UNIVERSITY OF AMSTERDAM
Amsterdam Business School
Faculty of Economics and Business
Earnings Conservatism
Impact of the 2007-2008 Financial Crisis
in common-law and code-law countries
Master Thesis
Student name: D. M. de Vogel
Student number: 10628320
Program: MSc Accountancy & Control
Academic Year: 2014-2015
Supervisor: dr. A. Sikalidis
Final version: June 19th 2015
II
Statement of Originality
This document is written by student David de Vogel, who declares full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
III
Abstract
This paper examines whether firms engaged in more earnings conservatism during the 2007-2008 Financial Crisis and whether this strategy was strengthened by the legal system under which the particular firm reported. Earnings conservatism is interpreted as the faster recognition of ‘bad news’ in earnings than ‘good news’. To control for earnings management, a measure of unmanaged earnings is used to measure earnings conservatism. I expect that earnings conservatism is more pronounced in the crisis period and more pronounced for common-law countries. Furthermore, I expect that the effect of the 2007-2008 Financial Crisis on earnings management is stronger for common-law countries. Contradictory to my expectations, I find that the crisis had a significantly negative impact on earnings conservatism. Results reveal that this finding is largely driven by Japan which is overrepresented in the sample. Furthermore, as expected, I find that common-law countries indeed report more conservative earnings. However, this does not strengthen the effect of the crisis on earnings conservatism but rather partly offsets this effect. These findings have implications for accounting research, accountants, financial analysts, students of corporate governance, and standard-setters.
Keywords: earnings conservatism, earnings-return relation, 2007-2008 Financial Crisis,
IV
Preface
With this master thesis, I complete the master program Accountancy & Control at the Faculty of Economics and Business of the University of Amsterdam.
It was important to me, to carry out a research on a topic that really interested me. For long time, I could not find such a topic, or, when I found it, lots of previous research had already been done on that topic. I was glad to come across the subject of earnings conservatism. I am sincerely interested in this subject, because it acknowledges that not everything can be reported precisely. Furthermore, it acknowledges that not everything is reported precisely, because of several reasons, e.g. managers not being fully rational. This contradicts with the efforts of auditors to have annual reports that present a true and fair view. Combining the topic of earnings conservatism with the 2007-2008 Financial Crisis and institutional characteristics makes this topic even more interesting. It’s fascinating and at the same time shocking to see how a crisis can be the cause of major companies going default and how these effects are different across countries. In my opinion I succeeded in combining these subjects, hereby making this thesis, at least for me, an interesting thesis.
During the writing of this thesis, I really appreciated all those who supported me in any way. First of all, my supervisor Alexandros Sikalidis, who assisted me during the writing of this thesis by providing helpful feedback. Furthermore, I would like to thank Deloitte and especially Niels Adegeest for granting me the opportunity to write my thesis at the office at The Hague, since this provided me with a glimpse of my future job as an auditor. Besides, I would like to thank all those who I did not mention, but who helped me to write this thesis.
Although I finished my thesis today, I will continue challenging myself in the future with everything I learn. To put it in the words of Winston Churchill: “This is not the end. It is not even the beginning of the end. But it is, perhaps, the end of the beginning”.
Koudekerk aan den Rijn, June 19th 2015, David de Vogel
V
Abbreviations and list of tables
Abbreviations
FASB Financial Accounting Standards Board GAAP Generally Accepted Accounting Principles IASB International Accounting Standards Board
NPV Net Present Value
PAT Positive Accounting Theory
SEC Securities and Exchange Commission
SIC Standard Industrial Classification
List of tables
Table 1 Sample
Table 2 Industry characteristics Table 3 Descriptive statistics Table 4 Correlation matrix
Table 5 Coefficients and adjusted R2s (%) from pooled cross-sectional regressions Table 6 Results using Basu model (1997)
Table 7 Impact of the 2007-2008 Financial Crisis on earnings conservatism Table 8 Influence of the legal system on earnings conservatism
VI
Table of Contents
1. Introduction ... 1
2. Literature review and hypotheses ... 4
2.1 Definitions of conservatism ... 4
2.2 Explanations for conservatism ... 5
2.2.1 Contracting explanation ... 6
2.2.2 Litigation explanation... 8
2.2.3 Income tax explanation ... 10
2.2.4 Regulatory explanation... 10
2.3 Types of accounting conservatism... 10
2.4 Conservatism over time ... 11
2.5 Conservatism and economic crises ... 13
2.6 Accounting conservatism and differences between countries ... 15
2.7 Earnings management ... 18
2.7.1 Definitions of earnings management ... 18
2.7.2 Real earnings management ... 19
2.7.3 Accrual based earnings management ... 19
2.7.4 Relationship between earnings management and earnings conservatism ... 20
2.8 Hypotheses development ... 21
3. Research design... 24
3.1 Sample selection ... 24
3.2 Measuring earnings conservatism ... 25
4. Results and analysis ... 34
4.1 The influence of earnings conservatism on unmanaged earnings ... 34
4.2 The impact of the 2007-2008 Financial Crisis on earnings conservatism ... 37
4.3 The influence of a legal system on earnings conservatism ... 39
5. Conclusion and discussion ... 43
1
1. Introduction
The 2007-2008 Financial Crisis which erupted in the summer of 2007 in the United States, has had a devastating impact on the global economy. Due to the escalation in the early fall of 2008, economies around the world were thrown in a recession, one after the other. In the United States large financial institutions like Lehman Brothers and Bear Stearns went bankrupt or, in the case of for example AIG, Fannie Mae and Freddie Mac, were acquired by the U.S. government. Governments worldwide took actions to improve liquidity and solvency in the financial sector. Though nowadays, many countries and industries are still not back at the pre-crisis level (Harding, 2013, November 13). It is hard to determine what the Financial Crisis has cost society, because the effects can be seen worldwide, and since the world is still in the aftermath of the Financial Crisis. Estimates for just the U.S. economy vary from 6 to 22 trillion dollar (Luttrell et al., 2013; U.S. Government Accountability Office, 2013).
However, the Financial Crisis did not only have an impact on economies, it had an impact on accounting as well. Research (e.g. Kothari and Lester, 2012; Vyas, 2011) is now emerging on the role of financial reporting during the Financial Crisis and how this role changed due to the Financial Crisis. For example, Kothari and Lester (2012) discuss the role of fair value accounting leading up to and during the Financial Crisis and Vyas (2011) discusses the timeliness of accounting write-downs by U.S. financial institutions during the financial crisis.
Literature on earnings conservatism is extensive as well, yet is does not analyze the impact of the 2007-2008 Financial Crisis on accounting. However, circumstances during a financial crisis are very different from a normal setting and therefore, the amount of conservatism applied in accounting can also be very different.
Although no previous research has been performed on the impact of the 2007-2008 Financial Crisis on accounting conservatism, Francis et al. (2013) investigated the influence of conservatism on shareholder value during the 2007-2008 Financial Crisis. They find that conservatism had a significantly positive and economically meaningful effect on firm stock performance during the financial crisis. Furthermore, the authors argue that conservatism is useful for shareholders in reducing information risks and agency problems. Moreover, previous research (e.g. Herrmann et al., 2008; Vichitsarawong, 2010) has addressed the impact of other crises on earnings conservatism. Likewise, without studying a specific crisis, Jenkins et al. (2009) demonstrate in their study that conservatism is higher during periods of
2 economic contraction compared to periods of expansion. However, these results are contradictory with research by Herrmann et al. (2008) and Vichitsarawong et al. (2010). The latter two studies focused on the impact of the 1997 Asian financial crisis and find that firms reported less conservative during this crisis. Above studies demonstrate that there is no general consensus on the impact of a financial crisis on earnings conservatism. This implies that more research in this field should be pursued.
One of the possible explanations for this lack of general consensus can be found in the influence of institutional characteristics of a specific country on the amount of conservatism applied. After all, the studies of Herrmann et al. (2008) and Vichitsarawong et al. (2010) focused solely on Asian countries. Even more, Ball et al. (2000) found that accounting income in common-law countries exhibits significant greater timeliness than code-law accounting income. Common-law countries are often characterized with the ‘shareholder’ model of corporate governance, whereas code-law countries are characterized with the ‘stakeholder’ model of corporate governance. This could be a reason why previous research is contradictory.
A further analysis will be performed in this study by investigating whether there is a difference between common-law and code-law countries and whether this difference strengthens the impact of the crisis on earnings conservatism. The research question resulting from this is: What was the impact of the 2007-2008 Financial Crisis on earnings conservatism and is this impact different for common-law countries compared to code-law countries?
I hypothesize that the 2007-2008 Financial Crisis has increased the amount of earnings conservatism applied by firms, because of increased information asymmetry between managers and other parties, including shareholders, debt holders and government. Furthermore, I hypothesize that the effect of the 2007-2008 Financial Crisis on earnings conservatism is strengthened by the institutional characteristics of a specific firm.
Contradictory to my expectations, I find that earnings conservatism significantly decreased during the 2007-2008 Financial Crisis. However, further analysis reveals that this result has to be interpreted with caution. Furthermore, I find that common-law countries report more conservative earnings and that earnings conservatism increased during the sample period. The results indicate that this increase partly offsets the decrease resulting from the crisis.
The results of this study lead to a deeper understanding of accounting practices and show how earnings conservatism differs across countries. These results have implications for accounting research, since this research examines the determinants of conservatism. The
3 finding that common-law countries are more earnings conservative has implications for accountants, analysts and students of corporate governance. They should consider that differences over time or between countries have an impact on financial reporting. Furthermore, this research has implications for standard-setters, because both IASB and FASB try to eliminate conservatism in financial reporting because it is not in line with the ‘true and fair view’ that financial reporting should aim to represent.
To improve the comprehensibility of this topic, it is important to delineate the subject properly. Therefore, this thesis will consist of five sections. The next section mainly consists of an extensive literature review in which relevant literature is discussed. Furthermore, in this part the conservatism principle will be defined as well as the reasons why earnings conservatism exists. Section two also develops the testable hypotheses. The third section describes the research design, followed by the sample selection process and a description of the data. The results and an analysis of the results will follow in the fourth section. Section five provides a discussion of the results and a conclusion.
4
2. Literature review and hypotheses
This study examines the influence of exogenous shocks on earnings’ reporting. It focusses especially on the impact of the 2007-2008 Financial Crisis. Accounting conservatism is used as a starting point, since it has proven to have a large influence on reported earnings (e.g. Basu, 1997; Watts, 2003a). After several definitions of conservatism are given, this chapter further elaborates on the explanations why conservatism exists. Furthermore, the developments over time, as well as the development of conservatism during crises will be discussed, followed by the differences between countries. The literature review will continue by focusing on earnings management. Emphasis will be placed on accrual-based earnings management, since Roychowdhury and Watts (2006) and Pae (2007) argue that conservatism is reflected mostly in the accrual component of earnings.
2.1 Definitions of conservatism
Several definitions have been developed to describe the same phenomenon. According to Bliss, as cited in Basu (1997), the general definition of conservatism is to anticipate no profit, but to anticipate all losses. In reality, this implies that all expenses and liabilities are recognized immediately when there is an uncertainty about the outcome and to only recognize revenue and assets when it is certain that they are being received. Basu (1997, p. 7) interprets this definition of conservatism as: “accountants’ tendency to require a higher degree of verification to recognize good news as gains than to recognize bad news as losses”. Basu’s findings point in the same direction. The author finds that the incorporation of bad news in accounting earnings is much timelier than good news. Watts (2003) reaches the same conclusion, since he states that the verification requirements for gains and losses are asymmetrical. This implies that gains require a much higher degree of verification than losses.
Both the Financial Accounting Standards Board (FASB) and the International Accounting Standards Board (IASB) describe conservatism. In the Statement of Financial Accounting Concepts No. 5, the FASB states that:
“[…] as a reaction to uncertainty, more stringent requirements historically have been imposed for recognizing revenues and gains than for recognizing expenses and losses, and those conservative reactions influence the guidance for applying the recognition criteria to components of earnings”. [emphasis added]
5 The IASB describes accounting conservatism by using the word prudence (IASB, 1999). Especially article 3.19 of the statement of principles for financial reporting provides a clear definition of what prudence is.
The above quotations indicate that, according to both FASB and IASB, accounting conservatism is mainly concerned with the higher verification requirement for recognizing revenues and gains compared to liabilities and losses. This corresponds to the definitions provided earlier. However, there is a difference between the definition of the FASB and IASB. Where the IASB focusses on both conservatism in the balance sheet and income statement, FASB seems to pay attention solely to earnings conservatism. This is not surprising, since Basu (1997), by making a reference to the Accounting Research Bulleting No. 2 1939, states that conservatism in the income statement is far more significant. Therefore, in this research, tests are conducted on earnings conservatism unless otherwise indicated.
2.2 Explanations for conservatism
According to the IASB (1999, p. 16), the objective of financial statements is “[…] to provide information about the reporting entity’s financial performance and financial position that is useful to a wide range of users for assessing the stewardship of the entity’s management and for making economic decisions”. This information has to be both relevant and reliable in order to be useful for financial statements users. Information can only be relevant when it is timely. Ball and Brown (1968) did an empirical evaluation of accounting income numbers. They find that 85 to 90 per cent of information in the annual income report is already known before the report is issued. In this case, the annual report is less timely, because almost all the
“Prudence is the inclusion of a degree of caution in the exercise of the judgements needed in making the estimates required under conditions of uncertainty, such that gains and assets are not overstated and losses and liabilities are not understated. In particular, under such conditions it requires more confirmatory evidence about the existence of, and a greater reliability of measurement for, assets and gains than is required for liabilities and losses”. [emphasis added]
6 information is already incorporated in the stock price. Conservatism also decreases the value relevance of accounting information, because good and bad news are not incorporated at the same time.
Several studies (e.g. Watts, 2003a, 2003b; Qiang, 2007; LaFond and Watts, 2008; Khan and Watts, 2009) have tried to explain the existence of accounting conservatism and found four main explanations, namely contracting, litigation, taxation and regulation. These four explanations will be further elucidated in the subsections below.
2.2.1 Contracting explanation
Jensen and Meckling (1976) argue that a firm can be seen as a nexus of contracting relationships between different parties. All those parties have different information, consequently causing a moral hazard. This is due to both the limited horizon of management because of their limited tenure with the firm, and the limited liability of managers. Managers can forgo positive Net Present Value (NPV) projects with negative earnings in the short term and positive earnings in the long term when the manager has left the firm. Furthermore, managers have an incentive to overstate earnings to obtain a higher compensation. By requiring a higher verification for good news, conservatism expands the horizon of the manager and limits his opportunistic behavior. Moreover, according to Watts (2003a), delaying the recognition of profits prevents managers from focusing solely on projects with a short term positive NPV. In addition, Watts claims that debt contracts often include earnings-based formulas. By doing this, the amount of dividend payments is restricted and a minimum of net assets within the firm is maintained to meet obligations of lenders.
In addition to issues with timeliness described above, both Watts (2003a) and LaFond and Watts (2008) describe the attribute of verifiability. Both studies argue that it is difficult to make accounting measures timely, because the information needed for this is hard to verify. Arguably, the best option would be to have a performance measure which includes both the future net cash flows from current operations as well as the net cash flows from new products. However, although this measure is timely, it is often near impossible to verify, and verification is necessary for the enforcement of a contract in a court of law.
Nevertheless, gains require a higher degree of verification than losses. This is called asymmetric verifiability (Watts, 2003a). This is largely due to the asymmetric payoffs from contracts that relevant parties within the firm have. This asymmetric payoff exists in both
7 debt contracts and executive compensation contracts. Debt-holders are primarily interested in getting their money back. Therefore, they demand timelier information for bad news compared to good news. When the loan is at maturity, the value of the assets has to be higher than the face value of the debt. However, when the value of the assets is lower than the face value of the debt, the maximum debt holders will receive is the value of all the net assets of the firm, which might be less than the contracted sum. Nonetheless, when the value of the firm’s net assets is far above the face value of the debt, debt holders will not receive additional compensation. The payoff is therefore asymmetric. These asymmetric payoffs exist in executive compensation contracts as well. Managers usually have more information about the future of the firm than the shareholders, auditor or board of directors. When information is not verified, a manager can increase compensation by biasing estimates of future cash flows upwards (Watts, 2003a). LaFond and Watts (2008) argue that this is especially the case when the manager has a limited tenure or limited liability. Following the same line of reasoning, in absence of conservatism, a dividend problem is likely to occur. In this case, shareholders may receive excess dividends which make the firm unable to repay its debts. The limited liability of the shareholders leads to an impossibility for the debt holder to recover these excess dividends. Without other drivers to report conservative, debt holders have to enforce the firm to report conservative. They can do this through conservative covenants in debt contracts (Beatty et al., 2008). Beatty et al. (2008) describe two modifications in their study which can be enforced by debt holders; income escalators and exclusion of purchased intangibles. When a certain percentage of positive net income is excluded from covenant calculations, this is called an income escalator. This reduces the slack of the company to distribute the positive net income in a way that suits them best, instead of the debt holder. The other way around, the threshold is not adjusted in case the borrower faces a loss, so covenant slack is fixed to the full amount of the loss. The exclusion of purchased intangibles is described in greater detail by Watts (2003b). Watts argues that debt contracts stimulate conservatism when GAAP requires that gains and assets are recognized when they are not verifiable, for example in the case of goodwill. Another possibility arises when GAAP requires the recognition of earnings to which the firm has no immediate legal claim, for example the earnings arising from the equity method of accounting for unconsolidated subsidiaries.
Firm governance is the last contracting attribute described by Watts (2003a). In order to prevent themselves from being fired, managers have an incentive to hide losses. Shareholders might push the Board of Directors to dismiss a manager when he executed
8 projects with a negative NPV or reports losses. With asymmetric verifiability, recognition of losses is speeded up. Moreover, it provides the Board of Directors and shareholders with an opportunity to investigate those losses. This explanation is supported by research of Francis et al. (2013), who find that the relation between conservatism and stock returns depends on the governance and information environments of firms.
For several centuries, contracting is used in accounting (Watts and Zimmerman, 1983). However, contracting is the most substantiated explanation for accounting conservatism in literature as well (e.g. Ahmed et al., 2002; Qiang, 2007; LaFond and Roychowdhury, 2008). For example, Ahmed et al. (2002) argue in their study that firms report more conservative when they have conflicts with debt-holders or shareholders to reduce the likelihood of litigation. Moreover, debt-holders appreciate conservative reporting firms and are therefore more likely to decrease their interest rates, which lowers the cost of debt. However, as argued by Bushman and Piotroski (2006), contracting can only lead to conservatism when this contract is enforced.
In summary, because all the parties within a firm have asymmetric information, asymmetric payoffs, and sometimes limited liability, the contracting explanation for accounting conservatism applies to most uses of accounting within the firm. The welfare of the manager depends on the way in which he reports the results of the firm in the financial statements. Therefore, the manager has incentives to bias, id est overstate, the results. This has an effect on his tenure at the firm, his compensation, the stock price of the firm, and the value of his stock options. By demanding a higher verification for gains compared to losses, the net assets and cumulative earnings are less likely to be overstated. This will make distributions that violate contracts and reduce the value of the firm less likely (Watts, 2003a).
2.2.2 Litigation explanation
The preceding attribute of contracting mainly discussed the reasons for managers to overstate the assets and gains, and to understate the liabilities. However, according to Watts (2003a), next to contracting, shareholder litigation is another major driver for conservatism. Though, unlike the contracting explanation, the litigation explanation is more recent. Shareholders can sue a firm when the information provided by the firm is misleading and shareholders suffer from the consequences of this false information. These costs of securities litigation are often substantial, so managers have an incentive to prevent their firm from being litigated. To
9 reduce litigation costs, a firm can decide to report more conservative financial statements. Graham et al. (2005) studied how managers justify their choice. They found that managers are willing to sacrifice value in the long term, in order to meet a short-term target. This implies that managers sometimes take litigation costs for granted to improve their current performance. Ball et al. (2000) make the inference that the incentive to follow accounting standards is largely dependent on the penalties resulting from not adhering to those standards. Since not all countries have the same penalties and enforcement institutions, conservatism varies between countries. In their study, the authors classify countries as either common-law or code-law. Countries with high political influence are classified as code-law systems, whereas common-law countries are characterized by accounting practices which are determined primarily in the private sector.
Watts (2003a, 2003b) argues that conservatism has a negative effect on shareholder litigation costs, because conservatism leads to assets and earnings which are more likely to be understated instead of overstated. Shareholders are more likely to sue managers and auditors when assets and earnings are overstated, in order to recover their losses. This is supported by research of Lys and Watts (1994) and Kellogg (1984). The latter study finds that in securities litigation, buyers start a lawsuit thirteen times more often than sellers. Furthermore, DuCharme et al. (2004) find that auditors have an interest in financial reporting being conservative, because in the cases where firms were accused of earnings inflation, the auditor was the codefendant in almost every case. This is supported by Basu (1997), who finds that the increases (decreases) in conservatism have coincided with increases (decreases) in the legal liability exposure of auditors.
Based on previously discussed literature, it can be concluded that both managers and auditors adopt a cautious approach when ex ante their firm has a higher litigation risk. As mentioned before, recognizing bad news as losses requires less verification than recognizing good news as gains.
Occasionally, political cost is a driving factor for litigation costs. Firms with exceptional profits draw attention to themselves from government or other stakeholders (Deegan and Unerman, 2006). They want to know whether this is due to exorbitant prices or due to production methods that harm the environment. Furthermore, labor unions will demand a salary increase. Therefore, reporting conservative has a negative effect on the litigation costs from a political cost perspective.
10
2.2.3 Income tax explanation
A third explanation for conservatism is the income tax explanation. Taxable income is dependent on reported earnings. Therefore, it has an influence on the calculation of earnings (Watts, 2003a). Shackelford and Shevlin (2001) argue that taxes provide an incentive for firms to conform their reported accounting income to taxable income when their reported accounting income is higher. At the same time, there is a financial reporting incentive to increase book income. According to the authors, firms changed their accrual behavior to conform book and tax. Deferring income leads to lower reported earnings, but this also reduces taxable income, leading to lower taxes in the current period.
2.2.4 Regulatory explanation
According to Watts (2003a), regulation is the fourth explanation for conservatism. Watts (1977) states that it is easier for politicians to attach blame to losses due to actions than to losses due to inactions. Therefore, he expects a bias in the regulation of disclosure of corporate financial statements. Ex post, it is easy for politicians to accuse the managers and auditor of the firm that the assets were overvalued. On the opposite, undervaluing the assets does not lead to a political crisis, because it is difficult for outsiders to observe losses due to undervaluation. Consequently, management and auditors are more likely to undervalue the assets in corporate financial statements.
2.3 Types of accounting conservatism
Two types of accounting conservatism can be distinguished (García Lara and Mora, 2004). The asymmetric timeliness in incorporating gains versus losses as described by Basu (1997) is defined as ‘earnings conservatism’ (Ball et al., 2000). Although related, this concept is different from ‘balance sheet conservatism’, which is defined by Ball et al. (2000) as the understatement of assets and/or overstatement of liabilities to report a low book value of equity.
Conservatism can also be divided into conditional and unconditional conservatism (Beaver and Ryan, 2005). Qiang (2007) states that the form of conservatism is dependent on the moment when it is applied. When it is applied after difficult-to-verify news occurs, than it is called conditional conservatism. The main criterion is that bad news is recognized in a
11 timelier fashion than good information. Examples of this include impairment accounting for long-lived assets and the lower of cost or market accounting for inventory. In the event of impairment, the book value of the asset is written down. However, when the value of an asset increases, the book value is not written up. The incorporation of value changes is therefore asymmetric. Since this type of conservatism occurs after a certain event, conditional conservatism is sometimes referred to as ex post conservatism or news dependent conservatism (Beaver and Ryan, 2005). Unconditional conservatism, also referred to as asymmetric timeliness (Basu, 1997), asymmetric verification (Watts, 2003), and ex ante conservatism or news independent conservatism (Beaver and Ryan, 2005), is applied before difficult-to-verify news occurs. Jenkins et al. (2009, p. 1044) puts it as “not conditioned on the economic reality”. This implies that the firm reports a lower book value than the estimated fair value during the lifetime of an asset, because it is difficult to give a correct estimation of the value of the asset. This type of conservatism includes i.a. expensing research and development costs and using accelerated depreciation methods.
Research of Qiang (2007) has shown that not all of the explanations described in paragraph 2.2 are applicable to both forms of conservatism. The litigation explanation is the only explanation which causes both conditional and unconditional conservatism. Contracting causes conditional conservatism, regulation and taxation induce unconditional conservatism.
2.4 Conservatism over time
Conservatism is ubiquitous in financial reporting. However, the reason of this presence is not so obvious. In the past, auditors regularly ignored conservatism, because it was not in line with the ‘true and fair view’ that financial reporting should aim to represent. Nevertheless, Basu (1997) argues that conservatism has had an influence on accounting practice and theory for over 500 years. Previous studies (e.g. Basu, 1997; Givoly and Hayn, 2000; Holthausen and Watts, 2001) suggest that accounting practice has become more conservative over the last decades. This is remarkable, since the IASB and FASB, but also many capital market regulators, consider conservatism as unfavorable or at least should be applied with caution (e.g FASB, 1980; Levitt, 1998).
In a speech at the New York University center for law and business, former chairman of the SEC Arthur Levitt describes five ways in which conservatism is misused (i.a. “Big Bath” charges and “cookie jar reserves”), making it earnings management (Levitt, 1998).
12 Likewise, FASB has its reasons against conservatism. FASB is not necessarily against conservatism, because business and economic activities are sometimes uncertain. However, they state that conservatism has “to be applied with care” (FASB, 1980, p. 23). As noted earlier, possible errors in measurement are preferred to be in the direction of understatement rather than overstatement. The main criticism for the FASB is that this introduces bias into financial reporting, because it conflicts with the qualitative characteristics of representational faithfulness, neutrality, and comparability. Although contradictory, several studies (e.g. Basu, 1997; Givoly and Hayn, 2000; Holthausen and Watts, 2001) find that conservatism increases under formal standard-setting regimes, with the FASB as most frequently mentioned. Holthausen and Watts (2001) suggest that this does not imply that the standards caused conservatism, but give as possible explanation that the implementation of those standards by preparers and auditors affect the degree of conservatism.
Basu (1997) studied differences in conservatism in the period from 1963 to 1990 for firms listed in the U.S. The author divided this timeframe in periods of high and low legal liability and finds that the amount of conservatism applied in financial reporting is associated with these periods. Periods of high legal liability indicate an increase in the amount of conservatism applied in financial reporting. This conclusion is consistent with the explanation of Watts (2003a) that managers use conservatism to reduce potential costs arising from litigation.
The results from Basu (1997) indicate that conservatism emerged around 1970 and subsequently increased. It is quite remarkable that Basu does not find significant evidence of conservatism in the period before 1970. However, Holthausen and Watts (2001) argue that conservatism already existed prior to both 1970 and formal standard setting in the U.S. The authors indicate that conservatism before 1970 was probably due to the contracting explanation and the income tax explanation. They confirm the result of Basu (1997), by stating that until the 1960s, litigation was not a major factor. Furthermore, Holthausen and Watts (2001) argue that conservatism has increased to the point where bad news drives the association between earnings and stock prices.
The developments in conservatism over time have also been investigated by Givoly and Hayn (2000) using a sample spanning from 1950 to 1998. The authors investigate the changing time-series properties of earnings, cash flows and accruals to find out whether financial reporting has become more conservative. The authors find that profitability generally declined over the last four decades, without a corresponding decline in cash flows. Moreover, the authors state that their evidence shows that negative non-operating accruals
13 were massively accumulated over this period. Furthermore, the authors argue that the distribution of earnings has become more dispersed and negatively skewed compared to the distribution of cash flows. In line with Basu (1997) and Watts (2003a), the authors claim that their evidence suggests an asymmetric timeliness in recognizing ‘good news’ relative to ‘bad news’. These results are in line with conservatism increasing over time.
2.5 Conservatism and economic crises
No previous research has been conducted on the impact of the 2007-2008 Financial Crisis on earnings conservatism. Though, Francis et al. (2013) used the recent financial crisis to study the benefits of conservative accounting to shareholders. Furthermore, research has been carried out on the influence of previous crises on conservatism in financial reporting. These influences will be discussed in this section. However, this section starts with an elaboration of the research of Jenkins et al. (2009), who studied the variance in earnings conservatism and value relevance across the business cycle.
Jenkins et al. (2009) used the business cycle instead of a specific crisis to study conservatism. They expect and find that management and auditors report more conservative earnings during economic recessions due to the increased focus on downside risk. The findings provided by the authors are in line with the conservatism explanations provided by Watts (2003a). First of all, there is a litigation risk. They posit that this risk is likely to be higher during periods of recession, because equity markets often experience sharp drops in stock prices. According to Watts (2003a), litigation risk can be reduced by reporting conservative numbers. Furthermore, Jenkins et al. (2009) state that during a contraction, investors demand more conservative accounting, because they are uncertain about the future. Negative outcomes are more likely during an economic recession. Therefore, investors require timely signals of the potential existence of negative net present value projects. The last reason provided by Jenkins et al. (2009) is the preference of internal sources of funding over external sources, and with external sources, the preference of debt over equity financing. The authors argue that internal funding is likely to be more difficult in a recession, because the profitability of the firm declines. Firms will therefore move to external funding. However, this creates information asymmetry between the manager and debt holder.
Recent studies (e.g. LaFond and Watts, 2008; Khan and Watts, 2009) show that conservatism is useful in reducing information asymmetry. Therefore, during economic recessions, more conservatism is required.
14 The results of Jenkins et al. (2009) are clear and convincing. Nevertheless, they are contradictory with the studies by Herrmann et al. (2008), Vichitsarawong et al. (2010) and Gul et al. (2002) described below, which use the 1997 Asian financial crisis as the center of their study. Herrmann et al. (2008) investigate conservatism in Thai firms during and subsequent to the 1997 Asian financial crisis. Specifically, they studied whether there was a difference between firms audited by Big 4 auditors or non-Big 4 auditors. The results show that, in general, Big 4 audit clients report more conservatively. Furthermore, their research states that Big 4 audit clients were more conservative in the crisis period than in the post-crisis period. According to the authors, this is due to the fact that Big 4 auditors demand more conservatism from their clients to reduce their legal risk and maintain audit quality. In the period after the crisis, the authors find no significant difference between conservatism applied by clients of Big 4 auditors or non-Big 4 auditors. Improvements in the accounting profession, accounting regulation, and corporate governance mechanisms are claimed to be potential reasons for this. Likewise, the impact of the 1997 Asian financial crisis has also been investigated by Vichitsarawong et al. (2010). Basically, the authors extended research of Herrmann et al. (2008) by not focusing solely on Thailand, but on Hong Kong, Malaysia and Singapore as well. However, their results are similar, since the authors find that firms reported less conservative during the financial crisis. Furthermore, the authors studied the timeliness of earnings, which is the time it takes to incorporate publicly available news in the financial statements (Basu, 1997). The results show that earnings are less timely during the crisis period compared to the period before and after the crisis.
Gul et al. (2002) examined the relation between audit fees and conservatism in Hong Kong during the 1997 Asian financial crisis. Just like Herrmann et al. (2008) and Vichitsarawong et al. (2010), Gul et al. (2002) find that conservatism decreased during the crisis. The authors argue that this is due to the fact that the crisis puts pressure on the manager to convey more positive news to investors. Furthermore, they find that less conservatism increases the audit fees charged to the firm, because of an increase in audit effort to counter aggressive reporting, since this increases the probability that the firm violates accounting standards.
The papers discussed above examined the effect of a financial crisis, i.a. the 1997 Asian financial crisis, on the extent to which manager use conservatism in the financial statements. The research of Francis et al. (2013) is different, since the authors did not study the impact of the crisis on conservatism, but the benefits of conservatism for shareholders in a crisis setting. The authors find strong evidence of a positive association between
15 conservatism and the value of a firm during the 2007-2008 Financial Crisis. The results of this study are important for this research, since it implies that in a crisis-setting, shareholders appreciate firms that apply conservatism to a certain degree. Moreover, the authors find that this relation mainly depends on the corporate governance environment of the firm. For firms with weaker corporate governance or higher information asymmetry, the relation between conservatism and firm value is more pronounced. Furthermore, the authors find that conservative firms have several advantages compared to less conservative firms including a lower default risk and cost of debt while issuing more debt. Besides, they reduce cash holdings, capital expenditures, and R&D expenditures less. Their increase of discretional accruals is lower as well.
2.6 Accounting conservatism and differences between countries
Empirical evidence of Basu (1997) shows that financial statements of listed firms in the U.S. are conservative. However, a lot of research has been conducted in other countries as well, often by making a comparison between multiple countries. This section highlights the most related studies in this domain.
A study that is very important for this research is the study of Ball et al. (2000). Their research is directly linked to this study, since they examined the effect of international institutional factors on properties of accounting earnings. Specifically, the authors make a distinction between common-law and code-law countries. They define code-law countries as countries with a high political influence (e.g. Japan) and common-law countries as countries where the private sector primarily determines accounting practices (e.g. United States). The authors hypothesize that the demand for timely and conservative accounting is weakened by the politicization of accounting standard setting and enforcement. This implies that there is a higher demand for an income variable with low volatility. The authors argue that there is a strong political influence on accounting practices in code-law countries. Accounting standards and enforcement is established and enforced by governments. The government represents several other stakeholders such as labor unions, banks and business associations that have contracts with the firm. According to the authors this leads to a ‘stakeholder’ governance model. Ball et al. (2000) describe current-period accounting income as a pie which has to be divided between not only shareholders, but also taxes to be paid to the tax authority and bonuses to managers and/or employees, i.e. to several other stakeholders. Holthausen and Watts (2001) complement this, by stating that conservatism reduces the
16 likelihood that resources are distributed unevenly to parties less important than the lenders. Ball et al. (2000) argue that information asymmetry is solved by insider communication because of the representation of these stakeholders in corporate governance. Conversely, common-law countries are characterized by a ‘shareholder’ governance model. According to the authors, this implies that the governing board is elected solely by the shareholders, not by other stakeholders. Furthermore, the payouts have a less close relation to accounting income of the current-period. The information asymmetry which exists between the board and the shareholders is largely resolved by public disclosures. Shareholders largely determine the desirable properties of accounting income. The authors hypothesize that this includes the timeliness of incorporating negative economic income. At the same time, the authors acknowledge that countries always have elements of both models and thus are never homogeneous. Their findings are in line with all their expectations. Accounting income in common-law countries is indeed timelier compared to code-law countries. This is due to the asymmetric verification of losses in accounting income compared to gains, since losses are recognized faster. This implies a higher level of conservatism in common-law countries.
Ball et al. (2000) only focus on earnings conservatism. Another cross-country study is the research by García Lara and Mora (2004) who examined both earnings conservatism and balance sheet conservatism across eight European countries. Hereby, they replicate and extend the research of Ball et al. (2000). The authors expect balance sheet conservatism to be less pronounced in common-law countries compared to code-law countries. Consistent with their expectations, the authors find that both earnings conservatism and balance sheet conservatism is present in all countries. Furthermore, the paper provides evidence that code-law countries are more balance sheet conservative. This result clearly indicates that there are two types of conservatism, since Ball et al. (2000) designated earnings conservatism as being more pronounced in common-law countries. The study of García Lara and Mora (2004) provides a possible explanation for this since their results suggest that earnings conservatism is reduced by balance sheet conservatism. According to the authors, this could be due to differences in the legal and institutional environment, an explanation which is widely supported (e.g. Ball et al., 2000; Watts, 2003a; Bushman and Piotroski, 2006). In addition, this is also consistent with the research of Beaver and Ryan (2005), who found that conditional conservatism is reduced by unconditional conservatism.
The aforementioned studies made a clear distinction between common-law and code-law countries. However, Ball et al. (2003) argue that it is misleading to classify countries as common-law or code-law, thereby ignoring the incentives of managers and auditors. These
17 incentives are i.a. influenced by institutional environments and enforcement mechanisms. Furthermore, the research of Ball et al. (2000) and García Lara and Mora (2004) shows that it is not always straightforward whether a country should be classified as either common-law or code-law. For example, Ball et al. (2000) designate the Netherlands as a common-law country whereas García Lara and Mora (2004) deem code-law as being more appropriate.
Ball et al. (2003) focus in their study on Hong Kong, Malaysia, Singapore and Thailand, which use standards derived from common-law sources. Those standards are widely viewed as higher quality than code-law standards. However, the authors argue that the incentives of the preparers imply low quality. For their research, Ball et al. (2003) use the timely recognition of economic income to measure reporting quality. They define economic income as the common stock return of a firm over the fiscal year. The results are in line with their expectations, since the authors find that the financial reporting quality of these East Asian countries is not higher than under code-law despite their alleged high quality standards. According to the authors, their results have three implications. First of all, determining the financial reporting quality of a country based on its accounting standards gives an incomplete and potentially misleading view. Second, the authors argue that changing the incentives of managers and auditors and other institutional factors is more important to increase financial reporting quality than adopting higher quality accounting standards. Lastly, comparing accounting information internationally cannot be achieved solely through homogenization of accounting standards.
Another study which includes several countries is the study of Bushman and Piotroski (2006). The authors focus in their study on the influence of legal and political institutions on conservative accounting, thereby extending the research of Ball et al. (2000) and Ball et al. (2003). The authors find that the quality of the judicial system has an influence on the timeliness of reflecting bad news. Firms in countries with a well-developed judicial system with strong investor protections reflect bad news in reported earnings more timely compared to firms in countries with an underdeveloped judicial system with weak investor protections. This leads to conservatism being more present in countries with a well-developed judicial quality and more diffuse ownership structures or high usage of public bonds. Furthermore, the authors claim that public enforcement actions of securities law have an impact on conservatism. When public enforcement is strong, recognition of good news is slowed down.
18
2.7 Earnings management
As stated in the introduction of this chapter, reported earnings are to a certain extent influenced by earnings management. This paragraph provides a better understanding of earnings management and tries to show the relationship between conservatism and earnings management.
2.7.1 Definitions of earnings management
Just like accounting conservatism, prior literature focused extensively on earnings management, leading to different definitions describing the same phenomenon. Many of these definitions have a negative view about earnings management (e.g. Healy and Wahlen, 1999; Roychowdhury, 2006). Healy and Wahlen (1999, p. 6) use the following definition:
Roychowdhury (2006, p. 337) provides a similar definition:
Above definitions give the impression that earnings management is about misleading. Several practices can be carried out which generally lead to reported income being less variable or the covering of deteriorating firm performance (García Lara et al., 2005). However, earnings management is not always about misleading. Ronen and Yaari (2008) argue that definitions of earnings management can be divided into white, grey and black definitions. Above definitions are examples of black earnings management. The authors consider this as the pernicious form of earnings management. When an accounting treatment is used that is either opportunistic or economically efficient, the authors consider it gray earnings management. Finally, the authors denote earnings management as white when it is used to take advantage of the flexibility in the choice of accounting treatment to signal the managers’ private
“Earnings management occurs when managers use judgment in financial reporting and in structuring transactions to alter financial reports to either
mislead some stakeholders about the underlying economic performance of the
company, or to influence contractual outcomes that depend on reported accounting numbers.” [emphasis added]
“Departures from normal operational practices, motivated by managers’ desire to mislead at least some stakeholders into believing certain financial reporting goals have been met in the normal course of operations.” [emphasis added]
19 information on future cash flows. This “white” definition describes the form of earnings management that is socially acceptable.
The frequently made dichotomy is the distinction between real earnings management and accrual based earnings management. The following two subsections will provide more information about these two types. Thereafter, the relationship between earnings management and conservatism will be described in greater detail.
2.7.2 Real earnings management
One of the papers describing real earnings management in great detail is the paper of Roychowdhury (2006). The author defines real earnings management as the departure from normal business practices by managers to meet certain financial reporting goals. Hereby, they mislead at least some stakeholders by letting them believe that those goals have been reached in the normal course of operations. The author agrees with the fact that some real activities manipulation methods, in certain economic circumstances, can be optimal actions. It becomes real earnings management when managers engage in these activities more than needed to meet or beat an earnings target.
2.7.3 Accrual based earnings management
Accruals are defined as the difference between reported earnings and cash flows from operations and can be decomposed into discretionary accruals and non-discretionary accruals (Healy, 1985). Healy (1985) argues that non-discretionary accruals (NDAt) originate when
accounting standard-setting bodies mandate accounting adjustments to the firm’s cash flows. When adjustments are selected by the manager, they are called discretionary accruals (DAt).
Healy argues that a manager can manage earnings by modifying the timing of reported earnings. Discretionary accruals enable managers to transfer earnings between periods
The focus of this thesis is on discretionary accruals. There are several possibilities for managers to transfer earnings between periods through discretionary accruals. Levitt (1998) describes three ways in which managers use accruals to understate assets. “Big Bath” charges can occur when companies restructure to remain competitive, efficient and profitable. According to Levitt, companies have incentives to overstate these charges, because the stock market only focusses on future earnings. By estimating these charges conservatively, they can be used as income when the estimates change or when future earnings prove insufficient. The second way described by Levitt is creative acquisition accounting. Firms activate an increasing portion of the acquisition prices as “in-process” Research and Development. By
20 doing this, this amount can be written off at once, which removes any future earnings drag. “Cookie Jar Reserves” are the final way in which accruals are used to decrease income. Firms make unrealistic assumptions to estimate liabilities for various items such as loan losses, sales returns and warranty costs. In good times, firms can store these accruals and reach to them in bad times.
The above results indicate that accruals reverse over time. This conclusion is supported by other empirical evidence (Ahmed et al., 2002; Cohen et al., 2008). Ahmed et al. (2002) argue that, in the long run, income before depreciation and amortization converges to operating cash flows. Because both positive and negative accruals reverse over time, net cumulative accruals should approach zero. However, the authors claim that conservative accounting leads to a persistent pattern of negative accruals over time. They correctly point out that, over a reasonably long period, a firm’s mean accrual provides an accounting-based, firm-specific proxy for conservatism.
2.7.4 Relationship between earnings management and earnings conservatism
Using the Jones (1991) model and the Basu (1997) model, García Lara et al. (2005) study whether earnings management affects the measures of earnings conservatism. The basis for their study is the previously cited study by Ball et al. (2000) about asymmetry in the recognition of good and bad news in earnings in common-law and code-law countries. Other studies (e.g. García Lara and Mora, 2004) fail to find significant differences between common-law and code-law based countries in Europe. The authors use France, Germany and the UK as their field of study to find out why other studies do not come to the same results. In a first step, they analyze the existence of earnings conservatism. The authors do not find significant differences in the level of conservatism between the three countries. In the second step, they control for the existence of earnings management. Using a measure of unmanaged earnings instead of observed earnings, the authors find that the differential recognition speed of bad news decreases significantly in France and Germany. When the authors adjust for discretionary accruals, the differences between France and Germany, referred to as code-law countries, and the UK, referred to as a common-law country, become significant. With their study, the authors show that in certain institutional contexts, the measures of earnings conservatism are significantly driven by earnings management.
By citing the IASB framework, García Lara et al. (2005) argue that the deliberate understatement of assets or income, or the deliberate overstatement of liabilities or expenses
21 is not allowed, since this contradicts the neutrality of the financial statements. Therefore, this should not be classified as conservatism. Contrary, the authors argue that these practices should be considered as earnings management practices, since the objective is not to protect investors, but to respond to the incentives of the manager. In this line of reasoning, conservatism should be reflected in the non-discretionary component of accruals, whereas earnings management is reflected in the discretionary component of accruals.
2.8 Hypotheses development
Several insights have been extracted from the literature review and accounting theories. In this paragraph, the research hypotheses arising from this information will be developed. Based on the literature review, there are several reasons to believe that the 2007-2008 Financial Crisis has had an impact on conservatism. Particularly, Watts (2003a) mentioned litigation as an important explanation for conservatism. During an economic recession, huge declines in stock prices are more likely (Jenkins, 2009). Watts argues that firms generally use conservatism to address the heightened securities litigation risk accompanied with huge declines in stock prices. In the same manner, auditors require that companies report more conservative in high legal liability regimes (i.a. Basu, 1997; Huijgen and Lubberink, 2005). Besides the litigation risk, Jenkins (2009) discusses the threat of increased regulatory scrutiny. This scrutiny is the result of the public being less certain about the future of the firm, especially the possibly existence and impact of hidden bad news. Regulatory scrutiny increases the risk of possible earnings restatements and the corresponding losses on investments. Firms can mitigate this risk by reporting more conservative. Third, as stated earlier, Jenkins (2009) argues that firms prefer internal sources of funding over external sources and, in case of external resources, debt over equity financing. However, during economic contractions, the possibilities for internal financing are limited, since profitability often declines, resulting in less free cash flows. Therefore, more firms will seek opportunities for external funding. As argued by Beatty et al. (2008), debt holders are primarily interested in getting their money back. Therefore, they demand more conservative accounting. Moreover, as argued by Zhang (2008), debt holders reduce the cost of debt for conservatively reporting firms, because conservatism reduces the downside risk. Besides of this, Francis et al. (2004) argue that conservatism leads to a lower cost of equity. Therefore, based on the aforementioned literature, a shift from internal to external funding which likely happens during a recession should increase the demand for conservative reporting.
22 The three arguments presented above all come down to the presence of information asymmetry between the firm and another party, including shareholders, debt holders and government. As argued by Johnson et al. (2000), this is especially the case during a financial crisis, since the expected return on investment is often lower as usual. Managers want to create a positive perception of their company, among others to justify their policies and rewards, and therefore have incentives to manipulate accounting numbers. Furthermore, the limited tenure and liability of managers provides incentives to recognize earnings that are currently unverifiable (LaFond and Roychowdhury, 2008). Simultaneously, the literature review demonstrated how conservatism can help to mitigate this opportunistic behavior of managers (i.a. Watts, 2003; LaFond and Watts, 2008; Francis et al., 2013). As argued by Watts (2003a), accounting information becomes more reliable due to the asymmetric verification requirements which are used in conservatism. Francis et al. (2013) consider conservatism as an efficient disciplining mechanism. By decreasing agency and information risks, conservatism helps to mitigate firm value losses during crisis periods.
Following these arguments, I expect that the 2007-2008 Financial Crisis had a positive effect on the use of earnings conservatism in the financial statements. Formally stated, this leads to the following hypothesis:
H1: The 2007-2008 Financial Crisis has increased the amount of earnings
conservatism applied by firms.
The development of the second hypothesis is aimed at the influence of institutional factors on conservatism. Ball et al. (2000) argue that the extent of political influence on both standard setting and enforcement is likely to be the most fundamental institutional factor causing accounting income to differ internationally. The authors claim that common-law countries resolve information asymmetry by public disclosure whereas code-law countries do this through private communication. Consequently, they find that common-law accounting income is timelier than code-law accounting income. This is because common-law countries show a greater sensitivity to economic losses. The authors conjecture that the early incorporation of losses increases the incentives of managers to attend to the sources of losses more quickly. When a company reports conservative, security analysts are more inclined to put pressure on managers. Furthermore, dividend restrictions and leverage are made binding more quickly, and bonuses of managers and employees are lower. Therefore, conservatism is used as a tool to monitor managers.
23 Ball et al. (2000) caution that it is impossible to make a precise distinction between common-law and code-common-law countries, since all countries have elements of both classes. The authors touch upon the Securities and Exchange Acts in the U.S. which codified its predominantly common-law reporting system and the allowance of consolidated financial statements prepared under common-law accounting standards in French and Germany. Comparative studies in a European context (e.g. García Lara and Mora, 2004), are not able to find a significant difference between common-law and code-law countries. García Lara et al. (2005) investigate why comparative studies do not achieve the same results as Ball et al. (2000). The authors find that, in European countries, accruals are to a large extent determined by earnings management. The authors argue that conservatism is reflected in the non-discretionary component of accruals, whereas earnings management is reflected in discretionary accruals. After controlling for discretionary accruals, García Lara et al. (2005) find a significant difference between conservatism applied in common-law and code-law countries.
One of the differences between common-law and code-law countries, as argued by Ball et al. (2000), is that the contracting explanation of conservatism is more applicable to common-law countries. In these countries, it is more common that the board of directors is elected by the shareholders only. As opposed to code-law countries, managers in common-law countries are less likely to have a substantial interest in the firm. Furthermore, debt and equity markets do monitor managers more and employees or lenders are less likely to be represented on the board. Therefore, all the stakeholders contracting with the firm operate at a larger distance, which leads to more information asymmetry. Therefore, in common-law countries, there is a larger demand for timely public disclosure.
Furthermore, in line with other studies (e.g. Kellogg, 1984; Lys and Watts, 1994; Watts, 2003a, 2003b), Ball et al. (2000) argue that securities litigation is another main explanation for the existence of conservatism. This is because shareholders are more likely to sue managers and auditors when assets and earnings are overstated. In addition, when firms fail to disclose material bad news on a regular basis, this can lead to litigation. Reporting in a more conservative way can lower this litigation risk.
Following these arguments, I expect that the 2007-2008 Financial Crisis incurs higher risks for common-law countries. This leads to the second hypothesis:
H2: The change in earnings conservatism as a result of the 2007-2008 Financial
Crisis is more significant for common-law countries compared to code-law countries.
24
3. Research design
3.1 Sample selection
The sample is collected from the COMPUSTAT Global and COMPUSTAT North America database. These databases contain a large amount of historical financial information of publicly traded companies worldwide. The countries included in this study are the same as the countries used in the research of Ball et al. (2000), being Australia, Canada, UK and USA as common-law countries and France, Germany and Japan as code-law countries.
The time frame consists of a pre-crisis period and a crisis period. The years 2004 till 2007 are classified as the pre-crisis period and the years 2008 till 2011 are classified as the crisis period. This results in an initial sample of 164,962 firm/year observations.
To calculate the discretionary accruals metrics which are employed in this analysis, it is required that each firm-year observation has all the data available. Therefore, each firm/year with a missing value for any variable is excluded. This results in the exclusion of 122,424 observations. A possible consequence of this restriction is a survivorship bias, whereby larger and more successful firms dominate. Furthermore, regulated utilities (SIC 4800-4999) and financial firms (SIC 6000-6999) are excluded because of their different nature of accruals (Pae, 2007). This leads to the exclusion of 14,702 observations. Therefore, the final sample exists of 27,836 firm-year observations as shown in Table 1. To reduce the influence of outliers on the results, data is winsorized by 5% from both tails. The industry distribution of the sample, based on the Standard Industrial Classification (SIC) is reported in table 2.
Table 1
Sample Sampling procedure
Total sample 2003-2011; 2003 lost due to lagged variables 164,962 (-) Exclusion of firms with missing values for any variable 122,424 (-) Exclusion of regulated utilities and financial firms 14,702
