ACCRUALS = β0 + β1 REVIEW + β2 BIG4 + β3 DEBT + β4 INVENTORY + β5 ROA + β6 INDUSTRY(t) + β7 YEAR(t) + ε
(5) ACCRUALS = β0 + β1 REVIEW + β2 INVESTOR PROTECTION + β3
REVIEW*INVESTOR PROTECTION + β4 BIG4 + β5 DEBT + β6 INVENTORY + β7 ROA + β8 INDUSTRY(t) + β9 YEAR(t)+ ε
5.1 Measures of Earnings Quality
16 2002; Dechow et al., 1995). In this study I use the definition constructed by Dechow and Dichev (2002). Dechow and Dichev (2002) state that estimation errors and subsequent corrections function as noise in financial statements which determine earnings quality.4 I measure earnings quality using the Modified Jones Model since it dominates other representative models in accurately computing earnings quality by the use of discretionary accruals (Dechow et al, 1995). The Modified Jones Model builds on the model originally introduced by Jones (1991), and on the modifications made by Dechow et al. (1995).5 The model I use does, however, measure earnings quality in a single year instead of every quarter. Although comparing quarterly reviews with annual earnings, instead of quarterly earnings, seems to put things out of perspective, there is reason to believe and justify using this method. The main argument stands in the work of Lee et al. (2014) and Bedard and Courteau (2015), that audit efforts in interim reviews reflect in the quality of the annual financial report. This means that the audit reviews on quarterly reports not only improve the earnings quality in quarterly reports, but also improve the overall earnings quality in that current year. A second reason to justify comparing quarterly audit reviews with annual earnings quality is of a more practical nature. Not all firms publish their quarterly results which include essential financial information to compute the level of earnings quality. I rely on annual levels of earnings management which are published more widespread by firms than quarterly reports. Lastly, using annual levels of accruals is a commonly used method to measure earnings quality, which makes this method reliable one to use (Dechow et al., 2010).
5.2 Quarterly Audit Review Variable
Quarterly audit reviews are timely activities performed by auditors on quarterly financial statements of a firm whereby auditors state an opinion on the quarterly statements. The data regarding the voluntary choice to review is hand collected by a comprehensive analysis of annual and quarterly reports for 514 firms in 15 countries. Reports which do not indicate whether they apply for quarterly audit reviews are omitted. I build a dichotomous variable REVIEW which takes the value of one if a specific firm purchases the auditor’s review on quarterly reports in four consecutive quarters in a current year and zero otherwise. Firms that only purchase reviews for the third and fourth quarter, or only the second quarter represent a value of 0.
5.3 Investor Protection
4 According to Dechow and Dichev (2002). earnings quality decreases as the magnitude of estimation errors rise. 5 Dechow et al. (1995) enriched the existing model by adjusting the change in revenues for the change in account
receivables in that current year. The original Jones Model tends to include estimation errors when managers exercise discretion over revenues, which are eliminated by the modification (Dechow et al., 1995; Dechow et al., 2010).
17 As stated earlier, I draw on the definition of investor protection which is developed by DeFond and Hung (2004), who define investor protection as: “the extent of the laws that protect investors' rights and the strength legal institutions that facilitate law enforcement.” (p.269). This interpretation of investor protection is also used by Leuz et al. (2003). It is important to mention that DeFond and Hung (2004) and Leuz et al. (2003) especially focus on the rights of minority shareholders. Minority shareholder rights refer to “how strongly the legal system favors minority shareholders against managers or dominant shareholders in the corporate decision making process, including voting rights” (La Porta et al., 1998, p. 1127). To measure the level of investor protection, I use the index measures developed by La Porta et al. (1998) and Spamann (2010). La Porta et al. (1998) computed an antidirector index for 49 countries out of 6 variables that comprise the definition minority shareholder rights, which are used in more than 100 studies to capture the level investor protection (Spamann, 2010).
For my research, however, I make use of the revised version of La Porta et al. (1998) antidirector index. Spamann (2010) reexamined the initial index computed by La Porta et al. (1998) to form a new index, which is more clearly defined and coded more reliably (Spamann, 2010). Spamann (2010) published the revised values for the 49 countries, which I use in my analyses.
Aside from the antidirector rights, I also include law enforcement variables in my analyses. Law enforcement could be seen as a substitute for weak legal rules (La Porta et al., 1998). When investors are abused by the management of firms, well-functioning courts can step in and save the investors (La Porta et al., 1998). To capture the level of enforcement, La Porta et al. (1998) developed a few proxies which together could be transformed into an index. For my research, I draw on the method of DeFond and Hung (2004) and Leuz et al. (2003). Leuz et al. (2003) state that the most important proxies to include for measuring law enforcements are La Porta et al.’s (1998) index of legal system’s efficiency, the rule of law index, and the corruption index.
Although Leuz et al. (2003) and DeFond and Hung (2004) included two separate measures (antidirector index and law enforcement index) of investor protection to perform their analyses, I prefer to combine all proxies for investor protection in a single measure. This can be justified by the interacting effect between laws and enforcement. Since laws and enforcement work as substitute to each other (La Porta et al., 1998), both measures apply to the same construct: investor protection. I therefore rearrange all proxies into a new variable by performing a principal component analysis. Another argument to combine all variables is to capture the essence of investor protection better, which consists of different law as well as enforcement aspects. Combining all proxies into a single measure does not leave me with two separate proxy indices for investor protection, but only a combined alternative.
18 component analysis to rearrange the proxy indices for investor protection based on its eigenvectors and eigenvalues. I then take the 2 eigenvectors, which jointly capture 99% of the variation between all index proxies. Lastly, I generate a new variable based on the both components which is called
INVESTOR PROTECTION.
5.4 Control Variables
Ball and Shivakumar (2008) performed a research studying firms that issue IPO’s (Initial Public Offerings) and its relations with earnings quality. The results show that firms that go public report a higher quality of earnings (Ball and Shivakumar, 2008). An important reason for the increase in earnings quality is that firms that go public in general attract more scrutiny, under which enhanced scrutiny by auditors (Ball and Shivakumar, 2008). DeAngelo (1981) find that the size of audit firms affects the audit quality. Big4 auditors have a higher perceived reputation, and therefore have more to lose than non-big4 auditors. The reputation of Big4 auditors serves as a collateral to provide higher audit quality than non-big4 firms (DeAngelo, 1981). DeAngelo (1981) measures audit quality by looking at the three variables under which auditors’ scrutiny. Since Big4 auditors deliver higher quality of audits and this higher level of scrutiny results in a higher quality of earnings, I shall control for firms that are audited by BIG4 firms.
Another perspective on Big4 auditors and its relation with earnings quality is provided by Francis and Krishnan (1999), who concluded that Big-N auditors are associated with reporting conservatism. Conservatism is a strategy for auditors whose clients are exposed to potential accrual estimation errors, leading to the point where auditors are more likely to issue modified reports on firms with higher accruals (Francis and Krishnan, 1999). Since only Big-N auditors are correlated with conservatism, clients of Big-N auditors have a higher perceived earnings quality.
A second variable that is proven to have influence on earnings quality is the level of debt. LaBelle (1990) find that firms with higher levels of debt are more likely to accelerate income recognition by accruals. Firms that borrow money are likely to be confronted with debt covenants, and therefore need to comply with certain imposed rules by lenders. LaBelle (1990) state that the most common covenants involve compliance to a certain maximum ratio between the level of debt scaled by tangible assets, and a floor to the interest coverage ratio. Firms that have attracted larger proportions of debt are more likely to violate covenants. When firms are more likely to violate debt covenants due to higher amounts of debt, they are tempted to use revenue increasing accruals as a mechanism to decrease the probability of violating debt covenants (LaBelle, 1990). Therefore, I control for debt ratio, which is calculated as the amount of long-term debt scaled by total assets which is indicated as DEBT in the regression models of equation (4) and (5).
19 The third control variable that I use is the inventory ratio. Ashbaugh-Skaife et al. (2009) initially use this variable to control for potential internal control deficiencies. According to Ashbaugh-Skaife et al. (2009), internal control problems lead to intentional as well as unintentional financial reporting failures which cause lower earnings quality. Higher levels of inventory are positively correlated with internal control deficiencies and therefore are expected to have a negative relation on earnings quality (positive on the amount of ACCRUALS). This negative relation is confirmed by Hope et al. (2013), who find that the inventory ratio has a significant negative relation on earnings quality. I therefore control for the inventory ratio (INVENTORY), which is measured by the amount of inventory scaled by total assets for a firm in a specific year.
Another control variable that I add is the effect of ROA (Return On Assets). Bedard and Courteau (2015) control for ROA to deal with the effects of profitability, since ROA is a common measure for performance. Bedard and Courteau (2015) find that ROA is positively associated with earnings quality. Therefore, I expect the ROA to be negatively associated with ACCRUALS. To measure the ROA I divide the net income before extraordinary items by the amount of total assets at the start of the year.
The fifth control variable that I use is a dummy variable called INDUSTRY(t). Like Bedard and Courteau (2015), I control for the effect of reviews on the level of earnings quality by taking differences between industries into account. Based on the method of Barth et al. (1998), I allocate all firms that are included in the sample into 14 different industries using the four-digit SIC code.
Lastly, the sample in this research covers a time period of 11 years (2005 till 2015). Therefore, I control for the potential differences in earnings quality between years: YEAR(t). I created a year dummy for each year which equals 1 if the observation took place in the specific year and otherwise 0. Table 1 (see appendix 1) provides a comprehensive overview of all variables including its measurement.
*Insert Table 1 here*
5.5 Sample and Procedures
I start with a sample of 5.593 firm-year observations. However, due to absence of the financial reports or information on the review I waive 1.980 firm-year observations. This results in a sample of 3.612 firm-year observation for which I am able to find information on the review. Due to missing information on control variables in DataStream, I have to adjust my sample and waive another 362 firm-year observations. The final sample that I use for the regressions consist of 3.251 firm-year observations with 514 different firms included.
20 Table 2 (see appendix 2) provides an overview on the base sample.
*Insert Table 2 here*
Earlier this chapter I explained how I collect data with regard to the described variables. I apply multiple data sources, from primary data which I collected myself, secondary data which I use from other researchers as well as tertiary data from databases. This mix of different data sources is called data triangulation. As Smith (2015) states, getting data from different sources will prevent over reliance on databases and provide a point of differentiation.
However, all collected data is rather pure and not perfectly suitable for running regressions due to outliers. With using financial data and specifically financial ratios, there is an inherent problem involved that certain firms under specific circumstances can report extremely high (or low) ratios. To handle this problem, I consciously bias my all continuous variables (ratio’s) with winsorization at 1%. This means that the ratios are corrected at the top 1% and bottom 1%, taking out the extreme values that can seriously bias the results and replace them with lower value at the 99% level. This essentially means that no data will be lost, only that the effect of outliers is mitigated. To test both hypotheses developed in chapter 4, I make use of an ordinary least squares (OLS) method, with standard deviations clustered by firm.
6. Results
6.1 Descriptive StatisticsIn Table 3 (see appendix 3), I present a combined oversight of the means and standard deviation per variable as well as the Pearson’s correlation matrix with included variables. As discussed in former chapters, the main variables of interest are ACCRUALS, REVIEW and
INVESTOR PROTECTION.
The results of the descriptive statistics show, that on average, the level of discretionary accruals is 8,4%. This essentially means that the reported earnings are on average inflated or deflated by 8,4% for a given firm. Besides, only 10% of the firms use quarterly reviews. Furthermore, I perform a Pearson’s correlation on the explanatory variables. According to Smith (2015), Pearson’s correlation is the most useful preliminary diagnostics technique. At first, it determines the direction of any relationship which should later on return in any regression analyses with corresponding signs (Smith, 2015). Pearson’s correlation provides us therefore with useful insights on the linear relation between two specific variables. Pearson’s correlation indicates that
21
ACCRUALS are negatively and significant correlated with REVIEWS, implying that firms who
purchase reviews have in general higher levels of earnings quality. INVESTOR PROTECTION is although insignificant expected to also have an earnings quality increasing effect. These results are in line with findings of previous research by Bedard and Courteau (2015), and Leuz et al. (2003).
With regard to the control variables, all correlations between accruals and the controls are in line with previous literature, except for DEBT leverage. The Pearson’s correlation matrix represent a negative relation between the amount of debt and the amount of accruals, although former research (LaBelle, 1990) suggested a reversed relation.
*Insert Table 3 here*
Furthermore, another perk of using Pearson’s correlation matrix is that it provides insight on potential multicollinearity problems, whereby high correlations among the competing explanatory variables indicate potential problems regarding multicollinearity (Smith, 2015). Absence of multicollinearity indicates that regression estimates with OLS regressions can be measured in a stable way, with high precision (Curto and Pinto, 2007).
Since the highest absolute correlation value amounts to 0,216, the risks of multicollinearity are limited among the variables in Table 1. However, untabulated analysis of the dummy variables indicate that year-dummy 2015 and SIC industry dummy 14 (Services) provide signs of potential multicollinearity problems. Therefore, I omit these specific dummy in further regression analyses.
6.2 Multivariate Regression Analysis
Considering the variables explained in the previous sections, I construct the following regression model to test the predictions of Hypothesis 1:
(6) ACCRUALS = β0 + β1 REVIEW + β2 BIG4 + β3 DEBT + β4 INVENTORY + β5 ROA + β6 INDUSTRY(t) + β7 YEAR(t) + ε
The main coefficient of interest in Equation 6 is the one of REVIEW. If significant and negative, it indicates that the voluntary choice to purchase the review is positively associated with increased earnings quality. Further, I add my measure of investor protection in the model in order to test the predictions of Hypothesis 2:
(7) ACCRUALS = β0 + β1 REVIEW + β2 INVESTOR PROTECTION + β3
22 + β8 INDUSTRY(t) + β9 YEAR(t)+ ε
The main coefficient of interest in Equation 7 is the one of the interaction term REVIEW*INVESTOR PROTECTION. If significant and negative, it indicates that the voluntary choice to purchase the review is positively associated with increased earnings quality especially in countries with strong investor protection.
Table 2 (see appendix 4) represent the results of the multivariate regression analyses. As stated in the former chapter, I run an ordinary least squares (OLS) method with standard deviations clustered by firm.
*Insert Table 4 here*
The first regression column in table 2 represents the regression model as drafted in equation 6, where I test if quarterly reviews are associated with higher earnings quality through the use of discretionary accruals. The second regression column represents the model as drafted in equation 7. The main variable of interest in the second regression column is the interaction effect between reviews and investor protection, testing the second hypothesis.
When looking at the regression in the first column regarding the first Hypothesis, the results show as predicted in Hypothesis 1 that firms having a review are associated with having less discretionary accruals (-0,017, p = <0,01). The coefficient of the REVIEW is significantly different from zero on a 1% probability level, thereby supporting as strong evidence to accept the first hypothesis.
The second OLS regression regarding the interaction effect between the review and investor protection on the amount of accruals, as represented in the second regression column in table 2, reflect great similarity with the first regression (without interaction terms). However, when I consider the main variable of interest, I observe a negative insignificant effect (-0,006, p = 0,28) on the amount of accruals.
On the control variables, firms having a Big4 auditor is significantly associated with lower amount of accruals, confirming prior research. The debt ratio is also significantly correlated (-0,030, p = <0,1) with lower accruals, contrary to prior research of Labelle (1990). A higher inventory ratio, while positively related to the amount of accruals as previous literature indicate, is not significantly different from zero. Lastly, the return on assets (ROA) is conform prior research significant negatively related to accruals.
23 In order to test the robustness of the results that I find, I perform an additional regression method. To be more specific, I also run a random effects panel regression analysis. As I collect information on the reviews over a time span of 11 years for a broad range of firms, panel data analysis is an appropriate way to perform regressions. Specifically, panel regression approach is likely to provide unbiased estimates of coefficients (Wooldridge, 2012). When considering panel regression, extinction is made between fixed effects and random effects analysis. To determine the most appropriate regression method, a Hausman test will provide a general implication. The Hausman specification test determines if biases inherent in random effect regressions are insignificant enough to ignore (Smith, 2015). Otherwise, a fixed effects model is the more appropriate one to use. The Hausman test basically compares the outcome of the random effect model with the fixed effects model whereby a null hypotheses is conducted that states that the differences between both models are insignificant. Therefore, significant results on the Hausman test suggest rejection of the null hypotheses, and the use of the fixed effects model is the most appropriate (Smith, 2015).
The results of the Hausman test show that the differences between the random and fixed effects methods are insignificant for the full model I use (p = 0,27). This means that the biases incurred in the random effects model can be ignored and that the random effects regression model is an appropriate one to use.
Table 5 represents the results of the random effects panel regression model. M4 and M5 respectively refer to the corresponding equation which I have developed in the methodology section. The first model refers to the equation for the first hypotheses, and the second model consequently to the equation connected to the second hypotheses.
*Insert Table 5 here*
The first panel regression show again that firms having a review are significantly negatively correlated to the amount of accruals (-0,015, p = <0,05). These results are in line with the results of the pooled OLS regression clustered by firms in the previous chapter, and therefore provide
additional evidence for the confirming the first hypotheses.
The second panel regression shows a negative and insignificant interaction effect of investor protection on the relation between reviews and accruals (-0,007, p = 0,61). This insignificant result is in line with the OLS regression in the previous chapter, providing additional evidence to reject the second hypotheses.
Also the control variables show great similarity to the OLS regression in the previous chapter. As predicted, BIG4 is significant and negatively correlated to accruals. The same applies to
24 ROA, which’ effect also show a significant and negative relation with respect to accruals. The inventory ratio shows in contrast to the OLS regression a significant effect. Although, in line with former research, both the OLS and the panel regression show a positive relation between the inventory ration and the amount of accruals. Most remarkable is the effect of the debt ratio on the accruals. Previous research shows a positive relation between the amount of debt and accruals. However, I find that the amount of debt is significant and negatively related to accruals.
8. Discussion and conclusion
8.1 Interpreting the ResultsIn this thesis, I analyze the effect of audit reviews on the quality of earnings in an international setting. I hereby try to explain a relation which former research failed to establish. Besides, I am adding a new variable to explain international differences between the effects of reviews on earnings quality. The interaction I try to establish constitutes the effect of investor protection on the relation between reviews and earnings quality. I measure earnings quality as the amount of discretionary part of total accruals according to the modified Jones model by Dechow et al. (1995).
The results of my empirical tests successfully confirm the accrual decreasing effect of firms that purchase quarterly audit reviews. Since the amount of accruals is the proxy I use for earnings quality, the results essentially show that as firm choose to purchase audit reviews, they seem to have an increased reporting quality. Therefore, this thesis is the first one to establish significant benefits attached to purchasing quarterly reviews. Bedard and Courteau (2015) did find a negative coefficient between reviews and accruals, although insignificant. The results that I find are not only significant, but are also robust. I tried different regression methods to determine the effect between reviews and earnings quality, and the results withstand multiple models.
My findings are also in line with the principles of the agency theory. Although auditors’ involvement is significantly lower on reviews compared to year end audits (Frankel et al., 2002), purchasing audit reviews seem to act to some extent as mechanism for monitor activity (Watts and Zimmerman, 1983).
When looking at the evidence for the moderating effect of investor protection on the relation between reviews and earnings quality, the evidence can’t confirm nor deny the existence of a moderating effect. However, the coefficients in both regressions predict a negative moderating effect, which would incline a tendency for reviews to increase the earnings quality even more when the concerned firm is subject to a strong investor protected regime. These findings, although
25 insignificant, are therefore likely to comply with previous research of Leuz et al. (2003), who predicted that auditors have greater incentive to increase the quality of reported earnings due to
litigation risks.
My findings, however, provide no real evidence for this relation. The potential cause for the insignificant moderating result is probably embedded in the set-up of the sample. When I analyze the sample which I used to perform the regressions, I conclude that the distribution of observations between countries in unbalanced. De distribution of firm-year observations per country differ between 0,2% and 50,1% of the total observations. Although this would be no problem when the sample is large enough. However, in my sample, this means that some countries are represented by just a hand full of observations. A representative sample per country cannot be drawn from my sample, and therefore it is no surprise that no conclusion can be drawn based on my findings with respect to the moderating effect of investor protection. The results remain negatively associated and insignificant across both regression methods that I used.
8.2 Implications
8.2.1 Practical implications. My thesis has important implications for managers as well as regulators. In the introduction, I describe how the benefits of reviews are lacking although audit fees incurred are significantly higher (Bedard and Courteau, 2015). The limited procedures which auditors perform in reviews compared to year end audits (Frankel et al., 2002) begged the question if these limited procedures actually add value. Despite inconclusive results on the benefits, some regulators mandated reviews. The results that I find help to justify regulators who mandate the reviews. I prove that firms which purchase reviews, report statistically significant higher quality of earnings in annual figures. Regulators therefore have an incentive to impose new legislation to mandate reviews for all listed firms.
Besides, managers can also benefit of these results by purchasing reviews on a voluntary basis. Since increased earnings quality is associated with lower cost of capital and increased access to finance (Garcia-Teruel et al., 2014a; Garcia-Teruel et al., 2014b), managers can achieve these benefits through the implementation of reviews. Secondly, investors will be provided with higher reported earnings. This will allow investors to valuate the firm more accurately, which results generally in increased stock prices (Gaio and Raposo, 2011).
8.2.2 Theoretical implications. The results of this thesis also contribute to the literature. First, I find evidence for the increased earnings quality as a result of purchasing quarterly reviews. These results add to the understanding of determinants of earnings quality. Besides, it also adds to
