Do Managers Smooth Earnings Paths

INTRODUCTION The topic of "income smoothing" (or "income normalization") has received much theoretical and empirical attention. Ronan, Sadan and Snow (1977), Eckel (1981) and Imhoff (1981) provide excellent reviews, and the book by Ronan and Sadan (1981) is thorough. From these works it is clear that there is no standard model of income smoothing, but rather a body of interrelated theories and tests. Articles differ on such fundamental matters as the motivation behind income smoothing, its measurement, and its impact on stockholders. Some of these differences may be attributable in part to multiple definitions of income smoothing. The type of smoothing that we study here "involves the repetitive selection of accounting measurement or reporting rules in a particular pattern, the effect of which is to report a stream of income with a smaller variation from trend than would otherwise have appeared." (Copeland (1968, p. 102), emphasis added). (1) We review theories of (1) management's alleged motivation for smoothing, (2) measures other than Net Income that may be the object of smoothing, and (3) the means by which smoothing may be achieved. Perhaps the most common theory of motivation is that management tries to induce other parties to believe the firm's income stream has low volatility in order to maintain personal position and income. Bartov (1993, p. 843) quoted Fortune when writing, "CEOs know that investors hate surprises, so they try to keep net income trending up a nice straight slope." We also review several empirical studies of income smoothing in order to contrast our empirical test with previous work. The test we conduct compares Net Income (NI) and Cash Flow from Operations (CFO) for 327 Fortune 500 companies for the 1986-1993 period. The CFO data have recently become available through changes in reporting requirements for publicly-held firms. We selected a comparison between CFO and NI because, for reasons we discuss below, the definition and construction of CFO makes it less subject than NI to accounting manipulations. We argue that if such manipulations are used to smooth the income stream, then NI data should exhibit less variability than CFO data. For each of the 327 firms in the sample, we compare the variabilities of the NI and CFO series, using two different measures of volatility for each series: (1) the standard deviation of the (index-based) variable around its linear trend, which is equivalent to a coefficient of variation, and (2) the standard deviation of the variable's annual growth rate. In contrast to the smoothing hypothesis, we find that NI variability exceeds CFO variability for either 81% or 74% of the firms, depending on which of the two variability measures is used. We concur with the assessment made by Albrecht and Richardson (1990, p. 713) who wrote that "The [income-smoothing] practice is conjectured to be widespread, but evidence in support of deliberate smoothing is not convincing." REAL VS. ARTIFICIAL SMOOTHING Dascher and Malcom (1970, pp. 353-354) made a useful distinction between "real" and "artificial" smoothing: "Real smoothing refers to an actual transaction that is undertaken or not undertaken on the basis of its smoothing effect on income, whereas artificial smoothing refers to accounting procedures which are implemented to shift costs and/or revenues from one period to another...Artificial smoothing may be achieved when management has the power to decide which, if any, research projects will be capitalized and the useful life and pattern of allocation for those projects which are capitalized." Similarly, Copeland (1968, p. 102) writes that (among other characteristics) a so-called "perfect" smoothing device "must not require a "real" transaction with second parties, but only a reclassification of internal account balances." He continued, A smoothing device ought to involve only the accounting interpretation of the event, not the event itself. …