Earnings Management to Exceed Thresholds Author(s): François Degeorge, Jayendu Patel and Richard Zeckhauser Source: The Journal of Business , Vol. 72, No. 1 (January 1999), pp. 1-33 Published by: The University of Chicago Press Stable URL: https://www.jstor.org/stable/10.1086/209601 JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected]. Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at https://about.jstor.org/terms The University of Chicago Press is collaborating with JSTOR to digitize, preserve and extend access to The Journal of Business This content downloaded from 206.253.207.235 on Thu, 21 May 2020 21:16:00 UTC All use subject to https://about.jstor.org/terms FrancËois Degeorge Hautes EÂ tudes Commerciales and the Centre for Economic Policy Research Jayendu Patel Boston University Richard Zeckhauser Harvard University and National Bureau of Economic Research Earnings Management to Exceed Thresholds* I. Introduction Earnings provide im- portant information for Analysts, investors, senior executives, and boards investment decisions. of directors consider earnings the single most im- Thus executivesÐwho portant item in the ®nancial reports issued by are monitored by in- publicly held ®rms. In the medium to long term vestors, directors, cus- tomers, and suppli- (1±10-year intervals), returns to equities appear ersÐacting in self- to be explained overwhelmingly by the ®rm's cu- interest and at times mulative earnings during the period; other plausi- for shareholders, have ble explanationsÐsuch as dividends, cash ¯ows, strong incentives to manage earnings. We introduce behavioral thresholds for earnings * We thank the David Dreman Foundation for fund- management. A model ing support; Degeorge also thanks the Fondation Hautes EÂ tudes Commerciales for research support. The data on ana- shows how thresholds lysts' forecasts of earnings were provided by I/B/E/S Interna- induce speci®c types of tional Inc. (post-1984 period) and by Q Prime (pre-1984). We earnings management. have bene®ted from helpful comments by Raj Aggarwal, Empirical explorations Shlomo Benartzi, Bengt Holmstrom, David King, Todd Mil- identify earnings man- bourn, Clyde Stickney, Richard Thaler, Kent Womack, and a agement to exceed each referee. We have also bene®ted from helpful comments from of three thresholds: re- seminar participants at the Behavioral Finance Working Group port positive pro®ts, of the National Bureau of Economic Research; Boston Uni- sustain recent perfor- versity; the Centre for Economic Policy Research European Summer Symposium in Financial Markets at Studienzentrum mance, and meet ana- Gerzensee, Switzerland; the European Finance Association lysts' expectations. The meetings, Vienna, 1997; the French Finance Association meet- positive pro®ts thresh- ings, Grenoble, 1997; Harvard University; the Hautes EÂ tudes old proves predomi- Commerciales; the European Institute of Business Administra- nant. The future perfor- tion (INSEAD); the Q Group; and the Amos Tuck School, mance of ®rms suspect Dartmouth College. for boosting earnings just across a threshold (Journal of Business, 1999, vol. 72, no. 1) is poorer than that of 1999 by The University of Chicago. All rights reserved. 0021-9398/99/7201-0001$02.50 control group ®rms. 1 This content downloaded from 206.253.207.235 on Thu, 21 May 2020 21:16:00 UTC All use subject to https://about.jstor.org/terms 2 Journal of Business or capital investmentsÐhave marginal correlations close to zero (Eas- ton, Harris, and Ohlson 1992; Kothari and Sloan 1992). Even for short- term equity returns, earnings are an important explanatory factor.1 The rewards to a ®rm's senior executivesÐboth employment deci- sions and compensation bene®tsÐdepend both implicitly and explic- itly on the earnings achieved on their watch (Healy 1985). But such executives have considerable discretion in determining the ®gure printed in the earnings report for any particular period. Within generally accepted accounting principles (GAAP), executives have considerable ¯exibility in the choice of inventory methods, allowance for bad debt, expensing of research and development, recognition of sales not yet shipped, estimation of pension liabilities, capitalization of leases and marketing expenses, delay in maintenance expenditures, and so on. Moreover, they can defer expenses or boost revenues, say, by cutting prices. Thus, executives have both the incentive and ability to manage earnings. It is hardly surprising that the popular press frequently de- scribes companies as engaged in earnings managementÐsometimes referred to as manipulation.2 This article studies earnings management as a response to implicit and explicit rewards for attaining speci®c levels of earnings, such as positive earnings, an improvement over last year, or the market's con- sensus forecast. We label as ``earnings management'' (EM) the strate- gic exercise of managerial discretion in in¯uencing the earnings ®gure reported to external audiences (see Schipper 1989). It is accomplished principally by timing reported or actual economic events to shift in- come between periods. We sketch a model that predicts how executives strategically in¯u- ence the earnings ®gures that their ®rms report to external audiences and then examine historical data to con®rm such patterns. Our model incorporates behavioral propensities and a stylized description of the interactions among executives, investors, directors, and earnings ana- lysts to identify EM patterns that generate speci®c discontinuities and distortions in the distribution of observed earnings.3 We do not determine which components of earnings or of supple- mentary disclosures are adjusted. Nor do we attempt to distinguish em- pirically between ``direct'' EMÐthe strategic timing of investment, 1. Ball and Brown (1968) is the classic early work; see Dechow (1994) and references there for subsequent research that details the relevance of earnings. 2. See, e.g., the multipage stories ``Excuses Aplenty When Companies Tinker with Their Pro®ts Reports,'' New York Times (June 23, 1996), and ``On the Books, More Facts and Less Fiction,'' New York Times (February 16, 1997). A recent studyÐBruns and Merchant (1996, p. 25)Ðconcludes that ``we have no doubt that short-term earnings are being manipulated in many, if not all, companies.'' 3. DeBondt and Thaler (1995, pp. 385±410) provide a discussion of behaviorally moti- vated ®nancial decisions by ®rms. This content downloaded from 206.253.207.235 on Thu, 21 May 2020 21:16:00 UTC All use subject to https://about.jstor.org/terms Earnings Management 3 sales, expenditures, or ®nancing decisionsÐand ``misreporting''Ð EM involving merely the discretionary accounting of decisions and out- comes already realized.4 We identify three thresholds that help drive EM: the ®rst is to report pro®tsÐfor example, 1 penny a share. This threshold arises from the psychologically important distinction between positive numbers and negative numbers (or zero). The second and third benchmarks rely on performance relative to widely reported ®rm-speci®c values. If the ®rm does as well or better than the benchmark, it is met; otherwise it is failed. The two benchmarks are performance relative to the prior com- parable period and relative to analysts' earnings projections. Perfor- mance relative to each benchmark is assessed by examining the sprin- kling of quarterly earnings reports in its neighborhood. A big jump in density at the benchmark demonstrates its importance. Burgstahler and Dichev (1997) examine the management of earnings to meet our ®rst two thresholds, though not in relation to analysts' estimates.5 Their analysis delves more deeply into accounting issues and identi®es the ``misreporting'' mechanismsÐfor example, the ma- nipulation of cash ¯ow from operations, or changes in working capi- talÐthat permit earnings to be moved from negative to positive ranges. We devote considerably more attention to the motivations for EM, con- sider direct EM (e.g., lowering prices to boost sales) in addition to misreporting, provide an optimizing model on how earnings are man- aged, and analyze the consequences of management for future earnings. In addition, we explore EM as the executive's (agent's) response to steep rewardsÐreaping a bonus or retaining a jobÐthat depend on meeting a bright threshold.6 Finally, we look at the hierarchy among our thresholds. Earnings management arises from the game of information disclo- sure that executives and outsiders must play. Investors base their deci- sions on information received from analystsÐusually indirectly, say, through a brokerÐand through published earnings announcements. To bolster investor interest, executives manage earnings, despite the real earnings sacri®ce. Other parties, such as boards of directors, analysts, and accountants, participate in this game as well, but their choices are exogenous to our analysis. For example, the contingent remuneration actions of boards are known to executives. Presumably such pay pack- ages are structured to take distorting possibilities into account and may 4. Foster (1986, p. 224) discusses mechanisms for misreporting transactions or events in ®nancial statements. 5. Payne and Robb (1997) show that managers use discretionary accrual to align earn- ings with analysts' expectations. 6. Burgstahler (1997) adds a model in which earnings are manipulated because the mar- ginal bene®t of reporting higher earnings is greatest in some middle range. This content downloaded from 206.253.207.235 on Thu, 21 May 2020 21:16:00 UTC All use subject to https://about.jstor.org/terms 4 Journal of Business have been adjusted somewhat to counter EM.7 If so, ®nding evidence of management is more signi®cant. Executives may also distort earnings reports in a self-serving man- ner, imposing an agency loss that reduces the ®rm's value if their incen- tives are not fully aligned with those of shareholders.