Research on the Relationship between Real Activities Manipulation and Earnings Management

This study investigates the relationship between following three kinds of real activities manipulation and earnings management with the purpose of avoiding losses. These three kinds of real activities manipulation are: (1) sales manipulation, (2) reduction of discretionary expense, and (3) overproduction. In addition to the above research, this study examines the cross-sectional features of firms in which managers use real activities manipulation to accomplish their earnings targets. There are six features specified in this study: (1) industry membership; (2) earnings management flexibility (3) stock price motivation; (4) bond covenant; (5) stakeholders motivation; (6) institutional ownership. This study applies the models of normal cash flows, discretionary expenses and production costs from Dechow, Kathari and Watts (1998) and Roychowdhury (2006) to measure the real activities manipulation, and use the pooled cross-sectional approach to detect the firm years which are suspicious for earnings management to avoid annual losses. Based on the samples of public firms in Taiwan from 1996 to 2005, the empirical results are summarized as follow:. Managers tend to manage earnings to avoid annual losses through sales manipulation.. Managers tend to manage earnings to avoid earnings decreases through reduction of discretionary expense.. No statistically significant relationship exists between reduction of discretionary expenses, overproduction and earnings management with the purpose of avoiding annual losses.. Firms with debt outstanding, high level of flexibility in earnings management tend to engage in real activities manipulation.. Firms with relatively high stock price motivation and reliability on stakeholders in the corresponding industry exhibit evidences of sales manipulation and overproduction.. Institutional ownership appears to be more short-term oriented and pressure managers to engage in real activities manipulation.. Industrial membership does not appear a statistically significant relationship with real activities manipulation.1. Introduction 1.1 Motivation 1.2 Objectives 2.3 Structure 3. Literature Review 4.1 Earning threshold and earning management 4.2 Real activity manipulation 7.3 Cross Sectional Variation in earning management 12. Research Design 16.1 Main hypothesis 16.2 Cross sectional variation in real activity manipulation 19.3 Research methodology 23.3.1 Normal cash flow estimation models 23.3.2 Real activities manipulation models 26.3.3 Test for abnormal CFO, discretionary expense, production cost 29.3.4 Test on cross sectional variation 30.4 Sample Selection 32.4.1 Sample selection 32.4.2 Selection of suspect firm years 33. Empirical Results 35.1 Models for real activity manipulation in suspect firm year 35.1.1 Descriptive statistics 35.1.2 Model parameters 38.1.3 The distribution of abnormal CFO, DISEXP and PROD n vicinity of zero earning 39.1.4 Univariate correlation 42.1.5 Empirical result in real activity manipulation 43.2 Cross sectional variation in real activity manipulation 47.2.1 Univairate correlation 47.2.2 Empirical result 51. Sensitivity Analysis 59.1 Another threshold: beat financial analyst forecast 59.2 Another threshold: avoiding earnings decrease 61. Conclusions 64.1 Conclusions 64.2 Research limitations 65ppendix A: Variable description 67eferences 69ablesable 1 Descriptive statistics 37able 2 Model parameters 39able 3 Correlation table 43able 4 Comparison of suspect firm years with rest of samples 44able 5 Comparison of suspect firm years in the vicinity of zero earnings 47able 6 Correlation among variables of cross-sectional variation 49able 7 Correlation among binary-rank variable of cross-sectional variation 49able 8 VIF Test of Cross-Sectional Variation Models 50able 9 VIF Test of Binary-Rank Cross-Sectional Variation Models 50able 10 Cross-sectional variation in veal activities manipulation 55able 11 Cross-sectional variation in real activities manipulation using binary rank variable 56able 12 Comparison of firm-years that beat financial analyst forecast 60able 13 Comparison of firm-years that just have slight EPS increases 63iguresigure 1 Number of firm years by earnings interval 34igure 2 Residual CFO by earnings interval 40igure 3 Residual DISEXP by earnings interval 41igure 4 Residual PROD by earnings interval 4