There is substantial evidence that executives engage in earnings management.1 One means of managing earnings is by manipulation of accruals with no direct cash flow consequences, hereafter referred to as accrual manipulation. Examples include under-provisioning for bad debt expenses and delaying asset write-offs. Managers also have incentives to manipulate real activities during the year to meet certain earnings targets. Real activities manipulation affects cash flows and in some cases, accruals. Much of the current research on earnings management focuses on detecting abnormal accruals. Studies that directly examine earnings management through real activities have concentrated mostly on investment activities, such as reductions in expenditures on research and development.2

My paper contributes to the literature on earnings management by presenting evidence on the management of operational activities, which has received little attention to date. Real activities manipulation is defined as management actions that deviate from normal business practices, undertaken with the primary objective of meeting certain earnings thresholds. The first objective of this paper is to develop empirical methods to detect real activities manipulation. I examine cash flow from operations (CFO), production costs, and discretionary expenses, variables that should capture the effect of real operations better than accruals. Next, I use these measures to detect real activities manipulation around the zero earnings threshold. I find evidence consistent with firms trying to avoid losses by offering price discounts to temporarily increase sales, engaging in overproduction to lower cost of goods sold (COGS), and reducing discretionary expenditures aggressively to improve margins.

There is predictable cross-sectional variation in real activities manipulation to avoid losses. In particular, the presence of sophisticated investors restricts the extent of real activities manipulation. This suggests that even though these activities enable managers to meet short-run earnings targets, they are unlikely to increase long-run firm value. Industry membership, the stock of inventories and receivables, growth opportunities, and the presence of debt are other factors that affect variation in real activities manipulation.

I develop several robustness tests to investigate if the evidence of abnormal real activities among firm-years reporting small annual profits reflect (a) earnings management to avoid losses, or (b) optimal responses to prevailing economic circumstances. The collective evidence from these robustness tests seems more consistent with the earnings management explanation. Finally, I document some evidence of real activities manipulation to meet/beat annual analyst forecasts.

Since Hayn (1995) and Burgstahler and Dichev (1997) found evidence of the discontinuity in frequency of firm-years around zero earnings, academics have had limited success in documenting further evidence of earnings management to avoid losses.3 For example, Dechow et al. (2003) fail to find evidence that firms reporting small profits manage accruals to cross the zero threshold. This paper contributes to the literature by providing evidence consistent with firms relying on real activities manipulation to meet the zero threshold. The evidence in this paper is particularly pertinent in the light of recent papers [Durtschi and Easton (2005), Beaver et al. (2004)] that question whether the observed discontinuities in firm-year distribution around zero can be attributed to earnings management.4

Section 2 discusses the definition of real activities manipulation and previous research. In Section 3, I identify firms that are likely to engage in real activities manipulation and develop hypotheses on how they should differ from the rest of the sample. I also develop hypotheses on cross-sectional variation in real activities manipulation. In Section 4, I discuss my data and estimation models, and present descriptive statistics. Section 5 presents my results. Section 6 discusses the implications of the evidence in this paper, as well as areas for further research.

According to Healy and Wahlen (1999), 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 practices. A number of studies discuss the possibility that managerial intervention in the reporting process can occur not only via accounting estimates and

To detect real activities manipulation to avoid losses, I investigate patterns in CFO, discretionary expenses, and production costs for firms close to the zero earnings benchmark. CFO represents cash flow from operations as reported in the statement of cash flows. Discretionary expenses are defined as the sum of (a) advertising expenses, (b) R&D expenses, and (c) selling, general and administrative (SG&A) expenses.

Production costs are defined as the sum of COGS and change in inventory during

I sample all firms in COMPUSTAT between 1987 and 2001 with sufficient data available to calculate the COMPUSTAT-based variables in Appendix A for every firm-year. I require that cash flow from operations be available on COMPUSTAT from the Statement of Cash Flows. This restricts my sample to the post-1986 period.

Given the primary focus on the zero target, I use annual data for my tests. Recall that the preliminary patterns in CFO detected by Burgstahler and Dichev (1997) are in annual data.

If firm-years that report profits just above zero undertake activities that adversely affect their CFO, then the abnormal CFO for these firm-years, calculated using the industry-year model described in Section 4.2, should be negative compared to the rest of the sample. To test this, I estimate the following regression:=+1(SIZE)-1+2(MTB)-1+3(Netincome)+4(SUSPECT_NI)+.

In this case, the dependent variable, Yt, is abnormal CFO in period t. Regression (6) is also estimated with abnormal 

This paper complements the existing literature on earnings management in several ways. First, this study develops empirical methods to detect real activities manipulation in large samples. In prior literature on real activities manipulation, the focus has mostly been limited to the reduction of discretionary expenditures. Second, the paper documents evidence consistent with real activities manipulation around earnings thresholds commonly discussed in the literature, in particular, the zero