2. Earnings Management: Origins

As well as in-depth interviews of CFOs.

Dechow et al. (2010) provide a comprehensive review of earnings quality studies.

Davidson et al. (1987) has been cited by Schipper (1989).

“Unlike economic transactions with an unrelated counterparty, in related party transactions, the same individual is on both sides of the transaction” (Gordon et al. 2007, p. 96).

IAS No. 24 defines a related party as a person or entity that is related to the entity that is preparing its financial statements (in this Standard it is referred to as the ‘reporting entity’). (a) A person or a close member of that person’s family is related to a reporting entity if that person: (i) has control or joint control over the reporting entity; (ii) has significant influence over the reporting entity; or (iii) is a member of the key management personnel of the reporting entity or of a parent of the reporting entity. (b) An entity is related to a reporting entity if any of the following conditions apply: (i) the entity and the reporting entity are members of the same group (which means that each parent, subsidiary, and fellow subsidiary is related to the others); (ii) one entity is an associate or joint venture of the other entity (or an associate or joint venture of a member of a group of which the other entity is a member); (iii) both entities are joint ventures of the same third party; (iv) one entity is a joint venture of a third entity and the other entity is an associate of the third entity; (v) the entity is a post-employment benefit plan for the benefit of employees of either the reporting entity or an entity related to the reporting entity. If the reporting entity is itself such a plan, the sponsoring employers are also related to the reporting entity; (vi) the entity is controlled or jointly controlled by a person identified in (a); (vii) a person identified in [a(i)] has significant influence over the entity or is a member of the key management personnel of the entity (or of a parent of the entity); (viii) the entity, or any member of a group of which it is part, provides key management personnel services to the reporting entity or to the parent of the reporting entity. The Statement of Financial Accounting Standards No. 57 (FAS No. 57) defines related party transactions as transactions between (a) a parent company and its subsidiaries; (b) subsidiaries of a common parent; (c) an enterprise and trusts for the benefit of employees, such as pension and profit-sharing trusts that are managed by or under the trusteeship of the enterprise’s management; (d) an enterprise and its principal owners, management, or members of their immediate families; and (e) affiliates.

Account schemes through which management commits fraud by manipulating financial statements are (among others): overvalued assets and understated expenses; omitted or understated expenses/liabilities; fictitious assets; other methods to overstate revenues; overvalued assets/equity; and misclassification (Gao and Srivastava 2007).

For example, under US GAAPs, the Statement of Financial Accounting Standards No. 95 (FAS No. 95) issued in 1987 became effective for the annual financial statements of fiscal years ending after July 15, 1988. While in 1992, the International Accounting Standards Board issued International Accounting Standard No. 7 (IAS No. 7), which became effective only in 1994, mandating that firms provide cash flow statements.

Hribar and Collins (2002) took both components of accruals directly from the statement of cash flows. Specifically, they took the following data items from the Compustat database: Compustat #123 to determine earnings before extraordinary items; and to determine cash flows from operations (CFO), they subtracted from net cash flow (Compustat #308) the amount of extraordinary items and discontinued operations (Compustat #124).

In a subsequent article, Healy (1996) changed the terminology and stated: “I regret that I bear much of the responsibility for the current labels, which I first used in my bonus plan paper (Healy 1985). If I were to rewrite that paper today, I would certainly change the terminology. What I termed ‘discretionary’ accruals would be renamed ‘unexpected’ accruals and what I called ‘nondiscretionary’ earnings would be relabeled as ‘expected’ earnings” (p. 114). The perspective has changed; the main point is not to detect earnings management but to forecast accruals. However, following conventional practice (Peasnell et al. 2000), I use the terms “managed accruals,” “discretionary accruals,” “unexpected,” and “abnormal accruals” interchangeably. Similarly, the terms “unmanaged accruals,” “non-discretionary accruals,” “expected,” and “normal accruals” are used interchangeably.

Healy (1985) specified: “These bodies require, for example, that companies depreciate long-lived assets in some systematic manner, value inventories using the lower of cost or market rule, and value obligations on financing leases at the present value of the lease payments” (p. 89).

As Healy (1985) pointed out, “the manager chooses discretionary accruals from an opportunity set of generally accepted procedures defined by accounting standard-setting bodies. For example, the manager can choose the method of depreciating long-lived assets; he can accelerate or delay delivery of inventory at the end of the fiscal year; and he can allocate fixed factory overheads between cost of goods sold and inventories” (p. 89).

Healy and Wahlen (1999) explained: “many studies begin with total accruals, measured as the difference between reported net income and cash flows from operations. Total accruals are then regressed on variables that are proxies for normal accruals, such as revenues (or cash collections from customers) to allow for typical working capital needs (such as receivables, inventory, and trade credit), and gross fixed assets to allow for normal depreciation. Unexpected accruals are thus the unexplained (i.e., the residual) components of total accruals” (p. 370).

Big bath accounting is a managerial stratagem (Walsh et al. 1991) based on the assumptions that “when circumstances are bad, making things just a little bit worse by cleaning out the rubbish does little harm to either reputation or prospects” and that “little damage will ensue when the market is so depressed that nothing can hurt it more” (Walsh et al. 1991, p. 174).

Under the income-smoothing hypothesis, “earnings are manipulated to reduce fluctuations around some level that is considered normal for the firm” (Bartov 1993, p. 840). Income smoothing is an earnings management technique and is defined as follows: “Income smoothing is the process of manipulating the time profile of earnings or earnings reports to make the reported income stream less variable, […]. To smooth income, a manager takes actions that increase reported income when income is low and takes actions that decrease reported income when income is relatively high; this latter aspect is what differentiates income smoothing from the related process of trying to exaggerate earnings in all states” (Fudenberg and Tirole 1995, pp. 75–76).

Perry and Williams (1994) argued that the principal reason for such contrasting results is sample size. Compared to the DeAngelo (1986) study, Perry and Williams (1994) examined a much larger sample of firms going private (175 management buyout proposals).

“Hurricanes Katrina and Rita caused widespread disruption to the US-based oil industry and were followed by large price increases for both crude oil and gasoline. These large price increases triggered a widespread public outcry that companies in the oil industry were engaged in price gouging. Various proposals were floated for investigations, regulations, and a windfall profits tax specifically aimed at companies in the oil industry. If passed, these proposals could have imposed large additional costs on these companies, thus adversely affecting their future profitability” (Byard et al. 2007, p. 734).

Issuing companies manage earnings upward through income-increasing accounting adjustments in order to increase the offering proceeds. However, the high earnings reported around SEOs temporarily overvalue issuing firms until the subsequent fiscal period in which discretionary accruals reverse.

The difference between the current earnings and analysts’ forecast earnings is called “earnings surprise” (Kinney et al. 2002, p. 1299).

Bartov et al. (2002) defined habitual beaters as firms that have met or beaten expectations in at least 9 (75%) of the previous 12 (100%) quarterly earnings forecasts.

Burgstahler and Eames (2006) “view earnings management as encompassing both actions that increase current earnings without decreasing future earnings and actions that increase current earnings at the expense of future earnings” (p. 635). The authors called the former type of earnings management “business management” and the latter “reporting management.” To proxy for business management (reporting management) the authors used changes in operating cash flows (discretionary accruals).

Overall, Cheng and Warfield’s (2005) results “suggest that CEOs with high equity incentives take more income increasing abnormal accruals than those with low equity incentives” (p. 467).

Chung et al. (2002) specified: “When managers have incentives to increase reported profits, institutional investors put pressure on them to limit the use of income-increasing DAC (discretionary accruals). Similarly, when managers have incentives to decrease reported profits, institutions apply pressure on them to limit the use of income-decreasing discretionary accounting accruals” (p. 46).

Koh (2003) used the level of institutional ownership to proxy institutional ownership types, where a lower (higher) ownership region approximates short-term-oriented (long-term-oriented) institutional ownership. Furthermore, he examined the association between levels of institutional ownership and income-increasing discretionary accruals. He found a positive (negative) association between levels of institutional ownership and aggressive accruals management in a lower (higher) institutional ownership region. More generally, Velury and Jenkins (2006) found a positive association between institutional ownership and several attributes of earnings quality. By examining the impact of institutional ownership on overall earnings quality, the authors provided evidence on whether the quality of earnings improves as investment by institutions increases. However, this positive association between institutional ownership and earnings quality is negatively affected by increased ownership concentration (Velury and Jenkins 2006).

Rosner (2003) used a sample of 51 s sanctioned and 242 non-sanctioned bankrupt firms. Within the non-sanctioned bankrupt firms, she created subsamples of stressed bankrupt firms (SB) and non-stressed bankrupt firms (NSB). Analyzing financial statements in a five-year window, she found evidence of earnings overstatement in the NSB which “resemble the SEC-sanctioned fraud firms” (Rosner 2003, p. 401).

Specifically, within the sample of firms chosen, the mean ratio of receivables to total assets was 28.7%, compared to 22.3% for the Compustat population, while the mean ratio of allowance for uncollectible accounts receivable to net income before extraordinary item was 29.4%, compared to 20.3% for the Compustat population.

Teoh et al. (1998b) grouped the depreciation methods into three categories: accelerated, straight-line, and a combination of straight-line and accelerated. Straight-line is viewed as the most income-increasing method, followed by a combination of straight-line and accelerated, while the accelerated method suggests income-decreasing accounting policies. The income-increasing group therefore includes any IPO that uses a more income-increasing method than the matched firm does.

The authors argued: “Tax expense provides a final opportunity to meet earnings targets after the firm has agreed to any re-tax adjusting entries required by the independent auditors. Tax expense also contains the complexity and discretion necessary for information asymmetry to persist. Thus, tax expense is a powerful setting in which to examine earnings management among a wide range of firms” (Dhaliwal et al. 2004, pp. 451–452).

Deviations from the normal levels of these proxies are termed abnormal. Roychowdhury (2006) followed Dechow et al. (1998) to determine normal levels of cash flow from operations, production costs, and discretionary expenses.

Cohen and Zarowin (2010) argued that “such discounts and lenient credit terms will temporarily increase sales volumes, but these are likely to disappear once the firm reverts to old prices. The additional sales will boost current period earnings, assuming the margins are positive. However, both price discounts and more lenient credit terms will result in lower cash flows in the current period” (p. 8).

Cohen and Zarowin (2010) argued that “managers can increase production more than necessary in order to increase earnings. When managers produce more units, they can spread the fixed overhead costs over a larger number of units, thus lowering fixed costs per unit. As long as the reduction in fixed costs per unit is not offset by any increase in marginal cost per unit, total cost per unit declines. This decreases reported cost of goods sold (COGS) and the firm can report higher operating margins. However, the firm will still incur other production and holding costs that will lead to higher annual production costs relative to sales, and lower cash flows from operations given sales levels” (p. 8).

Cohen and Zarowin (2010) argued that “reducing such expenses will boost current period earnings. It could also lead to higher current period cash flows (at the risk of lower future cash flows) if the firm generally paid for such expenses in cash” (p. 8).

During, or upon completion of, an investigation involving accounting and auditing issues, the Division of Enforcement of the US Securities and Exchange Commission (SEC) may take enforcement action against firms, managers, auditors, and other parties involved in violation of federal securities laws by requiring the firm to change its accounting methods, restate its financial statements, and pay damages (Dechow et al. 2011).

Beneish’s (1999a) model will be presented in Chap. 5 and not in Appendix since it is employed in the research design of the present study. Beneish (1997) presented a previous model for detecting earnings manipulation that differed from the M-score in three main ways: Beneish (1997) was estimated with 64 sample companies while Beneish (1999a) used 74 companies, the control sample was different, and the set of explanatory variables also differed.

O’Connor et al. (2006) explained: “CEO duality occurs when a single individual serves as both the CEO of a company and the chair of its board of directors” (p. 485).

McNichols (2002) modified and extended Dechow and Dichev’s (2002) measure.

Jensen (2005) argued: “Like an addictive drug, manning the helm of an overvalued company feels great at first. If you are the CEO or CFO, you are on TV, and covered by the press, investors love you, your options are increasing in value, and the capital markets are wide open to your firm. But as drug users learn, massive pain lies ahead. […]. So as time goes by it begins to dawn on managers of such overvalued firms that times are getting tough. You realize the markets will hammer you unless your company’s performance justifies the stock price. So after all value creating alternatives have been taken you start to take actions that destroy long run value that you hope will at least appear to generate the market’s expected performance in the short run. By doing this you postpone the day of reckoning until you are gone or you figure out how to resolve the issue. To appear to be satisfying growth expectations you use your overvalued equity to make long run value destroying acquisitions; you use your access to cheap debt and equity capital to engage in excessive internal spending and risky negative net present value investments that the market thinks will generate value; and eventually you turn to further accounting manipulation and even fraudulent practices to continue the appearance of growth and value creation” (pp. 9–10).

For example, according to Healy and Wahlen’s (1999) framework, incentives for managing earnings can be classified into three macro areas of motivation: (1) capital market motivations; (2) contracting motivations; and (3) regulatory motivations. The first category of incentives “include studies of earnings management in periods surrounding capital market transactions and when there is a gap between firm performance and analysts’ or investors’ expectations” (p. 371). Contracting motivation studies “test whether the incentives created by lending and compensation contracts can explain earnings management” (p. 375). To sum up, these “studies suggest that compensation and lending contracts induce at least some firms to manage earnings to increase bonus awards, improve job security, and mitigate potential violation of debt covenants” (p. 377). Lastly, regulatory motivation studies explore whether regulation, regulatory scrutiny, or potential regulatory scrutiny creates an incentive for managing earnings.

Specifically, Zhao et al. (2012) found that antitakeover provisions decrease the level of real earnings management.

It is generally assumed that fraudsters are sensitive to the risk of sanctions. According to Dellaportas (2013), taking advantage of an opportunity, when the probability of detection and the severity of the penalties are high, becomes less frequent. However, herein the non-exploitation of an opportunity depends not on extrinsic conditions such as sanctions, consequences, or penalties, but on intrinsic properties or intrinsic enabling conditions of the ‘thief.’ Chapter 3 will clarify this position. At this point, it is sufficient to recognize that some would exploit the opportunity while others would not.

Perry and Williams (1994) showed that managers manipulate discretionary accruals downwards in the year preceding the public announcement of management’s intention to go private, presumably to reduce the share price.

Healy (1985) suggested that when the bonus plan upper bound is largely met, deferring income that exceeds the upper bound does not reduce the current bonus and increases managers’ expected future award.

The modified Jones model proposed by Kothari et al. (2005) involves computing TACC as well.

Where: TACC _i,t is total accruals for firm i in year t; ∆CA _i,t is the change in current assets of firm i in year t; ∆Cash _i,t is the change in cash and cash equivalents of firm i in year t; ∆CL _i,t is the change in current liabilities of firm i in year t; ∆DCL _i,t is the change in debt included in current liabilieties (i.e., current maturities of long-term debt) of firm i in year t; and DEP _i,t is the depreciation and amortization expense of firm i in year t. Changes (∆) are computed between time t and t − 1.

Where: TACC _i,t is total accruals for firm i in year t; EBXI _i,t is earnings before extraordinary items of firm i in year t; and CFO _i,t operational cash flows of firm i in year t.

According to (Peasnell et al. 2000), I use the terms “non-manipulated normal accruals”, “unmanaged accruals”, “non-discretionary accruals”, “expected”, and “normal accruals” interchangeably.

Following conventional practice (Peasnell et al. 2000), I use the terms “manipulated abnormal accruals”, “managed accruals”, “discretionary accruals”, “unexpected”, and “abnormal accruals” interchangeably.

The first term α(1/TA_it−1) is a scaled intercept through prior-year total assets. Like the other variables, the intercept, is scaled by prior-year total assets to avoid heteroscedasticity. While prior research typically does not include a constant in the above model (e.g. Abarbanell and Lehavy 2003; Cohen et al. 2010; Dechow et al. 1995; Subramanyam 1996; etc.) some studies (e.g., Kothari et al. 2005; Zang 2012; etc.) include both, an intercept (α) as well as a scaled intercept α/TA _it−1) in the Jones (1991) and in the modifies Jones model (Dechow et al. 1995). Other studies (e.g., Gaver et al. 1995; Peasnell et al. 2000; Rangan 1998; etc.) do not scale the intercept.

The Jones (1991) model of normal accruals has been estimated cross-sectionally by DeFond and Jiambalvo (1994) first. Cross-sectional versions of the Jones (1991) model are estimated using data from firms matched on year and industry. The cross-sectional Jones model is similar to the original Jones (1991) model but estimated by using cross-sectional, not time-series, data. Therefore, the estimated parameters are industry and year specific rather than firm specific, and are obtained using data from all firms matched on year and two-digit Standard Industrial Classification (SIC) code (Bartov et al. 2001). Matching at a finer level of SIC code (e.g. the four-digit level) may cause a significant loss of firms (DeFond and Jiambalvo 1994). However, Kothari (2001) explains the pros of the cross-sectional approach: “cross-sectional estimation imposes milder data availability requirements for a firm to be included for analysis than time-series estimation. This mitigates potential survivor bias problems. The precision of the estimates is also likely higher in cross-sectional estimation because of larger sample sizes than the number of time-series observations for an individual firm” (p. 163).

Jones (1991) specifies: “ordinary last squares is used to obtain estimates a _i , b_1,i , and b_2,i , of α_i, β_1i , and β_2i respectively” (p. 212).

Accordingly, I use the terms “non-discretionary accruals” and “normal accruals” interchangeably (Peasnell et al. 2000).

More specifically, the value of the mean is a very small number (0,0000000000000000305311).

Following Jones (1991), Dechow et al. (1995) adopted a time-series approach. However, the modified Jones model can be estimated cross-sectionally as well.

Kothari et al. (2005) also add a constant (α₀) to the original Jones (1991) model. A constant provides an additional control for heteroscedasticity (not alleviated by using assets as the deflator), and mitigates problems stemming from an omitted size variable (Kothari et al. 2005).

The first term α₀ is an unscaled intercept while the second term α (1/TA_it−1) is a scaled intercept through prior-year total assets.

I did not provide an example for other well-known earnings management detection tools (except for the Jones 1991 model) since it would have been redundant. However, in this case, an example might be useful because applications of Benford’s Law in earnings management studies are uncommon. In the following example, I will use only a few items from the balance sheet. The interested reader should adapt this example and expand it to all financial data available for a company (e.g., income statement, statement of cash flows and the notes).