HUB HUB RESEARCH PAPER Earnings management: a literature review S. Verbruggen, J. Christaens, and K. Milis HUB RESEARCH PAPER 2008/14. FEBRUARI 2008 Hogeschool – Universiteit Brussel Stormstraat 2, 1000 Brussel, Belgium T: +32 2 210 12 11 F: + 32 2 217 64 64 Earnings management: a literature review S. Verbruggen, Hogeschool Universiteit Brussel Prof dr. J. Christaens, Universiteit Gent Prof. dr. K. Milis, Hogeschool Universiteit Brussel

Abstract Based on a literature review of major accounting journals, this paper attempts to offer a comprehensive overview of recent earnings management research and provide a critical classification of articles on the matter as well as a search for voids in current literature. A selection of leading journals was reviewed systematically from January 2000 onwards resulting in 145 articles examining ‘earnings management’. Each article was thoroughly screened in terms of research question, methodology and findings.

The paper attempts to be in keeping with prior review articles from around the turn of the century (Healy and Wahlen 1999; Fields, Lys and Vincent 2001) and focuses on the latest evolutions given that earnings management remained a focal research topic. The study resulted in four research categories: motives for earnings management, earnings management techniques, restrictions to earnings management, and research design issues. In every category, main research conclusions as well as methodological issues are discussed.

Screening and classifying earnings management literature did not only generate a structured overview of the work performed in this area, it also provided insights in some important voids, such as a focus on non-listed and small companies, ‘real’ earnings management and non-financial motives. Finally, this paper offers a systematic literature review and evidences that there is ample room for further research. 1 1. Introduction Researchers (re)directing their focus to earnings management are confronted with an extent body of literature regarding the subject.

Prior review articles such as those by Schipper (1989), Healy and Wahlen (1999) and Dechow and Skinner (2000) on earnings management and by Fields, Lys and Vincent (2001) on accounting choice have created structure in the enormous number of articles dedicated to the subject. McNichols (2000) has focused on design issues while reviewing recent literature. These prior review articles focused mainly on research done in the 1990’s. We extend this by examining the literature on earnings management in the early years of the new decade: January 2000September 2006.

We selected 11 major accounting journals and screened them systematically on earnings management in title, abstract and/or author supplied keywords. As such, we read 153 articles focusing on or relating to earnings management. The articles are quite evenly spread over the entire period. This indicates that there is an continuous interest in this field of research. The articles deal with a variety of issues related to earnings management. We narrow this broad research area down into different categories to provide structure in the existing literature.

This allows us to assess whether there is a shift in the research focus and to detect voids in current research. The remainder of the paper is organized as follows. In the next two sections we briefly discuss three prior review articles and our methodology (section 3). We discuss the first category (motives for earnings management) in section 4, followed by techniques of earnings management in section 5. How earnings management can be restricted is outlined in section 6. We discuss several research design issues in section 7.

A summary concludes our paper. 2. Prior review articles Three important review articles from around the turn of the century serve as a basis for our research. Healy and Wahlen (1999) have reviewed earnings management literature in respect to the usefulness of prior research for standard setters. Fields, Lys and Vincent (2001), have structured their analysis around three types of market imperfections. The third review paper (McNichols, 2000) discusses the trade-offs associated with three research designs commonly used in earnings management literature. 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’ (Healy and Wahlen, 1999, p. 365). Healy and Wahlen’s often cited definition sets the tone for several papers on earnings management. While it indicates that there are two motives and two techniques for earnings management, it leaves ample room to refine these goals and modi operandi.

Healy and Wahlen came across three different motives for earnings management: capital market expectations and valuations, contracts written in terms of accounting numbers and antitrust or other government regulation. They concluded that research has not been able to assist standard setters in their attempts to restrain earnings management nor to provide evidence on the extent and scope of earnings management practices. Even though Fields et al. 2001) review accounting choice research articles, their classification is also useful for earnings management studies: ‘Although not all accounting choices involve earnings management, and the term earnings management extends beyond accounting choice, the implications of accounting choice to achieve a goal are consistent with the idea of earnings management. ’ They organized the accounting choice literature into three groups based on as many market imperfections: agency costs, information asymmetries and externalities affecting non-contracting parties.

Once again, the motives for earnings management were made apparent. Managers want to influence the outcome of contracts (e. g. compensation agreements and debt covenants), stock prices and policies of third parties (e. g. taxes, industry specific regulations). They argued that progress in the field of accounting choice 2 has slowed. They defined three fields for further research: measuring the implications of alternative accounting methods, building analytical models that provide guidance to empiricists, designing more powerful statistical techniques and improving research designs?

This last issue is the main subject of the review paper by McNichols (2000). She discussed the characteristics of the three most commonly applied designs in the earnings management literature: aggregate accruals models, specific accruals models and the frequency distribution approach. One of the main arguments against using aggregate accruals models is that we do not have enough knowledge on how these accruals ‘behave’ in the absence of earnings management. That’s one of the reasons why McNichols argued that progress in earnings management research would come from specific accruals research.

The frequency distributions (of different earnings metrics) approach introduced by Burgstahler and Dichev (1997) is another often used method to distinguish between companies who are thought to be managing their earnings and those companies who are probably not. This method, although quite easy to put into practice, is also being criticized. This will be addressed in section 7. 3. Methodology and contribution We selected eight major accounting journals, based on research conducted by Ballas and Theoharakis (2003). They have ranked accounting and finance journals, according to their main focus.

Based on the top ten journals of their rankings, we selected the following: Accounting Horizons (AHO), Accounting, Organizations and Society (AOS), Accounting Review (TAR), Contemporary Accounting Research (CAR), Journal of Accounting, Auditing and Finance (JAAF), Journal of Accounting and Economics (JAE), Journal of Accounting and Public Policy (JAP) and Journal of Accounting Research (JAR). To make sure we conduct a worldwide screening, we also added Abacus (ABA) and European Accounting Review (EAR) to the list.

We selected articles that are dedicated to or linked to earnings management by systematically screening the chosen journals on ‘earnings management’ in either title, abstract and/or author supplied keywords. This selection process resulted in a list of 145 articles. Nine out of ten journals dedicate at least 1 article to earnings management. There are no articles selected in Accounting, Organizations and Society during the entire research period. We have added Review of Accounting Studies (rather new journal but with growing importance) to our list and selected 8 papers following the same selection procedure.

Thus, our selection extended to 153 articles. ABA AHO CAR EAR JAAF JAE JAP JAR RAS TAR Total 0 0 2 1 1 2 2 4 0 2 2000 14 0 1 1 2 0 2 6 1 2 4 2001 19 1 0 1 0 0 1 3 0 5 11 2002 22 3 3 5 5 1 4 6 1 3 2 2003 33 0 1 0 5 3 1 1 2 1 6 2004 20 1 0 0 3 1 1 3 0 3 5 2005 17 3 0 0 2 0 1 6 3 2 11 2006 28 8 Total 5 9 18 6 12 27 11 16 41 153 5% 27% 100% % 3% 6% 12% 4% 8% 18% 7% 10% Table 1. Number of selected earnings management articles per journal per year % 9% 12% 14% 22% 13% 11% 18% 100% Three journals (TAR, JAE and CAR) represent over 50% of the selected articles.

The ‘top four’ (inclusing JAR) of the list represents 2/3 of the articles read Since we want to build our review on earlier work and extend it, we focus on four major questions: (i) which earnings management motives have been investigated recently, (ii) which earnings management techniques have been dealt with, (iii) have possible restrictions (such as corporate governance mechanisms, specific accounting standards or audit issues been identified and (iv) has there been any progress in fortifying research designs? 3

These questions directed us to four categories to which we assign all 153 papers, based on their major focus: motives for earnings management, earnings management techniques, limiting earnings management and research design issues. Tables 2 and 3 represent the number of selected articles per journal and per year. The categories consist of 22 up to 51 papers and are spread over the entire research period. There have been no real ‘trends’ during this period, all categories are represented over the years. We cannot draw any conclusions about a ‘specialisation’ of journals, because the number of papers per journal is often quite limited.

According to the current search the number of articles dedicated to earnings management exceeds 10% of all published articles in the same period in Accounting Review, Journal of Accounting and Economics, Contemporary Accounting Research and Journal of Accounting, Auditing and Finance (based on unreported results). ABA AHO CAR EAR JAAF JAE Motives 2 1 6 2 10 12 40% 11% 33% 33% 83% 44% % Techniques 3 3 6 0 0 6 60% 33% 33% 0% 0% 22% % Restrictions 0 5 3 3 1 5 0% 56% 17% 50% 8% 19% % Research design 0 0 3 1 1 4 0% 0% 17% 17% 8% 15% % Total 5 9 18 6 12 27 Table 2.

Categories of earnings management articles per journal JAP JAR 4 3 36% 19% 0 4 0% 25% 5 6 45% 38% 2 18% 11 RAS 1 13% 1 13% 1 13% TAR Total 10 51 24% 33% 9 32 22% 21% 19 48 46% 31% 3 7% 41 22 14% 153 5 3 19% 63% 8 16 Category/Year 2000 2001 2002 2003 2004 Motives 6 5 6 16 4 43% 26% 27% 48% 20% % Techniques 3 4 3 7 6 21% 21% 14% 21% 30% % Restrictions 3 6 10 5 8 21% 32% 45% 15% 40% % Research design 2 4 3 5 2 14% 21% 14% 15% 10% % Total 14 19 22 33 20 Table 3. Categories of earnings management articles per year 2005 8 47% 2 12% 5 29% 2 12% 17 006 6 21% 7 25% 11 39% 4 14% 28 Total 51 33% 32 21% 48 31% 22 14% 153 Table 4 provides an overview of the methodology of the papers. We divide all papers into empirical – based on financial and/or accounting data, empirical – survey, experiments, analytical models, discussion papers and others. The vast majority of papers (72%) depend on financial and/or accounting data to find empirical evidence of earnings management. All other methods represent each less than 10% of all papers. discuss experim fin/acc model survey other total Motives 4 1 37 8% 2% 73% % Techniques 1 0 28 3% 0% 88% % Restrictions 4 5 30 8% 10% 63% % Research design 3 0 15 14% 0% 68% % Total 12 6 110 8% 4% 72% % Table 4: methodology used in the selection of papers 4. Motives for earnings management 7 14% 0 0% 3 6% 4 18% 14 9% 1 2% 1 3% 1 2% 0 0% 3 2% 1 2% 2 6% 5 11% 0 0% 8 5% 51 32 48 22 153 Engaging in earnings management is not a risk-free operation. Companies and CEO’s risk to damage their reputations and there are litigation risks involved with managing earnings.

Therefore, companies will only engage in earnings management when the benefits of this behaviour are higher then the risks and costs involved. We identify 5 broad categories of incentives: (i) stock market incentives, (ii) signalling / concealing private information, (iii) political costs, (iv) making the CEO look good and (v) internal motives. 4. 1 Stock market incentives Although there are several possible motives for managing earnings, the spotlight has been on those incentives that are related to the stock market.

The interaction between accounting numbers and stock markets reaction can indeed push management towards earnings management. Although the focus on listed firms seems logical and natural, we have to bear in mind that the majority of earnings management studies published in the journals that were under consideration, rely on US data. The US economy is known for its widespread ownership and liquid and efficient stock markets. (Cormier et al. , 2000) In several other countries, there are far less listed companies and privately owned companies set the tone.

We have to consider the fact that in those countries, there might be other important reasons for earnings management (such as tax avoidance) that haven’t been under the attention of researchers quite as much. Investing in stocks can be a risky venture. That is why investors often rely on the views and forecasts of stock market analysts to put together a portfolio of potentially successful firms. Meeting or beating the analysts’ forecasts seems to be of enough importance for companies to engage in earnings management.

Several research papers are dedicated to finding out why managers try to meet or beat expectations as well as to finding evidence consistent with earnings management to reach this benchmark. Meeting the analysts’ expectations is important because firms that meet or beat expectations enjoy higher returns, even when it is likely that this is achieved through earnings management or expectations management (Bartov et al. , 2002). Missing an earnings benchmark has negative implications for stock returns as well as CEO compensation. Matsunaga and Park, 2001). To be able to meet or beat the forecasts, managers turn to earnings management. Payne and Robb (2000) concluded that the more analysts agree, the stronger the incentive is to meet the consensus forecast. If pre-managed earnings are below the forecast, managers use income-increasing earnings management. If pre-managed earnings are higher then the forecast, managers can choose between income-decreasing earnings management (saving it for a rainy day) or not managing the earnings (hoping for an increase in stock return). The fact that there is evidence of a relationship between pre-managed, forecasted and reported earnings, makes it important for analysts, researchers as well as regulators to be able to detect those companies that are likely to engage/have engaged in earnings management to meet the targets. But is it also possible to do so? Burgstahler and Eames (2003) found evidence that firms engage in earnings management to avoid small losses and earnings decreases, but analysts seem to be unable to correctly identify those firms.

In order to help to identify those firms that might engage in earnings management to avoid negative earnings surprises, Matsumoto (2002) has tried to identify firm characteristics that are associated with this kind of behaviour. She finds that firms with higher transient institutional ownership are more likely to meet or beat expectations. So are firms who rely heavily on implicit claims with their stakeholders and who are in industries in which earnings have a higher value-relevance. These firms seem to use earnings management as well as expectations management to be able to meet the expectations.

Ghosh et al. (2005) conclude that companies that show an increase in earnings as well as in revenues are less susceptible to earnings management. To align shareholders’ goals with managers’ objectives and give less room to agency conflicts, CEO’s and senior management are often compensated by equity incentives. The previously mentioned papers introduce evidence of earnings management to meet or beat expectations and to influence the stock price. This kind of opportunistic behaviour might even increase when there is a direct link to these two incentives and the financial benefit of the firm’s management.

Earnings management’s link with insider trading is documented by Beneish and Vargus (2002), Park and Park (2004) and Cheng and Warfield (2005). Other studies document the relationship between earnings management and stock compensation through stock options (Baker et al (2003), Bartov and Mohanram (2004), Kwon and Yin (2006)). Recent research also considered earnings management in specific stock market situations, such as an initial public offering (DuCharme et al. ,2001) and seasoned equity offerings (Shivakumar, 2000). The opposite of an equity offering, a share repurchase, can also give way to earnings management.

Vafeas et al (2003) find weak evidence that managers decrease earnings through accruals prior to a share repurchase. Bens et al (2003) document that corporate executives use stock repurchases as an earnings management tool when earnings are below the level required to achieve the desired growth of earnings per share. This conclusion is questioned by Larcker (2003) because (among other arguments) managers are supposed to be extremely myopic to engage in such decision-making. Several researchers have also used analytical models to determine stock market incentives for earnings management.

Ronen et al. (2003) and Feltham and Pae (2000) use the earnings response coefficient (ERC) in their models in combination with earnings management. The use of ERC to proxy for earnings quality is questioned by Fischer and Stocken (2004). Ronen et al (2006) have studied the effect of directors’ equity incentives on the likelihood of earnings management and the stock price as well as the firm’s value. They show that ‘earnings management distorts the stock price because the market cannot undo the bias in the accounting report.

Furthermore, it reduces the firm’s value because of its unfavourable effect on the manager’s effort. ’ ‘If aspects of a firm’s financial well-being depend on how it compares to similar firms, then firms may wish to manage their earnings in order to reap comparison benefits’ is the conclusion of Bagnoli and Watts (2000). They modelled earnings management in a game theoretic approach and show that earnings management emerges relatively easy and firms engage in earnings management simply because they expects their rivals to do so. 4. Signalling or concealing private information 6 Earnings management is, by definition, a process of altering financial information in order to achieve certain goals. i To make a category dedicated to the motive of ‘signalling or concealing private information’ might seem strange. Some articles, however, are dedicated to that motive, without obvious links to other second-stage goals such as those discussed above. Rosner (2003) has examined whether failing firms engage in earnings management and alter their annual accounts to conceal their financial struggle.

Without immediately measuring the consequences on stock price or CEO compensation, Rosner has shown the existence of earnings management to simply conceal information. Louis and Robinson (2005) suggest that the accrual signal, combined with another signal such as a stock split might be an effective way of communicating private information. Rather than assuming that managers manage earnings for their personal benefit (opportunism), they have looked into the possibility that it is the only (or optimal) way to communicate their optimism.

Sometimes it is difficult for firms to communicate the goals of their accounting practices to market participants. Shane and Stock (2006) find evidence of the fact that analysts fail to recognize shifts in earnings as optimal tax planning. When the market doesn’t see through this form of earnings management, these firms might be penalised for their strategic tax planning. Tucker and Zarowin (2006) also shed light on the discussion whether managers manage their earnings to their own benefit (garbling) or to improve informativeness. We document empirically that an important effect of managers’ use of financial reporting discretion is to reveal more information about firms’ future earnings and cash flows. ’ 4. 3 Political costs Next to changing financial statements in order to influence shareholders’ opinions and decisions, firms can also manage reported earnings in response to other stakeholders that make use of financial reports. Governmental regulations and tax laws, when they make use of financial reports, are obvious candidates to be analyzed as possible sources of earnings management motives.

It can be valuable to companies to seem more/less profitable to escape from governmental interference. Haw et al (2005) studied income-increasing earnings management in China as a response to governmental regulations demanding a minimum of 10% return on equity for firms wanting to issue bonds or offer shares. Johnston and Rock (2005) examined income-drecreasing earnings management for firms threatened by the Superfund Act. When accounting numbers are the basis for tax calculation, there might be large tax avoidance incentives for earnings management (Monem, 2003).

But not only governments can be misinformed by managed earnings. D’Souza, Jacob and Ramesh (2001) provide evidence for firms using earnings management to reduce labour renegotiation costs. More unionized companies are more likely to choose the immediate recognition adoption method of SFAS 106 (postretirement benefits), thereby influencing reported income and labour negotiation strength. In summary, political costs, although less under researchers’ attention in recent years, seem to be a strong incentive for firms to manage their earnings.

This is even proven in economies where there are no liquid en efficient stock markets and CEO’s are appointed by the government. 4. 3 Making the CEO look good There might be other than financial motives for the CEO to manage earnings. Two articles present evidence of earnings management when there is a change in CEO (Godfrey et al. , 2003) or when the CEO is retiring (Reitenga and Tearny, 2003). A new CEO can be inclined to downwards earnings 7 management in the year of change and upwards earnings management in the following years. Retiring CEO’s use upwards earnings management to leave in style and keep a seat on the board. . 4 Internal motives Finally, we sum up motives for earnings management that are not linked to external stakeholders (such as shareholders, government or unions) but are intra-company. Within a company, it might also be useful to alter financial reports or to structure transactions in such a way that budget ratcheting is avoided or performance standards are met. Leone and Rock (2002) investigated the accruals of several business units within the North American division of a large multinational company to study the effect of budget ratcheting on earnings management.

Their evidence is consistent with ratcheting, meaning that ‘budgets change in response to the prior year’s variance from the budget and the absolute change is greater for positive variances’. They test this in an environment of both transitory and permanent earnings innovations. The hypothesis that under the ratchet effect, managers will choose to use income-decreasing unexpected accruals when the earnings innovations are transitory is supported by their empirical evidence. Murphy (2001) examined the relationship between the nature of performance standards in incentive contracts and earnings smoothing.

He concludes that companies using externally determined standards (i. e. relatively unaffected by participants such as peer group standards, fixed standards or cost of capital) are less likely to smooth earnings than those companies that use internal standards (budget goals, prior year, subjective standards). 4. 5 Do practitioners agree? The results of a survey (Graham et al. , 2005) indicates that managers believe that earnings are important to external constituents and that meeting or beating analysts’ expectations and prior period earnings is important to build credibility with the capital markets and uphold stock prices.

They do not agree with a compensation motivation hypothesis. Beating important benchmarks and issuing equity are also motives brought forward in the review by Dechow and Skinner (2000). 5. How do companies manage earnings? While the majority of papersii uses aggregate unexpected accruals (using the Jones (1991) model, the or a similar procedure to estimate expected accruals, compare those with actual accruals and use the difference as a proxy for earnings management) a number of papers specifies an alternative route to detect managed earnings.

To provide clarity in this category of articles, we chose the following subcategories: earnings management through specific accruals, through cost allocation or cost shifting, through disclosure and through ‘real action’. The first three ‘methods’ all refer to altering financial data, the last method refers to the structuring of transactions in order to reduce/enlarge reported earnings. 5. 1 Earnings management through specific accruals The use of specific accruals to manage earnings is often linked to a specific situation, a specific industry or a specific accounting standard.

The accruals that have been studied might be defined as ‘the usual suspects’: researchers direct their attention to a specific accrual that is likely to be managed because there is some room for accounting choice and the accrual is important enough to manage earnings above a certain level. 8 Marquardt and Wiedman (2004) studies specific accruals used in specific situations (equity offerings, management buyouts and earnings decrease). Among the specific accounts investigated are also the tax expense (Dhaliwal et al. 2004) and restructuring charges (Moehrle, 2002). Among the specific industries studied are banks (Capalbo(2003) and Gray and Clarke (2004)), insurance (Beaver et al. , 2003) and investment property companies (Dietrich et al. , 2001). In each of these industries a specific accrual was managed, respectively allowance for loan losses, pension accounting, loss reserve accrual and valuation of property. When an accounting standard leaves a manager with interpretation or application choices, earnings management can also occur.

This was investigated for SFAS 89 (Picconi, 2006) and SFAS 109 (Schrand and Wong, 2003). 5. 2 Earnings management through cost allocation or income shifting Companies can try to manage earnings by allocating joint costs to those activities that are appreciated by the public, by shifting costs to below-the-line items or by shifting costs/revenue to/from other subsidiaries that are located in an area with another tax or accounting regime. Even in charitable organizations, reasons for and methods of earnings management can be found.

Jones and Roberts (2006) show that charities use the allocation of joint costs to smooth the program ratio, an often used indicator of charity efficiency. Most income statements separate core expenses and revenue from financial expenses and revenue and special items. Since the attention of users of financial reports goes out to the core financial data, it might be useful for companies to shift expenses from ‘core expenses’ to ‘special items’. Jaggi and Baydoun (2001) have examined this in Hong Kong (prior to SSAP2) and McVay (2006) found evidence of such practices in the US.

Intra-group earnings shifts are documented by Beatty and Harris (2001) and Krull (2004). They found evidence that income shifting from one subsidiary to another is used to optimize taxes and reported earnings, using gain realizations and permanently reinvested earnings. 5. 3 Earnings management through disclosure (other than balance sheet and income statement) Balsam et al. (2003) find evidence consistent with companies managing the pro forma stock option expense, under SFAS No 123. This standard encouraged firms to expense the estimated fair value of employee stock options.

Almost no firms did so prior to 2002, but instead chose to report the pro forma impact of stock option grants in a footnote. Even when the expense is disclosed instead of recognized, firms seem to manage this expense when their CEO’s compensation and value of stock option grants are relatively high in order to reduce public criticism about these compensation practices. Schrand and Walther (2000) document that ‘managers strategically select the prior-period earnings amount that is used as a benchmark to evaluate current-period earnings in quarterly earnings announcements’.

Although this can be defined as perception management, rather than earnings management, we still want to mention the results of this research in our earnings management review. Both managing activities are clearly related to each other and often will serve the same purpose. 5. 4 Earnings management through ‘real activities’ Instead of using accounting discretion to alter financial reports, companies can also turn to real activities to manage earnings upwards or downwards. They can also structure their transaction in a way that a certain accounting standard does/does not apply resulting in a more opportune income statement or balance sheet.

Among the ‘real earnings management’ investigated are : selling price cuts, just-in-time adoption, R&D budget cuts, etc. 9 ‘Real activities manipulation is defined as management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain thresholds’ is the definition by Roychowdhury (2006) in a recent article on earnings management. He found evidence that companies give price discounts in order to increase sales, engage in overproduction in order to reduce the cost of goods sold and keep a tight rein on discretionary spending to improve margins.

Kinney and Wempe (2004) documented that the decision to adopt just-in-time practices (JIT)- a fundamental operational decision- is influenced by the relationship of firms’ LIFO reserves with their income smoothing, debt covenant and tax incentives. Mande et al (2000) documented a substantial decrease in R&D-expenses in Japanese firms during the 1990’s recession. These firms, who had a reputation of long-term R&D vision, showed signals of myopic income-increasing behaviour. The evidence suggests that these cutbacks in expenses are a product of earnings management instead of optimal business decisions.

Hribar et al (2006) investigated stock repurchases as a tool to increase earnings per share. While most earnings management studies focus on the numerator (earnings), these researchers have investigated whether managers increase EPS by decreasing the denominator (number of shares). They found an unexpectedly large number of stock repurchases among firms that would have missed the EPSforecast without the repurchase. Another way of managing earnings through real activities, rather than accounting decisions is the timing of ‘payins’ to corporate foundations.

Companies donate money to a foundation, which, in turn, makes grants to a charity. This timing gap between payins and payouts creates earnings management possibilities. Petrovits (2006) documented that firms with high stock price sensitivity and small increases in earnings make the most income-increasing foundation funding choices. Firms with increasing earnings despite of large income-decreasing foundation funding choices in the current year are more likely to increase earnings in subsequent periods, consistent with the use of cookie jar reserves and earnings smoothing.

Barton (2001) showed that derivatives and discretionary accruals are used as earnings management tools. He states that managers can smooth earnings by smoothing cash flows (through hedging) and by increasing/decreasing accruals. His empirical results show that ‘firms holding derivative portfolios with large notional amounts also have lower absolute levels of discretionary accruals, suggesting that derivatives and discretionary accruals are partial substitutes for smoothing earnings’.

Following Nelson et al (2003) who find that firms will use transaction structuring when accounting rules are very precise and through judgement when rules can be interpreted, Marquardt and Wiedman (2005) showed that firms use contingent convertible bonds (COCO’s) to increase diluted EPS. Firms use these COCO’s because, under SFAS No 128, these bonds do not have to be part of the denominator of diluted EPS. 5. 5 Do auditors and managers agree? We can compare the research conclusions to the experience of 515 auditors, surveyed by Nelson et al (2003).

They documented that the following earnings management attempts are frequently observed: recognizing reserves (such as loan losses in banks), recognizing asset impairment, capitalizing/deferring too much or too little, reducing previous accrual (such as deferred tax asset valuation allowance), modifying depreciation, cut-off manipulation, deferring revenue, bill-and-hold sales, sale-and-lease-back transactions, misestimating percentage-of-completion, income statement classification, avoiding consolidation etc.

The list of earnings management attempts seems endless and is too long to discuss in this review paper. We can only conclude that the techniques discussed by researchers and lived by auditors are very alike. The list of Nelson et al. could be a source of inspiration for researchers trying to detect earnings management by specific accruals. 10 There are more discrepancies between the direction of research (most often discretionary total accruals and specific accruals) and the opinion of managers, surveyed by Graham et al (2005).

Managers seem to turn more to ‘real activities earnings management’ than to earnings management by accruals management. 6. Restraining earnings management The third category consists of articles that focus on or are related to the restriction of earnings management. Engaging in earnings management can lead to different consequences. As discussed above, companies can try to increase their stock price or escape from political scrutiny. Managers can increase bonus pay or the value of their stock options or simply enhance their reputation.

These advantages have to outsize the possible negative consequences, such as litigation costs, decrease in stock prices and loss of reputation, in order for companies to engage in earnings management. The negative outcomes of revealed earnings management are, in themselves, a strong restriction to this type of behaviour. ‘If earnings management is considered unethical by financial statement users, then managers’ and companies’ reputations may suffer and companies’ credibility in the financial markets may be damaged’ (Kaplan,2001).

Kaplan investigated whether shareholders and non-shareholders of a company perceive earnings management as more or less unethical depending on the intent and technique of earnings management. His experiment shows that we cannot simply assume that nonshareholders always perceive earnings management as unethical and shareholders’ opinion depends on the earnings management’s intent (individual managerial benefit versus company benefit). When we assume that it is necessary to prevent the occurrence of earnings management and managers use accounting techniques to manage reported earnings, we have to rely on well-defined accounting standards.

However, there will always be some degree of choice in accounting practices in a complex and global business environment. The next step is then to turn to the auditors of a company to make sure that financial reports are a reliable and unedited picture of the company’s financial status. For listed companies other watchdogs, such as the Security and Exchange Commission (SEC) in the US or CBFA in Belgium, also control the financial reporting process. The Enron debacle and other evidence of accounting malpractices provoked yet another way of restricting earnings management.

Corporate governance initiatives surged and audit committees as well as outside directors on the board may present new limits to earnings management. 6. 1 Accounting standards, SEC and state control to constrain earnings management One way to reduce earnings management (by accounting techniques) is setting more rigorous accounting standards. However, this may have the unwanted effect that managers turn to ‘real earnings management’, which consists of abnormal, suboptimal, business practices in order to change reported earnings.

Tan and Jamal’s experiment (2006) leads to the conclusion that although high foresight managers ‘can continue to report smooth earnings in order to communicate with shareholders, the use of this form of earnings management can reduce their firm’s productive capacity and growth prospects’. This implies that, when earnings management is used to communicate with shareholders but hampered by strict accounting standards, managers may turn to decisions that smooth earnings in the short run but are damaging to the company in the long run.

This idea can also be found in the message of Arya et al (2003): ‘Accounting research shows that income manipulation is not an unmitigated evil; within limits, it promotes efficient decisions. (…) Earnings management and managerial discretion are intricately linked to serve multiple functions. Accounting reform that ignores these interconnections 11 could do more harm than good. (…) The implicit role of regulators is to make earnings management challenging, not impossible. ’ Several analytical models also question the idea of total eradication of earnings management (Liang, 2004; Ewert and Wagenhofer, 2005; Dutta and Gigler, 2002).

Although these authors challenge the idea that total prohibition of earnings management is necessary at all times, most studies look for ways to impede this kind of behaviour. When standards are written with earnings management avoidance in mind, they might turn out very precise, leaving managers with almost no accounting choice. Is this the way to avoid earnings management and/or increase the relevance and informativeness of reported earnings? The research results do not seem to agree.

Ramesh and Revsine (2001) document that companies (in this case: banks), when given a choice in adoption method and timing of two new accounting standards (SFAS 106 and 109), consider the effect on reported earnings and balance sheet. Nelson et al. (2002) report their results of a questionnaire in which auditors describe attempts of earnings management by their clients. They find support for their hypothesis that ‘managers are more likely to attempt earnings management with structured transactions when standards are precise and with unstructured transactions when standards are imprecise’.

This also indicates that the precision of an accounting standard in itself is not enough to rule out earnings management. Lang et al. (2006) present evidence that cross-listed non-US firms report smoother earnings, show a greater tendency to manage towards an earnings target and are less timely in the recognition of losses than matched listed US firms. This evidence of earnings management is stronger for companies coming from countries with low investor protection.

Their evidence suggests that accounting standards indeed influence the level of earnings management but that SEC regulation and reconciliation with US GAAP do not entirely eliminate local GAAP managing effects. Baber et al. (2006) argue that investors are more able to interpret earnings (and correct them for earnings management) when additional information on balance sheet data and cash flows is disclosed. Healy et al. (2002) use a simulation model that examines the trade-off between objectivity (in accounting for R&D investments) and relevance of the accounting information.

A simple capitalisation rule (in which R&D investments are capitalized and written down when unsuccessful or amortized over the expected life when successful) which leaves more room for earnings management, stays more informative than expensing all R&D investments, even when earnings management is widespread. This tradeoff can also be found in the paper by Altamuro et al (2005). They find evidence that the introduction of Staff Accounting Bulletin (SAB) n° 101 decreased earnings management, but also caused a decline in earnings informativeness.

One of the most obvious trends in accounting is the harmonisation of standards. Several authors have investigated its effect on earnings management. Chen et al. (2002) tried to explain the earnings gap between local GAAP and IAS, even after harmonisation efforts were made. One of the explanations of the continuing gap is that, due to inadequate supporting infrastructure and low quality auditing, earnings management was manifestly present. The lesson to be learned here is that a high quality level of standards in itself does not suffice to rule out earnings management.

This is also the conclusion of Van Tendeloo and Vanstraelen (2005) who questioned whether the voluntary adoption of International Financial Reporting Standards (IFRS) is associated with lower earnings management in code-law countries with low investor protection rights. They found no evidence to support this hypothesis. Next to accounting standards, state regulations and state oversight can also influence the level of (potential) earnings management. Dowdell and Press (2004) reported that SEC scrutiny seems to have altered financial reporting practices in the ‘right’ direction.

Gaver and Paterson (2000) showed that the level of under-reserving by weak insurers declined after states adopted certain standards aimed to improve the quality of insurers’ financial statements. But not all state rules have the same rate of 12 success. Chen and Yuan (2004) document that although Chinese regulators appeared to scrutinize firms’ earnings management attempts to reach the required 10% return on equity, many firms were still able to gain state approval to issue additional shares by using excess amounts of non-operating income. . 2 Audit as a constraint to earnings management If accounting standards as well as governmental scrutiny do not completely eradicate earnings management and earnings management is as pervasive as many believe, then auditors should be confronted with attempts to alter financial reports. Nelson et al. (2002) have questioned several auditors within a big 5 firm on their experience with earnings management attempts. Several authors have investigated the relation between audit(or) characteristics and earnings management by their client.

Increased audit quality could or should lead to increased quality of reported earnings. Audit quality can be proxied in different ways. Krishnan (2003) and Van Caneghem (2004) used auditors’ industry expertise as an indicator for audit quality. Both found a negative correlation between the auditors’ specialisation and the emergence of earnings management (measured by discretionary accruals and earnings rounding up behaviour).

Another proxy for audit quality that is often used is auditor size. Van Caneghem using UK evidence and Vander Bauwhede and Willekens (2004) using Belgian evidence found no evidence that auditor size (either as dichotomous Big x – NonBig x, or as continuous measure of size) decreases earnings management practices. Kim et al (2003) indicated that Big 6 auditors seem to be more effective in deterring earnings management only when managers prefer income-increasing accrual choices. Contrary to conventional wisdom, we find Big 6 auditors are less effective than non-big 6 auditors when both managers and auditors have incentives to prefer incomedecreasing accrual choices and thus no conflict of reporting incentives exists between the two parties. ’ One of the most important characteristics of an auditor is his independence. This independence can be compromised when the auditor also supplies non-audit services to the client, when the client is very important for the auditor, when the audit partner is linked with a client for too long or when a client hires a CFO directly from its auditing firm.

All of these ‘threats’ were investigated in recent years in relation to earnings management. Frankel et al. (2002) and Ferguson et al. (2004) documented a positive relationship between the purchase of non-audit services and earnings management proxies (earnings surprises, discretionary accruals, public criticism in financial reports, restatements). Chung and Kallapur (2003) did not find a statistically significant association between client importance and discretionary accruals. Auditor independence can also be reduced when an audit partner is associated with a client for too long.

Carey and Simnett (2006) investigated the association between audit partner tenure and audit quality. They found evidence that long tenure is positively associated with a lower propensity to issue a going-concern opinion and some evidence of just beating earnings benchmarks. Finally, Geiger et al (2005) investigated the effect of the so-called auditor-to-client revolving door on audit quality. This concept refers to the situation in which a company hires a senior financial reporting manager directly from its external auditing firm.

In these situations, earnings management could occur immediately before or after the hiring when auditor independence is impaired towards the future employer (before) or when the new CFO uses audit-firm-specific knowledge (after). However, the results of this study do not support such a hypothesis, since discretionary accruals are no greater in these situations compared to control groups. Overall, the relation between potential weakened auditor independence and a decline in audit quality (measured by earnings management proxies) hasn’t been proven.

Heninger (2001) showed that abnormal accruals are positively associated with auditor litigation risk. This indicates that shareholders and other external stakeholders hold auditors responsible when management inflates earnings to make the company look healthier. For auditors, this is an important motivation to inspect the income-increasing accruals of their clients, especially when they are financially distressed or large companies. 13 Firms can also abuse materiality to manage earnings. Libby and Kinney (2000) report on the result of an experiment with audit managers.

They find that auditors expect a majority of clients to make full correction only if the forecast will not be missed. One of their conclusions is: ‘our results imply opportunistic correction of quantitatively immaterial misstatements to manage earnings to forecasts, and auditor acceptance of the practice. ’ The experiment by Nelson et al. (2005) indicates that the approach to materiality (cumulative versus current-period materiality approach) influences the decision of the auditor whether or not to require adjustments to the financial statements.

Their suggestion to standard setters is to expect auditors to require adjustment whenever a misstatement is material under either approach. 6. 3. Company characteristics that restrict earnings management Two recent evolutions in business communication and business ethics might decrease the occurrence of earnings management. Wasley and Wu (2006) documented that management not only forecasts earnings, but also cash flow to meet the demand of investors. In doing so, they ‘precommit to a certain composition of earnings, thus reducing the degree of freedom in arnings management. ’ Klein (2002) examined whether audit committee characteristics and board characteristics are associated with earnings management. She found that there is a negative relation between the independence of both and the presence of abnormal accruals. Kim and Yi (2006) found evidence on the association of ownership structure, group affiliation and listing status with earnings management. They documented that when ownership becomes more disperse, firms tend to engage more in earnings management, measured by the magnitude of unsigned discretionary accruals.

This is also the case when the company is part of a business group and is publicly traded. 7. Methodological issues and analytical models 7. 1 Most frequently used models and methods to detect earnings management The most important ‘flaw’ of earnings management research stems from the fact that earnings management and managerial intent aren’t directly observable. Researchers use different methods to try to detect earnings management. Many empirical studies use the hypothesis of accruals manipulation to detect earnings management.

Since accounting manipulation is said to be less costly than real transactions earnings management (because it doesn’t affect cash flow and it doesn’t impede growth of a company), researchers have focused on accounting methods (cf. category 2). Within the different accounting manipulation methods available, accruals seem to be the favorite instrument. Accruals management is less obvious and detectable than, for example, changing accounting methods that have to be explained in the financial statements of the company.

The accruals that are thought to be manipulated are most often referred to as discretionary accruals (or also unexpected or abnormal accruals). The Jones (1991) model and modified Jones model, although not free of criticism, are by far the most popular models used. They are based on earlier work by Healy (1985) and DeAngelo (1986) who used (the change in) total accruals from the estimation period to proxy for expected nondiscretionary accruals in the event period. This implies however that, if non-discretionary accruals change from period to period, their models will measure non-discretionary accruals with error.

Jones has relaxed the assumption of constant non-discretionary accruals and has created a model that controls for the effect of the company’s economic evolution on nondiscretionary accruals. The original Jones model was modified by Dechow et al. (1995). The reason for the modification was the fact that some discretionary accruals are measured as non-discretionary accruals when part of revenue is also managed (e. g. managers could accrue revenue that is in fact not yet earned and not yet received in cash, which would lead to a change in revenue, but also to a change in receivables). 14

Next to these aggregate accruals models, there are also models using specific accruals. These have been discussed above and are often related to a specific industry or a specific situation. iii Another method often used in earnings management research (in combination with discretionary accruals) is the distribution of reported earnings around certain earnings benchmark (zero earnings, last quarter’s earnings) introduced by Burgstahler and Dichev (1997) and Degeorge et al. (1999). Assuming a smooth probability distribution, the expected number of observations in an interval is the average of the two adjacent intervals.

The difference between the expected number and the actual number of observations, divided by the estimated standard deviation of the difference, indicates whether there is a significant discontinuity in a certain interval. Discontinuities around the previously mentioned earnings benchmarks are considered to be signaling earnings management. 7. 2 Discussion of discretionary accruals models Although the Jones and modified Jones model are popular models to detect earnings management, researchers are aware that these models may be biased or even misspecified.

We discuss the following issues in order to improve the detection of earnings management: (i) time-series versus cross-sectional design, (ii) balance sheet versus cash flow statement to calculate accruals, (iii) the use of performance matching and (iv) linear versus non-linear models. McNichols (2000) examined some research design issues related to the use of the aforementioned models and methods. One of the main arguments against using aggregate accruals models is that there is little knowledge of how accruals behave in the absence of earnings management.

McNichols argues that the amount of papers in which there is evidence of earnings management (22 out of 23 she has taken up in her list of references) may indicate aggressive earnings management or bias in the earnings management tests. It may also reflect bias in the editorial process where papers finding no evidence of earnings management are less likely to be published. The bias in the earnings management tests can have several causes. For one, these models often measure discretionary accruals by industry.

If companies manage their earnings simply because they expect their competitors to do so, the level of discretionary accruals will be understated because the average level of discretion exercised by the industry is included. Furthermore, estimation of discretionary accruals requires specification of a prediction period (estimation period) and a test period (event period). The assumption is that there is no earnings management during the prediction period. Given the aforementioned numerous motives for earnings management, this assumption might be difficult to be maintained.

Although there are disadvantages to measure discretionary accruals by industry, Bartov et al. (2001) concluded that the cross-sectional Jones and modified Jones model outperform their time-series counterparts. Accruals can be measured in two ways, depending either on the balance sheet or on the cash flow statement of a company. ‘Despite the availability of accurate accruals data in the statement of cash flow since 1988, the majority of these studies use an indirect balance sheet approach to calculate accruals’ is one of the observations of Hribar and Collins (2002).

They argued that using the balance sheet approach to predict accruals leads to estimation errors when non-operating events occur, such as acquisitions and accounting changes. They found evidence that accruals estimates are biased when the partitioning variable is correlated with either mergers/acquisitions or discontinued operations. Kothari et al. (2005) examined the power of ‘traditional’ discretionary accruals tests and tests based on performance-matched discretionary accruals.

Their results suggest that performance matching ‘enhances the reliability of inferences from earnings management research when the hypothesis being tested does not imply that earnings management will vary with performance, or where the control firms are not expected to engage in earnings management’ Ball and Shivakumar (2006) argued that using linear models to predict accruals leads to substantial bias because these models do not take into account the loss recognition asymmetry. Therefore, these 15 models are a poor specification of the accounting accruals process.

They showed that nonlinear models that incorporate timelier loss recognition (conservatism) explain more variation in accruals then linear specifications. Philipps (2003) et al. suggested using the deferred tax expense to detect earnings management, next to discretionary accruals. They provided evidence that this expense is incrementally useful beyond total accruals and abnormal accruals derived from two Jones-type models in detecting earnings management to avoid a small loss or an earnings decline. 7. Discussion of the distribution approach Ayers et al. (2006) investigated whether the association between discretionary accruals and beating earnings benchmarks hold for comparisons of groups segregated at other points in the distribution of earnings, changes in earnings and earnings surprises. Their results suggest that the positive association between discretionary accruals and beating the earnings benchmark extends to other points (they call pseudo targets) in the distribution of both earnings and earnings changes.

This means that the mere positive association between discretionary accruals and meeting or beating earnings targets is not sufficient to conclude that discretionary accruals detect earnings management. Dechow et al. (2003) conclude that, since both small profit firms and small loss firms show similar positive discretionary accruals, it is unlikely that they are the only explanation for the kink in the earnings’ distribution. They argue that researchers have to consider other explanations, such as real action to beat the benchmark and the effect of the denominator.

Durtschi and Easton (2005) conclude that the shape of the distribution in se is not enough to evidence earnings management. They provided evidence that the shape of frequency distributions around zero earnings is affected by deflation, by sample selection criteria and/or by differences between the characteristics of observations to the right and to the left of zero. They found that, opposed to the findings of Burgstahler and Dichev (1997) for the distribution of deflated earnings, there are no discontinuities around zero in the distribution of net income, earnings per share and diluted earnings per share.

This might be caused by the deflator, if it is different for firms to the left and the right of the earnings benchmark. Additionally, they found that the proportion of small loss firms being deleted because the beginning-of-the year-price isn’t available on Compustat is larger than the proportion of small profit firms. Also, the proportion of small loss firms followed by I/B/E/S is much smaller than the proportion of small profit firms. This can also lead to a discontinuity in the earnings distribution which is not related to earnings management. 7. Conclusion Although there is little progress in determining new models to measure earnings management, the model created by Jones seems to be the subject of improvement efforts. Recent research has come up with a number of suggestions to enhance the model’s use, such as using cross-sectional data instead of time-series, matching on performance, using cash flow statements instead of balance sheet data to calculate accruals. The suggestion made by Ball and Shivakumar to use nonlinear models to predict accruals instead of the linear models that have been used so far might shed a complete new light on the subject.

Building such a model and applying it to issues that have been investigated before is a possible new direction for future research. 8. Concluding remarks We have read and categorized 153 articles focusing on or related to earnings management in the period from January 2000 to September 2006. We have created four categories of papers, depending on their main subject: incentives for earnings management, earnings management techniques, restrictions to earnings management and research design issues. Using these categories, we have identified voids in current earnings management literature. 6 The focus on listed firms, often in the US, has left room to investigate non-listed, non-US firms more closely. This is also the case for non-US regulations. Recent articles have shed some light on non-for-profit firms trying to manage earnings. We believe that there are research opportunities in that area. Researchers have proven that firms manage their earnings in order to avoid taxes or political scrutiny. We haven’t found research regarding earnings management to obtain subsidies or grants from the government, which might be the case for non-profit firms.

We have found little research concerning corporate governance initiatives and the introduction of IAS/IFRS in European listed companies. It would be helpful to find out whether these new initiatives limit the existence of earnings management. On the methodological front there has been little progress. However, recent articles such as those by Ball and Shivakumar (2006) and Kothari et al. (2005) might bring new dimensions into earnings management models that still have to be explored. There are limitations to our research stemming from the selection of journals as well as the time span under review.

Also, some articles are difficult to assign to one of the categories, because they deal with motives and/or techniques and/or restrictions and/or methodological issues. To our opinion, this hasn’t threatened neither the logical build up of the categories, or the overall overview and determined voids. i Earnings management by structuring transactions is discussed in section 2 62% of papers using fiancial data in the first category, 52% of papers using financial data in the second category iii Cfr. Section 5. 1 ii 17 References Abarbanell, J. & Lehavy, R. (2003).

Can Stock Recommendations Predict Earning Management a