1 Amsterdam Business School
What are the Effects of Audit Quality
and Tax Alignment on Accrual-based
and Real Earnings Management
within Privately Held Companies?
Name: Ninh Pham
Student number: 10281746
Date: June 22nd, 2015
Word count: 15,534
Thesis supervisor: dhr. dr. J.J.F. van Raak
MSc Accountancy & Control, specialization Accountancy & Control Faculty of Economics and Business, University of Amsterdam
2 Statement of originality
This document is written by student Ninh Pham, 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 Economic and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Abstract
The objective of this study is to examine the relations between audit quality, accrual-based earnings management and real earnings management. The study focuses on privately held companies in the Netherlands and in Belgium and utilizes data from the period 2012-2013 to analyze the fiscal year 2013. To provide an answer on what effects audit quality has on earnings management within privately held companies, multiple relationships have to be examined. First, the effect of audit quality on accrual-based earnings management practices is examined, using auditor size and auditor industry expertise as proxies for audit quality. This relation is examined in a high tax alignment country (Belgium) and a low tax alignment country (the Netherlands). These different tax alignment settings are chosen due to the fact that prior literature has documented that audit quality differs across different tax alignment settings. Using the performance-adjusted modified Jones model provides results indicating that privately held companies’ ability to practice accrual-based earnings management is constrained in high tax alignment settings when audited by a Big 4 auditor. Second, to examine whether privately held companies behave in the same way as publicly traded companies, I examine whether privately held companies also use real earnings management as a substitute when accrual-based earnings management is constrained. Contradictory to publicly traded companies, I expect no substitution effect due to the differences in the fundamental structures and incentives of privately held companies. Using the sales manipulation method, I find no significant evidence regarding this effect. This study suggests that higher audit quality provided by a Big 4 auditor in a high tax alignment setting constrains accrual-based earnings management when compared with a low tax alignment setting but no evidence for a substitution effect could be found, which suggests that privately held companies do not shift to real activities manipulation in order to manage earnings.
Keywords: Audit Quality, Accrual-based Earnings Management, Real Earnings
Management, Privately Held Companies, Big 4 Auditor, Auditor Industry Expertise, Tax Alignment.
4 Table of Contents
1. Introduction ... 5
2. Literature Review and Hypothesis Development ... 8
2.1 Differences between Privately Held Companies and Publicly Traded Companies ... 8
2.1.1 Structure ... 8
2.1.2 Agency Theory ... 9
2.1.3 Moral Hazard ... 9
2.1.4 Information Asymmetry ... 10
2.2 Incentives for Earnings Management ... 12
2.2.1 Avoiding Small Losses and Earnings Declines ... 12
2.2.2 Tax Incentives ... 13
2.3 Earnings Management ... 14
2.3.1 Audit Quality and Accrual-based Earnings Management... 15
2.3.2 Audit Quality, Accrual-based and Real Earnings Management ... 17
3. Research Design ... 18
3.1 Sample ... 18
3.2 Empirical Models ... 20
3.2.1 Accrual-based Earnings Management ... 20
3.2.2 Real Earnings Management ... 23
4. Results ... 25
4.1 Descriptive Statistics ... 25
4.2 Accrual-based Earnings Management in Different Settings ... 31
4.3 Real Earnings Management – Sales Manipulation ... 34
4.4 Sensitivity Analysis ... 36
5. Conclusion ... 38
References ... 41
5 1. Introduction
Numerous accounting scandals have taken place in recent years in both the United States and the European Union. The U.S. had to deal with WorldCom and Enron among others, while the EU had its own problems with Parmalat and Ahold. These major accounting scandals raised concerns and exerted pressure on the accounting and auditing profession. From a societal point of view, these accounting scandals must be prevented and any sign of fraud should be noticed timely. Audit quality is becoming more of importance in order to provide investor protection. A lack of audit quality has led to the implementation of various laws and regulations. In the U.S., for example, the Sarbanes-Oxley Act (SOX) was implemented in 2002 to ‘fix auditing’ and its primary goal was to improve audit quality and reduce fraud on a cost-effective basis (Coates, 2007). In the EU the 8th EU Company Law Directive was introduced to strengthen the consistency and credibility of the external audit. However, these laws and regulations are primarily focused on publicly traded companies while privately held companies are mostly neglected. Privately held companies differ fundamentally from publicly traded companies, in that they have different ownership, governance, financing, management and compensation structures (Ball & Shivakumar, 2005). However, research regarding privately held companies is not as extensive as research regarding publicly traded companies. Therefore this study deals with the effects of audit quality on earnings management within privately held companies.
The objective of this study is to examine the relations between audit quality, accrual-based earnings management and real earnings management. To provide an answer on what effects audit quality has on earnings management in privately held companies, multiple relations have to be examined. First, the link between audit quality and accrual-based earnings management is examined. This link is then used as the setting in which the relation between accrual-based earnings management and real earnings management is examined.
This study focuses on privately held companies. The motivation behind this is the fact that even though privately held companies constitute a significant part of the economy and audit market, there is still a gap in the literature. The differences between privately held companies and publicly traded companies lead to some differences in incentives for earnings management. Prior research has found conflicting evidence when looking at the earnings quality between publicly traded and privately held companies (Vander Bauwhede et al., 2003 2004; Ball & Shivakumar, 2005; Burgstahler et al., 2006). Agency conflicts are expected to be higher for publicly traded companies due to their structure. Capital market forces, such as
6
demand from public equity markets for information, influence the reporting behavior of these companies. They are required to provide higher quality financial statements to deal with the information asymmetry between principal and agent (Burgstahler et al., 2006). Privately held companies, on the other hand, do not broadly distribute their financial statements to the public and information asymmetry is often resolved through private communication (Ball & Shivakumar, 2005). These financial statements are, however, often used to determine taxes which give management incentives to manage earnings. The earnings of privately held companies are therefore mainly affected by management’s practices to manage taxes.
The complete sample is collected from the Amadeus database (2013) provided by Bureau van Dijk. The Amadeus Database contains detailed global private company information. The sample consists of privately held companies from Belgium and the Netherlands. These countries are selected due to the fact that Belgium represents a high tax alignment setting while the Netherlands represents a low tax alignment setting. Furthermore, the legal system and enforcement in both countries are quite similar (La Porta et al., 1998). The selection therefore controls for these factors as well. The final sample consists of 6,615 firm-year observations for the year 2013.
Using the final sample, the relation between audit quality and accrual-based earnings management is examined in two differing tax alignment settings. The differentiation between whether financial reports also serve as a base for tax purposes (high tax alignment) or not (low tax alignment) plays an important role. Prior research finds that low tax alignment is associated with lower audit quality while high tax alignment is associated with high audit quality. This is caused by the fact that when a country has high tax alignment, the financial statements are examined closely by tax authorities and causes a higher detection risk of audit failure (DeAngelo, 1981; Van Tendeloo & Vanstraelen, 2008). Auditors are expected to be more conservative because they want to avoid litigation risks and any damage to their reputation and do so by providing high audit quality. High audit quality implies a constraint on the ability of management to practice accrual-based earnings management. As a response to the call of Van Tendeloo & Vanstraelen (2008) to use multiple proxies to measure audit quality, this study will use auditor size and auditor industry expertise at the national-level. Results from the regression show that accrual-based earnings management is indeed constrained by Big 4 auditors in a high tax alignment setting (Belgium). No significant results are found regarding auditor industry expertise and accrual-based earnings management. Furthermore, a sensitivity analysis using a subset of firms identified as suspect firms show the same significant results which further indicate the robustness of the results.
7
The second stage examines the relation between accrual-based earnings management and real earnings management. When accrual-based earnings management has been constrained due to high audit quality, the expectation is that management will find other ways of achieving their ‘management goals’. Another way to do so is through real earnings management, which is the manipulation of real activities. Real earnings management potentially imposes greater long-term costs on shareholders than accrual-based earnings management because it has negative consequences on future cash flows and might hurt firm value in the long run (Roychowdhury, 2006; Cohen et al., 2008; Cohen & Zarowin, 2010; Chi et al., 2011). Prior research on publicly traded companies has evidence supporting a substitution effect between accrual-based earnings management and real earnings management. In this study, I test whether this substitution effect is also present within privately held companies. Measurement of real earnings management is limited to the sales manipulation method. Following the regression, no evidence could be found regarding the relation between audit quality, accrual-based earnings management and real earnings management. Big 4 auditors constraining accrual-based earnings management does not show a significant increase in real earnings management using sales manipulation methods.
This study contributes to the knowledge on privately held companies and fills the knowledge gap that exists in the literature. This study contributes to prior literature by providing detailed information and a better view on the differences, incentives and practices of privately held companies regarding earnings management. Since privately held companies differ from publicly traded companies and constitute the majority of economies, it is important to provide further insight and guidance in this field. As opposed to prior research, which only documents the accrual-based earnings management aspect, this study includes real earnings management practices to give a better understanding of the effects of audit quality on accrual-based earnings management and real earnings management. The findings of this study provide insight and guidance to regulators, standard-setters, auditors, researchers and users of the financial statements that have an interest in privately held companies.
The remainder of the paper is organized as follows. Section 2 consists of a literature review on the differences between privately held and publicly traded companies, incentives for earnings management, earnings management itself and audit quality. Based on this literature, the hypotheses are developed. Section 3 introduces the sample selection and the research design. Section 4 discusses the results and finally, section 5 summarizes the most important findings and provides a conclusion.
8 2. Literature Review and Hypothesis Development
2.1 Differences between Privately Held Companies and Publicly Traded Companies
Both publicly traded and privately held companies within the European Union face the same accounting standards because accounting regulation is based on legal form (Ball & Shivakumar, 2005). It is known that privately held companies constitute the majority of the economy and researchers acknowledge the importance of providing research on privately held firms as well. Publicly listed companies are far less dominant, both in numbers and size, in various continental European economies (Vander Bauwhede & Willekens, 2004; Langli & Svanström, 2013). However, prior research has been focused on uncovering problems and phenomena regarding publicly traded companies. When looking at privately held companies independently, the literature is still in its early stages. This is mainly caused due to the fact that there is limited data available for privately held companies while there is plenty of data on publicly traded companies. The results derived from research based on publicly traded companies are difficult to generalize to privately held companies. Reasoning behind this is that privately held and publicly traded companies are fundamentally different. The most important differences lie in the fundamental structure and the way these companies deal with information asymmetry issues (Ball & Shivakumar, 2005; Coppens & Peek, 2005; Burgstahler et al., 2006). The following subparagraphs provide more insight on the differences between these companies.
2.1.1 Structure
The fundamental differences between privately held and publicly traded companies give reason to consider these two separately. The ownership, governance, financing, management and compensation structures of privately held companies differ from publicly traded companies (Ball & Shivakumar, 2005). Private companies are held more closely, have greater managerial ownership, and major capital providers take a more active role in management with insider access to corporate information (Van Tendeloo & Vanstraelen, 2008). Another major difference can be found when looking at the ability to generate capital from the general public (Burgstahler et al., 2006). Publicly traded companies have the ability acquire capital publicly by issue shares to the public while this is not allowed for privately held companies. The financial statements of privately held companies are therefore not widely distributed to the public because users of privately held companies’ financial statements are stakeholders other than equity investors. The stakeholders’ interests are protected by the requirement that
9
European companies that exceed certain size criteria are required to have their financial statements audited. Due to differences in the use of financial statements, it is suggested that the financial statements (and therefore the audits as well) fulfill different economic roles and the demand for quality varies (Ball & Shivakumar, 2005).
2.1.2 Agency Theory
According to Beatty & Harris (1998), earnings management occurs due to two related control difficulties, information asymmetry and moral hazard. These control difficulties occur when equity ownership is separated from the day-to-day operation of the corporation. The separation of ownership and control in a company leads to a divergence between the interests of the managers and the interests of shareholders. When information asymmetry problems exist between managers and shareholders, the managers have the opportunity to promote their own self-interest rather than the shareholders’ interests (Burgstahler et al., 2006). When a company has outside owners, which is the case with publicly traded companies, there is a separation of ownership and control. The principal (owner) hires the agent (manager) to run the company and gives the agent the authority to make its own decisions. Fama & Jensen (1983b) argue that managers, who are not major residual claimants but have the authority to make decisions, are likely to take actions that follow their own self-interest that deviates from the interests of shareholders. An explanation for this is that these managers are not the ones who bear the effects of their decisions and the risk is diversifiable for them.
2.1.3 Moral Hazard
The incentive for publicly traded companies to manage earnings should be greater than that for private firms because private companies are held more closely and have greater managerial ownership (Van Tendeloo & Vanstraelen, 2008). Beatty & Harris (1998) state that agency costs are expected to decrease as managerial ownership increases, which supports the view that privately held companies have less agency conflicts and, therefore, less incentive to manage earnings. Privately held companies are given as a solution to moral hazard. The interests of managers and owners in privately held companies are assumed to be better aligned and moral hazard lessens. Managers bear the consequences of their decision-making properly and are less inclined to take high risks because they are also the ones bearing the risks (Fama & Jensen, 1983a). Should outside ownership increase, the fraction of consequences and risks they bear would decrease again and provide incentive to act opportunistically and make
10
decisions in their self-interest (Schulze et al., 2003). Privately held companies are often companies that are owned by the founders and their families. These family firms reduce agency costs associated with monitoring agents because there is familiarity and the intimate knowledge gained from long association facilitate communication and cooperation among family owners and family agents. However, Schulze et al. (2003) claim that family firms do not necessarily resolve the agency problems experienced by privately held companies. They argue that interest alignment would be difficult to attain and sustain. Agency conflicts and moral hazard are therefore still pervasive within privately held companies and incentives for earnings management remain.
2.1.4 Information Asymmetry
Another important difference between privately held and publicly traded companies arises from capital market pressures. These companies face different demands for accounting information (Burgstahler et al., 2006). It is assumed that higher quality financial statements are required from publicly traded companies compared to privately held companies. This is primarily caused by the ways companies tend to communicate to their stakeholders to deal with information asymmetry. Users of financial statements demand information to evaluate and monitor the company. As was discussed earlier, information asymmetry is one of the difficulties that enable earnings management (Ball & Shivakumar, 2005).
Not all information is communicated through financial statements because managers are either not able to or not willing to do so when information is confidential or sensitive. Publicly traded companies have difficulty using private communication channels to address information asymmetry because they have an almost unlimited number of shareholders with high shareholder turnover (Ball & Shivakumar, 2005). It is inefficient for such companies to attempt to communicate with their stakeholders through private communication channels. Due to the fact that stakeholders continually change, the relation between company and stakeholders is more ‘distant’. The only way the information asymmetry between the stakeholders of public companies and the company themselves can be effectively dealt with, is through higher quality financial statements.
Privately held companies are in comparison held more closely, have a lower shareholder turnover, and have a more direct relation with shareholders because they take a more active role in management (Van Tendeloo & Vanstraelen, 2008). According to Ball & Shivakumar (2005), privately held companies communicate via private channels with their
11
shareholders, creditors, suppliers, customers and others, which reduce the demand for higher quality financial statements compared to public firms. The way companies communicate is typically associated with the structure of the firms themselves. Bharath et al. (2008) mention, for example, that debt-contracting differences between private firms and public firms arise because private firms are able to communicate privately with lender banks. The banks are not fully reliant on the published financial statements, which reduces the demand for financial reporting quality.
The demand for higher quality financial statements of publicly traded companies can be contributed to the fact that there is a larger separation between ownership and control. (Vander Bauwhede & Willekens, 2004; Van Tendeloo & Vanstraelen, 2008; Ball & Shivakumar; 2005; Burgstahler et al., 2006). According to Van Tendeloo & Vanstraelen (2008), agency conflicts are weaker in private firms compared to public firms because of the structure. Ownership and control are closer in private firms and affects the demand for financial statements for monitoring managers and the demand for a high quality audit (Fama & Jensen, 1983a). However, privately held companies are not always completely run by owner-managers which could still lead to agency conflicts and agency conflicts are not eliminated because banks and private firms are able to communicate privately. The absence of market-based measures of firm-value makes high quality reporting and high quality financial statements stay relevant for evaluation of managerial performance and to support personnel and compensation decisions (Van Tendeloo & Vanstraelen, 2008).
In summary, publicly traded companies have a higher demand for financial reporting quality due to the structure and information. A private company is held more closely and has less shareholder turnover and more active shareholders. The general consensus is that privately held companies have less agency conflicts and therefore less incentive for earnings management, but there are studies that contradict this conclusion so it is still not entirely certain. Privately held companies are also able to disclose more information through private communication channels while publicly traded companies do not have this option. The financial statements of both types of companies fulfill different economic roles (Burgstahler et al., 2006). Due to these differences, these companies also have different incentives and objectives when preparing their financial statements and reporting their earnings, which is discussed in the next paragraph.
12 2.2 Incentives for Earnings Management
Due to harmonization efforts in the European Union, the accounting standards for privately held and publicly traded companies are largely the same. The application of accounting standards, however, still involves judgment and leaves space for discretion (Collins et al., 2012). Managers are able to use their own knowledge about the business and its opportunities to select reporting methods, estimates, and disclosures. This discretion gives management the option to either make the reported earnings more informative about the firms’ economic performance or use these opportunities to manipulate the reported accounting numbers (Burgstahler et al., 2006). An example of the latter can occur when management may have incentive to mislead some stakeholders about the underlying economic performance of the company or to influence contractual outcomes that depend on reported accounting numbers (Healy & Wahlen, 1999). Both publicly traded and privately held companies can face situations where management uses judgment in financial reporting and in structuring transactions to modify financial reports.
Companies have various incentives for earnings management. Management may want to make sure they comply with debt covenants, meet earnings benchmarks or avoid reporting losses. Earnings management can also be applied to improve terms of trade between the firm and other stakeholders such as suppliers, customers, and employees (Coppens & Peek, 2005). The following paragraphs discuss important incentives for earnings management that are shared between publicly traded and privately held companies but also incentives that are primarily related to privately held companies.
2.2.1 Avoiding Small Losses and Earnings Declines
Generally, managers have incentives to avoid reporting losses or reporting declines in earnings. This comes from the observation that shareholders and creditors occur a higher disutility from an accounting loss of one dollar than they experience utility from an accounting profit of one dollar, which means that they might be averse to accounting losses (Coppens & Peek, 2005). This assumption is derived from prospect theory. Another explanation can be derived from the Transactions Costs theory, which predicts that stakeholders, such as customers, suppliers, creditors and employees use cut-offs at zero earnings or zero earnings changes to value their implicit claim on the firm and to determine the terms of transactions with the firm (Coppens & Peek, 2005).
13
earnings to mask or misstate economic performance. They overstate their reported earnings to achieve certain targets or meet certain benchmarks, such as analyst forecasts or prior year performance. Various studies have examined the distribution of reported earnings and a higher frequency of firms that reported barely positive earnings were observed and a lower frequency of firms with slightly negative earnings. Discontinuities around zero earnings and last year’s earnings are commonly found in prior literature for publicly traded companies (Healy & Wahlen, 1999; Burgstahler & Dichev 1997; Gunny, 2010). For example, irregularity around zero earnings is illustrated in the figure below.
Figure 1 - The distribution of annual net income. The horizontal axis show the Earnings Interval and the vertical axis named ‘frequency’ represents the number of observations in a given earnings interval (Burgstahler & Dichev, 1997). The interval widths are equal to 0.0025.
2.2.2 Tax Incentives
Financial reporting of privately held companies is less focused on informing users, this means that other underlying factors affect the financial reporting. One of these underlying factors is the tax income policy of these companies. Privately held companies’ reported earnings can be affected by their efforts to manage taxes (Sercu et al., 2002). Prior literature find that tax incentivizes privately held companies to engage in earnings management practices (e.g., Van Tendeloo & Vanstraelen, 2008; Vander Bauwhede & Willekens, 2004; Burgstahler et al, 2006; Ball & Shivakumar 2005).
Coppens & Peek (2005) identify two factors that determine how tax incentives influence financial statements. First, the dependability on financial statements in contracting and communication with stakeholders is negatively associated with the severity of tax
14
management because they are conflicting reporting incentives. Publicly traded companies are more dependent on financial statements as a means of communication for contracting and stakeholders and resolving information asymmetry. Tax incentives are therefore less likely to play a significant role for publicly traded companies. Burgstahler et al. (2006) find that because privately held companies are less dependent on using financial statements to communicate with their stakeholder groups, it is easier and has fewer consequences for them to diminish the informativeness of their earnings for tax purposes. Privately held companies will practice earnings management for tax purposes, if possible. Second, tax incentives play a bigger role when there’s a high alignment between tax practice and accounting practice (Vander Bauwhede & Willekens, 2004; Coppens & Peek, 2005; Van Tendeloo & Vanstraelen, 2008). In high tax-alignment settings, the financial reports of privately held firms are also the financial reports on which taxes are determined and their accounting choices are, therefore, often driven by tax determination.
In countries where tax alignment is high the financial statements are examined closely by tax authorities and results in a higher detection risk of audit failure. It is especially important for auditors in these countries, specifically the Big 4 auditors, to constrain the ability of private firms to manage earnings. Van Tendeloo & Vanstraelen (2008) find that privately held companies that are audited by Big 4 auditors in high tax alignment settings engage less in earnings management. An explanation is that Big 4 auditors cannot afford to face the high costs associated with audit failures. If audit failures were discovered at Big 4 auditors, it would have significant consequences such as reputation damage, litigation costs and loss of potential client-specific quasi-rents (Vander Bauwhede & Willekens, 2004). In a high tax alignment setting, tax authorities can basically be considered as a direct stakeholder and user of the financial statements, which increases the probability of audit failure detection because financial statements are more scrutinized.
2.3 Earnings Management
Managers can engage in earnings management through the manipulation of financial reporting figures when judgment is needed in accounting estimates and methods. This is only one way of engaging in earnings management and is called accrual-based earnings management. Managers, however, can also manage earnings through operational decisions. First off, accrual-based earnings management is achieved by manipulating the numbers without actually changing the actual cash flows related to the transactions being manipulated.
15
Accruals are the difference between the cash flows and net income and can be separated in discretionary accruals and non-discretionary accruals. Accrual-based earnings management occurs through managerial discretion. Examples of how accrual-based earnings management can be applied are: changing depreciation methods and estimates of provisions by applying different accounting policies, which do not affect the actual cash flows.
Real earnings management, on the other hand, is based on real activities manipulation. According to Zang (2012) it a purposeful action that is used to manage reported earnings in a particular direction, and is accomplished by changing the timing or structuring of an operation, investment, or financing transaction. It can be seen as a departure from normal operational practices. As was determined before, it is often used by managers to mislead stakeholders about the underlying economic performance of the firm. Examples of real earnings management are provided by Roychowdhury (2006), he mentions the act of decreasing R&D investments, price discounts and reductions, but also other examples such as over-production and general discretions in investments.
2.3.1 Audit Quality and Accrual-based Earnings Management
Research done on privately held companies, audit quality and earnings management have almost solely focused on accrual-based earnings management. Van Tendeloo & Vanstraelen (2008) examine the relation between audit quality and the earnings quality in private firms. In their research, they defined audit quality with the measure auditor size. Auditor size has been used consistently as a means of measuring audit quality (e.g., Vander Bauwhede & Willekens 2004; Burgstahler et al., 2006; Van Tendeloo & Vanstraelen 2008). The assumption that auditor size is related to audit quality is derived from the reputation rational concept of DeAngelo (1981). The concept explains that auditors suffer reputational damage and lose support from clients if their audit service practices are deemed as below-standards. This can lead to clients switching auditors, audit fees being renegotiated and thus an auditor will want to prevent bad publicity. The auditor size is directly related with this, the bigger the firm, the bigger the stakes and consequences are. Large auditor firms (Big 4 auditors), therefore, have greater incentive to provide audit quality as compared to small auditor firms (non-Big 4 auditors).
High tax alignment settings, where financial reporting also serves for tax determination, are therefore expected to lead to higher audit quality because of scrutinization of reporting by tax authorities. Lower tax alignment, which are associated with lower audit
16
quality, provide more ‘accounting flexibility’ and therefore more room for accrual-based earnings management (Becker et al., 1998). Van Tendeloo & Vanstraelen (2008) find evidence consistent with the aforementioned. Big 4 auditors , when compared to non-Big 4 auditors, generally engage less in accrual-based earnings management in countries with high tax alignment. This is consistent with the assumption that high tax alignment increases the probability of detection of audit failure and risks of litigation and causes Big 4 auditors to constrain the private firms’ ability to engage in earnings management. Coppens & Peek (2005) also examined earnings management practices in private firms. The results of this research also gave indication of a constraint in the capability of private firms to engage in earnings management in high tax alignment countries. Burgstahler et al. (2006) findd, contrary to recent allegations, that the audit quality in private firms compared to public firms is still considered low and therefore earnings management is still more pervasive in private firms than in public firms, even when outside market factors (such as auditors) constrain earnings management in high tax alignment countries.
Another important factor affecting accrual-based earnings management is the strength of legal systems and enforcement in countries. Countries with weaker legal systems and enforcement lead to more pronounced earnings management. The distinction between the strength of legal systems and enforcement will be discussed later on.
Besides auditor size, other factors have been identified as well which have an impact on audit quality, factors that are often used to identify audit quality are auditor industry expertise and audit fees. Auditor industry expertise on national-level is of concern in this study whereas audit fees will be left out due to data unavailability. Auditor industry expertise is identified as a critical indicator of audit quality (Reichelt & Wang, 2010). They found that auditors that were industry specialists (at both national and city-specific level) had clients with the lowest amount of abnormal accruals. This suggests that auditors’ network synergies and deep industry knowledge are important for delivering higher audit quality. Chi et al. (2011) include this in their research on real earnings management and Van Tendeloo & Vanstraelen (2008) suggested future research to include other measures of audit quality to contribute to prior research by widening the current limitations.
In summary, privately held companies have incentives for managing earnings through accruals. According to Burgstahler et al. (2006), audit quality when dealing with privately held companies appear to be lower than publicly traded companies but evidence of Big 4 auditors constraining accrual-based earnings management within high tax alignment settings is still found. Big 4 auditors do so in order to avoid any reputational damage and potential
17
negative consequences of having an audit failure. High audit quality leads to less discretion and accrual-based earnings management is constrained. The first hypothesis of this study is based on the aforementioned and is as follows:
H1: Big 4 auditors provide higher audit quality and constrain management’s ability to practice accrual-based earnings management in high tax alignment settings.
2.3.2 Audit Quality, Accrual-based and Real Earnings Management
Due to high audit quality, accrual-based earnings management is mostly constrained in private firms by Big 4 auditors. Prior literature has not looked at what kind of effect this has in private firms and real earnings management practices, while in public firms this could lead to a substitution effect (Cohen et al., 2008; Zang, 2012). In public firms, this substitution effect was mainly caused by the implementation of the Sarbanes-Oxley Act (SOX) in 2002. SOX is a new set of enhanced standards that resulted from the numerous accounting scandals that occurred in the past few decades. It was implemented to enhance corporate governance and audit quality for public firms, and had measures such as penalties for certain misconduct. Cohen et al. (2008) find a trend whereby post-SOX, the accrual-based earnings management decreased and is avoided due to the strict standards and consequences of misconducting behavior. Instead, companies start to shift from accrual-based earnings management to real earnings management. Zang (2012) also found a substitution effect between accrual-based and real earnings management. Chi et al. (2011) examine the association between enhanced audit quality and real earnings management. They find that due to auditors constraining the ability of firms to practice accrual-based earnings management, these firms substituted it with real earnings management. This is identified as an unintended consequence of higher audit quality and could even be harmful to the organizational value.
No prior research has been conducted about the relation between audit quality, accrual-based earnings management and real earnings management within privately held companies. This study will extend prior literature by analyzing the consequences of the limitation of accrual-based earnings management. The second hypothesis is focused on the consequences of the first hypothesis. Based on prior literature about publicly traded companies, we test for a substitution effect between accrual-based earnings management and real earnings management within privately held companies. If this effect is present within privately held companies, research has to find evidence separately because of the fact that it is
18
difficult to generalize results found when researching publicly traded companies. In contrast to what is expected within publicly traded firms, the expectation for the second hypothesis is that privately held companies do not see real earnings management as an option. It would be more likely to expect no substitution effect due to the fact that real earnings management potentially destroys firm value. Private companies are held more closely and their main incentives for earnings management seems to be derived from tax incentives Based on the literature discussed in this study, it would also not be logical for privately held companies to engage in real earnings management in the first place. Furthermore, another explanation on why real earnings management is not used by privately held companies is that real earnings management does not seem to be something that can be applied in order to satisfy the tax incentives that they have. It makes no sense to hurt your own company in order to pay less tax or to increase your earnings to pay more tax at the cost of future earnings. The second hypothesis can be stated as:
H2: Privately held companies do not substitute accrual-based earnings management with real earnings management when their ability to practice accrual-based earnings management is constrained.
3. Research Design
3.1 Sample
The sample used for this study is collected from the Amadeus database provided by Bureau van Dijk. The Amadeus database is the largest database which contains information on public and private companies from 43 countries, including all the EU countries. It is one of the few databases that provide private company financials in a standardized format for up to 10 years. Data regarding auditors, however, is only provided for the most recent year which is necessary information. The period of analysis is the most recent year for which data is available, which is 2013. Furthermore, the initial sample consists of all large and very large privately held companies from the Netherlands and Belgium. The reasoning behind selecting large and very large companies is that these companies are required by law to have their financial statements audited.
The selection of the Netherlands and Belgium is based on the fact that Belgium is considered a high tax alignment setting while the Netherlands is considered a low tax alignment setting (Van Tendeloo & Vanstraelen, 2008). As was previously discussed, the
19
distinction between these settings plays an important role for this study because tax incentives are presumed to affect the reporting behavior of private companies. In addition, these countries are also specifically selected to control for legal system or enforcement influence on earnings management. Institutional factors are found to have an influence on the reporting behavior of publicly listed and privately held companies (Ball, Kothari & Robin, 2000; Burgstahler et al., 2006; Leuz et al., 2003; Bushman et al., 2004; La Porta et al., 1998). La Porta et al. (1998) assembled a data set covering legal rules pertaining to the rights of investors, and to the quality of enforcement of these rules. The Netherlands and Belgium are, based on this data set, very similar when comparing the legal system or enforcement.
Another important modification is the exclusion of privately held companies that are actually subsidiaries of publicly listed companies. These subsidiaries are excluded because the parent company is able to control management and operations which can lead to reporting decisions that deviate from those of ordinary privately held companies (Burgstahler et al., 2006; Fenn, 2000; Van Tendeloo & Vanstraelen, 2008). Similar to prior research, this study also excludes banks, insurance companies and other financial holdings (SIC codes between 6000 and 6799), and public administrative institutions (SIC code 43 and SIC codes above 9000) due to the unique characteristics of these industries. The Amadeus database already excludes a large portion of the aforementioned industries.
Table 1 Sample Selection
Search Strategy The Netherlands Belgium Total
Data for Very Large + Large companies (2012-2013) 29,109 51,823 80,932 Excluding:
- Subsidiaries of publicly traded companies 27,101 47,116 74,217
- Missing data on auditors 4,673 28,920 33,593
- Publicly listed companies 4,673 28,728 33,401
- Financial institutions (SIC code 6000-6799), and public administrative institutions (SIC code 43 and SIC codes > 9000)
4,673 23,904 28,577
- Missing financial data and/or duplicates 3,342 9,880 13,222
Total unique firm-year observations 3,342 9,880 13,222
Full sample of observations in year 2013 1,671 4,940 6,611
20 3.2 Empirical Models
3.2.1 Accrual-based Earnings Management
The first stage of this study is to capture accrual-based earnings management. To determine the level of accrual-based earnings management, a model is used to estimate the measure discretionary accruals. Prior literature has used various models for estimating the discretionary accruals. This study considers a cross-sectional model use of the modified Jones model, which is believed to be able to effectively capture discretionary accruals. Dechow et al. (1995) evaluate the ability of alternative models (i.e. Healy model, DeAngelo model, Jones model & Industry model) to detect earnings management and find that the modified version of the model developed by Jones (1991) provides the most powerful tests of earnings management.
The Jones model implicitly assumes that revenues are nondiscretionary (Jones, 1991). This assumption is acknowledged as a limitation. It is more likely that discretion is exercised over the revenue recognition of credit sales than it is to manage earnings through exercising discretion over the revenue recognition of cash sales (Dechow et al., 1995). Therefore, the modified Jones model adjusts the change in revenues for the change in receivables. With this modification, the Jones model now implicitly assumes that changes in receivables result from earnings management. The results would not be biased towards zero anymore in a sample where earnings are managed through revenues.
The first step in measuring accrual-based earnings management is determining the total accruals. Consistent with prior literature (Burgstahler et al, 2006; Dechow et al., 1995; Healy, 1985; Jones, 1991; Van Tendeloo & Vanstraelen, 2008), total accruals (TACC) is estimated as follows:.
𝑇𝐴𝐶𝐶 = (Δ𝐶𝐴 − Δ𝐶𝐿 − Δ𝐶𝐴𝑆𝐻 + Δ𝑆𝑇𝐷𝐸𝐵𝑇 − 𝐷𝐸𝑃) (1) Where:
ΔCA = Current assets in year t less current assets in year t-1; ΔCL = Current liabilities in year t less current liabilities in year t-1; ΔCASH = Cash in year t less cash in year t-1;
ΔSTDEBT = Short term debt in year t less short term debt in year t-1; DEP = Depreciation in year t.
21
The estimated total accruals are then used to estimate the discretionary accruals cross-sectionally. The model is estimated at the 3-digit SIC industry level when more than 20 observations are available for an industry group, if not, the 2-digit (1-digit) SIC codes are used (See Appendix A). This is done to control for any industry effects on accruals. Each of the variables is scaled by lagged total assets in order to reduce heteroscedasticity in the residuals (Kothari et al., 2005).
TACCit Ait-1 = 𝛽0+𝛽1( 1 Ait-1) +𝛽2( ΔREVit-ΔRECit Ait-1 ) +𝛽3( PPEit Ait-1 ) +𝛽4( EBXIit-1 Ait-1 ) + ε (2) Where
TACC = estimated total accruals in year t (scaled by total assets at t-1); ΔREV = Sales in year t less sales in year t-1 (scaled by total assets at t-1);
ΔREC = Receivables in year t less receivables in year t-1 (scaled by total assets at t-1); PPE = Gross property plant and equipment in year t (scaled by total assets at t-1); EBXIt-1 = Earnings before extraordinary items and discontinued operations at year t-1 (scaled
by total assets at t-1), also referred to as ROAt-1;
At-1 = Lagged Total Assets (Total assets at t-1);
This model varies slightly from the original modified Jones model; it is augmented and includes a performance measure (ROAt-1) to adjust for performance in the accruals regression
(Cahan et al., 2011; Kothari et al., 2005).
The error term(ε), also called residual term, is used as an estimate of the discretionary accruals. The dependent variable of the empirical model is the estimation of absolute discretionary accruals |DACC|. The absolute values for discretionary accruals are used in further models due to the fact that the hypothesis is not concerned about any specific direction for earnings management. It is therefore unnecessary to look at whether discretionary accruals for income-increasing or income-decreasing earnings management.
The first independent variable included is the dichotomous variable TAX. This variable takes on the value of one if the observation is from a high tax alignment setting (Belgium) and zero if it is from a low tax alignment setting (the Netherlands). The next independent variable included is BIG4, this variable takes on the value of one if the company is audited by a Big 4 auditor and zero if it is audited by a non-Big 4 auditor. Consistent with Van Tendeloo & Vanstraelen (2008), an interaction term called BIG4*TAX is included.
Next, an independent variable EXP is included. This variable represents auditor industry expertise. Auditor industry expertise is calculated by determining the market share of
22
the auditor in an industry. The two-digit SIC codes identified earlier are too narrow to function as an indicator of auditor industry expertise. Utilizing industries based on two-digit SIC codes, which are indicators of major groups, would unnecessarily lead to a smaller sample due to the requirement of having more than 10 observations per industry while most of these can be retained by using a broader industry group. It is therefore necessary to define industry groups. An industry classification originally from Campbell (1996) and also used by Van Tendeloo & Vanstraelen (2008) is followed in the process of determining auditor industry expertise (See Table 4 for the industry classification). This industry classification is used to indicate broader industry groups as a base for determining auditor industry expertise. The original data collected from the Amadeus database provided core SIC codes which were three-digit. A transformation from three-digit to two-digit codes was therefore necessary to be able to make this industry classification. Next, auditor industry expertise is calculated as the total assets of an auditor’s clients in an industry divided by the total assets of all companies within that industry classification. The variable EXP takes on the value of one if this market share of the auditor is above 15%, and zero otherwise. The threshold of a market share of 15% is one that has been used before in Krishnan (2003).
The variables BIG4 and EXP are used as proxies for audit quality. Following the first hypothesis, the expectation is that Big 4 auditors provide higher audit quality in high tax alignment settings and therefore constrain the ability of management to practice accrual-based earnings management. Thus, the expected sign for the coefficient of the interaction term BIG4*TAX is negative.
Several factors can have an influence on the discretionary accruals, it is therefore important to include additional control variables. Besides using ROAt-1 to adjust for
performance during the previous regression to estimate accruals, another control variable is included afterwards, which is GROWTH (e.g. Ball & Shivakumar, 2005; Van Tendeloo & Vanstraelen, 2008). This control variable represents the change in sales. Next, as discussed before, earnings management practices can be incentivized by debt covenants. It is therefore reasonable to control for any leverage differences (e.g. Van Tendeloo & Vanstraelen, 2008; Watts & Zimmerman, 1990). This is done by including the control variable LEV, calculated as total liabilities divided by total assets in year t. Another control variable often used in prior research is the size of a company. The size of the company is associated with accounting choices and thus, the level of earnings management (Watts & Zimmerman, 1990). Prior literature on publicly listed firms for example, use the market value of equity or the Market-to-Book ratio but these measures are not available for privately held companies seeing as
23
there is no direct capital market available (e.g. Kothari et al., 2005; Cohen et al., 2008). For privately held companies, this is measured by total assets or the natural logarithm of total assets instead (e.g. Van Tendeloo & Vanstraelen, 2008; Cahan et al., 2011). As a result, the control variable LNASSETS (natural logarithm of total assets) is included.
Following the aforementioned, the determinants of accrual-based earnings management are examined by the following regression:
|DACC| = β0 + β1TAX + β2BIG4 + β3BIG4*TAX + β4EXP + β5LEV
+ β6GROWTH + β7LNASSETS + β11-22IND + ε (3)
3.2.2 Real Earnings Management
According to Roychowdhury (2006) there are three manipulation methods for real earnings management. The three methods are sales manipulation, reduction of discretionary expenses and overproduction. These proxies have been used in various researches such as Cohen et al. (2008), Chi et al. (2011) and Zang (2012).
Sales manipulation is the acceleration of the timing of sales and/or generating additional unsustainable sales. This is achieved though increased price discounts or more lenient credit terms (Roychowdhury, 2006). Managers can increase sales during the current period by offering price discounts and essentially accelerate the sales that would have occurred in the next period into the current period or achieve additional sales (Roychowdhury, 2006; Cohen et al., 2008). The additional sales will inflate current period earnings but lead to lower cash flows in the current period. Second, reduction of discretionary expenses such as advertising, R&D and SG&A expenses, will also boost current period earnings. Oveproduction could be used to spread fixed overhead costs over a larger number of units which would increase the profit margin and increase earnings.
Studies that have used these proxies are limited to publicly traded companies. This is most likely caused by the fact that not much data is available on privately held companies. This study tries to incorporate part of real earnings management practices on privately held companies and uses the sales manipulation method to detect real earnings management. The latter two methods (discretionary expenses and overproduction) are not included in this empirical model due to limited data availability. For example, Advertising, R&D and SG&A expenses are not commonly reported for privately held companies and neither are the costs of
24
goods sold. There are studies that deem the use of SG&A expenses only adequate as an estimation but even these expenses are barely reported consistently for privately held companies (Cohen et al., 2008). Consequently, the results will be limited to one of three manipulation methods documented in prior research.
Sales manipulation is measured by estimating the abnormal level of cash flow from operations (CFO). As was briefly mentioned before, CFO is calculated using the balance sheet approach: EBXI – TACC, where TACC is defined in formula (1) (Burgstahler et al., 2006). The cross-sectional regression is run at the 3-digit SIC industry level when more than 20 observations are available for an industry groupfor each industry, if not, the 2-digit(1-digit) SIC codes are used (See Appendix A). As was mentioned before, this is to control for industry effects. Each of the variables is caled by lagged total assets as well in order to reduce heteroscedasticity in the residuals (Kothari et al., 2005). The model is as follows:
𝐶𝐹𝑂𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1 = 𝛽1( 1 𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1) + 𝛽2( 𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1) + 𝛽3( 𝛥𝑆𝑎𝑙𝑒𝑠𝑖𝑡 𝐴𝑠𝑠𝑒𝑡𝑠𝑖𝑡−1) + 𝜀 (4)
Abnormal CFO (AB_CFO) is calculated as the actual CFO minus the normal level of CFO estimated using the coefficients from equation (4). AB_CFO is then used in the next regression to estimate the determinants of the level of real earnings management as the dependent variable. Obervations where the absolute value of ΔSALES/Assetst-1 exceeds a
threshold of 30% are excluded to control for extreme changes and stabalize the sample. The independent variables are similar to those used in the previous regression to determine the determinants of the level of accrual-based earnings management. The only difference is that ROA is used as a control variable after the abnormal-regression instead of before. See the previous paragraph for the description of these variables.
The expectation of the various variables are as follows. TAX is expected to have a positive sign if a substitution effect would be present. The expectation is that when accrual-based earnings management is constrained, privately held companies will substitute this with real earnings management. This is, however, only expected when the auditor is a Big 4 auditor, so the expected sign of BIG4*TAX is positive. EXP is also seen as a proxy for audit quality. Audit quality is expected to be negatively related with accrual-based earnings
management and thus, the expectation for the coefficients for terms related to EXP are also to be negative. Even when the auditor has industry expertise, it cannot prevent companies from
25
practicing real earnings management. If this was possible, then the expectations would be reversed. The determinants of the level of real earnings management are estimated by the following regression:
AB_CFO = β 0 + β1TAX + β2BIG4 + β3BIG4*TAX + β4EXP + β5LEV
+ β6GROWTH + β7LNASSETS + β8ROA + β9-20IND + ε (5)
4. Results
4.1 Descriptive Statistics
The descriptive statistics of the necessary variables to measure earnings management are presented in Table 2. Pearson and Spearman correlations are reported in Appendix B. Panel A of Table 2 shows the descriptive statistics of the variables used in the calculations. These descriptives provide insight on the characteristics of the complete sample. For this large cross-section of firms the influence of the few extreme observations is reduced. This is accomplished by winsorization. The extreme small and large observations are set equal to the values of less extreme small and large observations, respectively. The data has been winsorized at the 1st and 99th percentiles of their distributions. There is still, however, a large standard deviation when analyzing the variables for all firms. For example, when looking at total assets it is visible that the size of the companies in the sample is very diverse (Mean 101.019 with a standard deviation of 1,134.166). This is consistent with prior literature on privately held companies (e.g., Ball & Shivakumar (2005) report a mean of £419.9 with a standard deviation of 1041.3 when looking at privately held companies in the UK). The high level of standard deviation is mainly caused by the fact that privately held companies vary significantly in size and ranges from small family companies to large multinationals. It is therefore important to control for the effect of firm size when running the models (Watts & Zimmerman, 1990; Kothari et al., 2005; Cohen et al., 2008; Van Tendeloo & Vanstraelen, 2008; Cahan et al., 2011).
Panel B of Table 2 summarizes the descriptive statistics of the regression used to calculate discretionary accruals. As was mentioned before, the first step in doing so is by estimating the total accruals (TACC). The mean level for TACC is -1.244 with a standard deviation of 11.388. The expected sign and the level of standard deviation is consistent with what prior literature on publicly traded companies has found (e.g. Kothari et al., 2005 report a mean of -3.03 with a standard deviation of 11.62). The total lagged assets (At-1) has a lower
26
mean (96.710) when compared with total assets (101.019, this indicates a growth in total assets. When TACC is scaled by total lagged assets (TACC/At-1) the mean is -0.023 with a
standard deviation of 0.192, which is also similar to what is reported by Cohen et al. (2008) and Kothari et al. (2005). The variable |DACC| represents the absolute value of discretionary accruals estimated as the residuals from the regression. The absolute values for discretionary accruals are used in further models due to the fact that the hypothesis is not concerned about any specific direction for earnings management. The mean for |DACC| is 0.121 and corresponds with the statistics found in prior literature as well (e.g. Cohen et al. (2008) found an average for ABSDA of 0.11).
Panel C of Table 2 provides the descriptive statistics of the variables used in the regression for estimating the abnormal level of cash flow from operations. The variable CFO has a mean of 8.673 and is consistent when compared to Roychowdhury (2006). When scaled with lagged total assets (CFO/At-1), CFO is on average 0.066 or 6.6%. The mean of SALES/A
t-1 shows that sales are about 2 times the amount of lagged total assets and ΔSALES/At-1 is 0.037
or 3.7%. The variable AB_CFO is the residual of the regression and an estimation of the abnormal level of cash flow from operation scaled by lagged total assets, and has a mean of -0.001 with a standard deviation of 0.191.
Panel D of Table 2 shows the descriptive statistics on the independent variables used in the regression to test the first hypothesis. The mean of the variable TAX is 0.747 which shows that 74.7% of the sample is comprised of Belgian companies and 25.3% are Dutch companies. The mean of the variable BIG4 is 0.281, 28.1% percent of the companies are audited by a Big 4 auditor. This distribution is similar to the distribution of the sample used by Van Tendeloo & Vanstraelen (2008). BIG4*TAX displays that 9.5% of the sample is audited by a Big 4 auditor in a high tax alignment. The variable that measures auditor industry expertise on a national-level, EXP, has a mean of 0.054 and indicates that 5.4% of the sample is audited by a specialist.
Appendix B and C report the correlations tables regarding both accrual-based and real earnings management variables. The tables report a large amount of significant correlations but with low values. The correlation between BIG4 and TAX is -0.590 which seems high, this is due to the fact that ther are more BIG4 observations in the Netherlands. Variance Inflation Factors (VIF) are used to test for multicollinearity. No values above 10 are found for the variables in these tables. All values for VIF are below 4.52 (BIG4).
27 Table 2
Descriptive statistics for all firms (n= 6,615)
Variables Mean Standard Deviation Lower Quartile Median Upper Quartile
Panel A – General
Total assets 101.019 1,134.166 8.191 16.460 36.239
Property, Plant & Equipment 54.402 819.994 0.740 2.959 10.780
Current Assets 46.616 469.797 5.401 10.309 21.826 Receivables 21.463 228.907 1.859 3.964 8.897 Cash 7.645 67.824 0.181 0.803 3.127 Total Liabilities 59.334 728.254 4.479 8.963 20.556 Noncurrent Liabilities 26.042 471.705 0.072 0.737 3.328 Current Liabilities 33.293 343.256 3.487 6.895 14.596 Creditors 8.168 52.658 0.801 2.118 4.982 Sales 112.986 1,416.186 13.221 23.168 47.854 Depreciation 1.518 11.360 0.028 0.254 0.852 Net Income 4.575 69.764 0.029 0.434 1.468 Extraordinary Income -0.272 32.118 -0.000 0.000 0.004
Earnings before Extraordinary Items 4.846 68.857 0.026 0.437 1.455
CFO 8.673 186.926 -0.212 0.817 2.780
Earnings before Extraordinary items is calculated by subtracting extraordinary income (EBXI) from net income. CFO stands for cash flow from operation and is calculated using the balance sheet approach (Burgstahler et al., 2006). The calculation is as follows: EBXI – TACC (See panel B).
28 Table 2 (Continued)
Variables Mean Standard Deviation Lower Quartile Median Upper Quartile
Panel B – Estimation of Accruals
TACC -1.244 11.388 -1.794 -0.350 0.730 TACC/At-1 -0.023 0.192 -0.106 -0.024 0.050 At-1 96.710 1,054.707 8.126 16.084 35.788 1/ At-1 0.090 0.091 0.028 0.062 0.123 (ΔREV - ΔAR)/At-1 0.023 0.549 -0.116 0.004 0.140 PPE/At-1 0.525 16.817 0.066 0.212 0.462 ROAt-1 0.040 0.093 0.003 0.031 0.075 |DACC| 0.121 0.130 0.037 0.080 0.154
Panel C – Estimation of Cash flows from operations
CFO 8.673 186.926 -0.212 0.817 2.780 CFO/At-1 0.066 0.204 -0.016 0.057 0.153 1/ At-1 0.090 0.091 0.028 0.062 0.123 SALES/At-1 2.094 1.743 0.987 1.774 2.714 ΔSALES/At-1 0.037 0.569 -0.108 0.004 0.145 AB_CFO -0.001 0.191 -0.087 0.007 0.087
TACC is the total accruals calculated as the change in non-cash current assets minus the change in current liabilities excluding the current portion of long-term debt minus
depreciation (ΔCurrent assets - ΔCurrent liabilities – ΔCash + ΔShort term debt - Depreciation). TACC/At-1 is the total accruals scaled by lagged total assets (At-1). (𝛥𝑅𝐸𝑉 −
𝛥𝐴𝑅)/At-1 is the change in sales minus the change in accounts receivables (Modification of the Jones model) scaled by lagged total assets. PPE/At-1 is the gross property, plant
and equipment scaled by total lagged assets. ROA is the lagged return on assets calculated as earnings before extraordinary items divided by lagged total assets in year t-1.
|DACC| is the absolute value of residuals, which represent discretionary accruals, calculated with the Modified Jones model. CFO is the cash flow from operations and is
calculated as EBXI – TACC. CFO/At-1 is the cash flow from operations scaled by lagged total assets (At-1). SALES/At-1 is the sales level scaled by lagged total assets and
29 Table 2 (Continued)
Panel D – Determinants of Earnings Management
Variables Mean Standard Deviation Lower Quartile Median Upper Quartile
TAX 0.747 0.435 0.000 1.000 1.000 BIG4 0.281 0.450 0.000 0.000 1.000 BIG4*TAX 0.095 0.293 0.000 0.000 0.000 EXP 0.054 0.226 0.000 0.000 0.000 LEV 1.608 79.105 0.440 0.652 0.809 GROWTH 1.488 14.228 0.925 1.007 1.092 LNASSETS 2.967 1.253 2.103 2.801 3.590 ROA1 0.063 1.472 0.002 0.028 0.075
TAX is the dichotomous variable representing the tax alignment setting and takes on the value of one when there is high tax alignment and the value of zero when there is low
tax alignment. The BIG4 variable takes on the value zero if the auditor is a non-Big 4 auditor and one if it is. EXP is an indicator for auditor industry expertise and takes on the value one if the industry market share of an auditor is more than 15%. BIG4*TAX is an interaction term between the variable BIG4 and TAX is calculated by multiplying the interacting variables. LEV is a control variable which measures the leverage of a firm, calculated as total liabilities divided by total assets. GROWTH represents the growth, or change of sales and LNASSETS, which is the natural logarithm of total assets, controls for firm size.
1ROA is another control variable included in the regression for determinants of Real earnings management and is not used in the regression for determinants of accrual-based
earnings management due to the fact that it has already been included when estimating the discretionary accruals (Performance-Adjusted Modified Jones model). The other variables are similar.
