UNIVERSITY OF ECONOMICS ERASMUS UNVERSITY ROTTERDAM HO CHI MINH CITY INSTITUTE OF SOCIAL STUDIES VIETNAM THE NETHERLANDS VIETNAM – THE NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS EARNINGS MANAGEMENT BY REAL ACTIVITIES MANIPULATION: A LOOK AT VIETNAM BY NGUYEN DUY ANH MASTER OF ARTS IN DEVELOPMENT ECONOMICS HO CHI MINH CITY, November 2016 UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES HO CHI MINH CITY THE HAGUE VIETNAM THE NETHERLANDS VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS EARNINGS MANAGEMENT BY REAL ACTIVITIES MANIPULATIONS: A LOOK AT VIETNAM A thesis submitted in partial fulfilment of the requirements for the degree of MASTER OF ARTS IN DEVELOPMENT ECONOMICS By NGUYEN DUY ANH Academic Supervisor: Vu Viet Quang HO CHI MINH CITY, November 2016 DECLARATION This is to certify that this thesis entitled “Earnings Management By Real Activities Manipulations: A Look At VietNam”, which is submitted by me in fulfillment of the requirements for the degree of Master of Art in Development Economics to Viet Nam – The Netherlands Programme (VNP) To the best of my knowledge, my thesis does not infringe on anyone’s copyright nor violate any proprietary rights and that any ideas, techniques, quotation, or any other material from the work of other researchers in my thesis, published or otherwise, are fully acknowledge in accordance with the standard referencing practices HCMC, November 30th, 2016 Nguyen Duy Anh i ACKNOWLEDGEMENT I would like to thank my supervisor, Dr Vu Viet Quang for his comprehensive guidance, great support and valuable advice he has given through my research study I have been very lucky to a supervisor who took a high cared about my work and who respond to my question He consistently allowed this paper to be my work but steered me in the right the direction whenever he thought I needed it His careful editing contributed enormously to the production of this thesis I also would like to thank my co-supervisor Dr Truong Dang Thuy for his enthusiastic support, availability and constructive suggestion, which help me overcome the challenge and highpressured situation during the time of research I would like to express my gratitude to all lecturers of the Vietnam- Netherlands Program who have provided an interesting lesson to build my economic knowledge during this program Besides, completing this work would have been difficult if it is not supported by my best friends I am indebted to them for their help Moreover, I wish to thank all my friends who are in VNP 21 who share unforgettable memories in this program Finally, there are also words of deep gratitude for my family who support and encourage me when I implement my postgraduate studies ii ABSTRACT This research thesis tests three hypothesizes: (i) In Viet Nam, the listed companies that meet earnings target (zero earnings and zero earnings growth) exhibit the proof of real activities manipulation (ii) Hypothesis 2: there is the difference between the extent of using real activities manipulation of the listed firm that meet benchmark and those that meet earning benchmark and have a high value of an asset (iii) there is no relationship between firm using real activities manipulation to just meet earnings benchmark and future performance Our tests were based on data included 2374 firm-year observation covering 2005 to 2015.We focus on companies that satisfy one of the criteria: zero earnings or zero earnings growth The firms meet the criteria call suspected firms or belong to suspected firms group The rules identify the firms that more likely to using real activities manipulation We examine three types of real earnings management: (1) cutting discretionary expenditures (2) acceleration of timing of sales (sale manipulation) (3) reducing the cost of production To measure real earnings management, we follow the cross-section model developed by Roychowdhury (2006), Gunny (2010); and estimate abnormal production cost, abnormal discretionary spending (sum of SG&A, R&D, and advertising) and abnormal CFO Our finding is that Vietnamese listed firmed apply real activities management to meet earnings benchmark Besides, the degree effect of cutting production cost of the suspected firms with a high value of assets is highest (11.14%) among three types of real activities management (2.54% for sale manipulation and 0.398% for reducing discretionary expenses, which suggest the firm with a good reputation prefer employ cutting production cost to meet companies target Final, the companies which engage in CFO manipulation or cutting discretionary expenses, as real earnings management to just meet earnings benchmarks have no impact on subsequent performance In contrast, the companies which engage in production cost as real earnings management to just meet earnings benchmarks have negative impact on subsequent performance Keyword: Capital markets; Accounting choice; Earnings manipulation iii Table of Contents DECLARATION i ACKNOWLEDGEMENT ii ABSTRACT iii CHAPTER INTRODUCTION 1.1 Problem statement 1.2 Research objective 1.3 Research questions 1.4 Structure of study CHAPTER LITERATURE REVIEW 2.1 Key concept 2.1.1 Definition of earnings management 2.1.2 Real activities management definition 2.1.3 Accruals 2.2 Models to detect the use of earnings management 2.2.1 2.3 Jones model Incentive to earnings management 11 2.3.1 Debt Covenants 11 2.3.2 Compensation 12 2.3.3 Insider Trading 14 2.3.4 Management buyout 14 2.4 Beating benchmark (zero earnings) 15 2.5 Empirical research on real earnings management 19 2.5.1 Type of real earnings manipulation 21 2.5.2 Impact of real activity management on firm future performance 23 2.6 Hypotheses development 24 CHAPTER 26 METHODOLOGY AND DATA DESCRIPTION 26 iv 3.1 Selection of suspect firm-year 26 3.2 Real earnings proxies 26 3.3 Empirical model 28 3.3.1 Empirical model to test hypothesis 28 3.3.2 Empirical model to test hypothesis 29 3.3.3 Empirical model to test hypothesis 30 3.3 Data collection 31 3.3.1 Data collection 31 3.3.2 Variable descriptions 32 CHAPTER 35 RESULTS AND DICUSSION 35 4.1 Selected suspected firm-year 35 4.2 Descriptive statistics 36 4.3 Testing the assumptions of panel data regression 40 4.3.1 Multicollinearity 40 4.3.2 Autocorrelation 41 4.4 Empirical and discussion 42 4.4.1 Empirical evidence on the real activities manipulation of Companies listed in the Viet Nam stock market 42 4.6.2 Discussion about real activities manipulation 45 4.6.3 Size effect on real earnings management 49 4.6.4 The accociation between using real activites managmet to meet earnings benchmark and future performance 52 CHAPTER 55 CONCLUSION, CONTRIBUTION, AND LIMITATION 55 5.1 Main finding 55 5.2 Contribution 57 5.3 Limitation and further research 57 Reference 59 Appendix 63 v List of Table and Figure List of Tables Table 1: Earnings management definition Table 2: Calculating abnormal accruals 11 Table 3: Variable descriptions 32 Table 4: Description statistics for sample firms in 2005-2015 period 37 Table 5: Pearson Correlation matrix 40 Table 6: Variance inflation factor 41 Table 7: Comparison of suspect firm and the rest of sample 45 Table 8: Capture net effect when firms combine one more type of real earnings management 49 Table 9: Multivariate regression analyses of size effect on earnings management in Vietnam listed firms 50 Table: 10 Regression between real activities and future performance 53 List of Figure Figure 1: Framework for understanding the practice of Accounts Manipulation Figure 2: A hypothetical value function 16 Figure 3: The losses have more impact on than an equivalent amount of gain 17 Figure 4: The distribution of changes in net income divided by market value of equity at the beginning of the year 18 Figure 5: histogram of distribution of net income scaled by lagged total assets (Figure 5A) and histogram of distribution of change in net income scaled by lagged total assets (Figure 5B) 35 Figure 6: Comparing mean of the suspect firm group (include observation of firm with zero earnings and firm with zero earnings growth) and men of the rest of sample 39 vi ABBREVIATION RM REM Real management Real earnings management vii CHAPTER INTRODUCTION 1.1 Problem statement With the explosion of the Dot-Com bubble in 2000, the previous stock which used to be bullish now became going down Then, in the end, the awful reality kept up with the firms which were supposed to try to hide the unpleasant truth in their financial reports The beginning of the long list big scandal happened in 2000 when Xerox admitted that over a four year – period, their income had been overstated by US$1 billion They boosted their income by reported revenue from the lease of printers and copier in the long-term too early Unfortunately, it was not an isolated example Enron Corporation uses a special purpose accounting entities which help them hide billion dollars worth of debt away its balance sheet WorldCom employed a simple scheme to change more than US$ 11 billion of cost to assets Tyco International executives were accused of covering million US dollar debt, which they borrowed from employees with interest – free or very low-interest loans and not disclose these loans After investigation of the U.S Securities and Exchange Commission, Quest Communication was forced to adjust their profit by US $2.4 billion because Quest Communication reported impressive income from the transaction is booked as revenue without receiving money The series of financial accounting scandal still goes on As a result of the scandals, the collapse of cooperation lead to hundreds of billion dollars in loss for investors, thousand job losses The collapse of WorldCom in May 2002 is the biggest, with approximately US $180 billion in the loss and also 30.000 lost jobs Besides financial accounting scandals, it also highlights the failures to audit financial statement correctly For example, Arthur Andersen LLP, which was one member of Big accounting firm in the world, prepared Enron’s financial report Andersen also collapses after Enron scandal According to the Securities and Exchange Commission investigation, firstly, Andersen found out many “trouble” transactions which were highly risky, but Andersen audit firm received a million fees, so Andersen did not give their opinion of these risk Further, Andersen ordered their company’s Houston office destroy a thousand of the document to prevent Securities and Exchange Commission from investigating Enron’s bankruptcy These Arthur Andersen scandals reduce the faith of investor in the integrity of the audit firm After all, United States of America enacted Sarbanes-Oxley Act in July 2002 to improve the business environment and protect investors Ab_CFO𝑖,𝑡 = 𝛼0 + 𝛽1 𝑆𝐼𝑍𝐸𝑖,𝑡−1 + 𝛽2 𝑀𝑇𝐵𝑖,𝑡−1 + 𝛽3 𝑅𝑂𝐴𝑖,𝑡 + 𝛽4 𝑆𝑢𝑠𝑝𝑒𝑐𝑡2𝑖,𝑡 + 𝜀𝑖𝑡 xtgls uhat_cfo (4-6) size mtb netincome1_la suspect_changeinnetincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 301 uhat_cfo Coef size mtb netincome1_la suspect_changeinnetincome1_la _cons 0073065 -.003265 6024235 0202111 -.2013306 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 Std Err .0015304 0044133 0332212 0060113 0304875 end of do-file 73 z 4.77 -0.74 18.13 3.36 -6.60 = = = = = = = P>|z| 0.000 0.459 0.000 0.001 0.000 1001 301 3.325581 18005.65 0.0000 [95% Conf Interval] 004307 -.0119148 5373111 0084292 -.261085 010306 0053848 6675359 0319931 -.1415761 6.5 Results of RM_1 RM_2 RM_3 RM_1 RM_2 RM_3 with suspected firms that meet zero earnings benchmark RM_1 xtgls uhat_cfo_disexp size mtb netincome1_la suspect_netincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 299 uhat_cfo_disexp Coef size mtb netincome1_la suspect_netincome1_la _cons -.0030223 0140639 -.4807488 -.0229409 0896416 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 Std Err .001525 006285 0477939 0052404 0317875 end of do-file 74 z -1.98 2.24 -10.06 -4.38 2.82 P>|z| 0.047 0.025 0.000 0.000 0.005 = = = = = = = 993 299 3.32107 121.87 0.0000 [95% Conf Interval] -.0060112 0017455 -.574423 -.033212 0273393 -.0000334 0263823 -.3870745 -.0126699 1519438 RM_2 xtgls uhat_prod_disexp size mtb netincome1_la suspect_netincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 233 uhat_prod_disexp Coef size mtb netincome1_la suspect_netincome1_la _cons 0164289 -.0130221 -.5136565 -.0662248 -.2946097 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 Std Err .0011333 0038515 0273267 00634 0251083 75 z 14.50 -3.38 -18.80 -10.45 -11.73 P>|z| 0.000 0.001 0.000 0.000 0.000 = = = = = = = 672 233 2.88412 1235.61 0.0000 [95% Conf Interval] 0142076 -.020571 -.5672159 -.078651 -.343821 0186501 -.0054733 -.4600971 -.0537986 -.2453984 RM_3 xtgls uhat_cfo_prod_disexp size mtb netincome1_la suspect_netincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 233 uhat_cfo_prod_disexp Coef size mtb netincome1_la suspect_netincome1_la _cons 0151158 -.0298553 -.9272691 -.0800977 -.2371513 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 Std Err .0035066 0132836 0925559 0089739 0720495 76 z 4.31 -2.25 -10.02 -8.93 -3.29 P>|z| 0.000 0.025 0.000 0.000 0.001 = = = = = = = 661 233 2.83691 4183.83 0.0000 [95% Conf Interval] 008243 -.0558908 -1.108675 -.0976861 -.3783656 0219887 -.0038199 -.7458628 -.0625093 -.0959369 RM_1 RM_2 RM_3 with suspected firms that meet zero earnings growth benchmark RM_1 xtgls uhat_cfo_disexp size mtb netincome1_la suspect_changeinnetincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 299 uhat_cfo_disexp Coef size mtb netincome1_la suspect_changeinnetincome1_la _cons -.0046401 0126615 -.4613856 -.0255184 1248876 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 Std Err .0018928 006076 0462706 0067869 0387566 77 z -2.45 2.08 -9.97 -3.76 3.22 = = = = = = = P>|z| 0.014 0.037 0.000 0.000 0.001 993 299 3.32107 141.04 0.0000 [95% Conf Interval] -.0083499 0007528 -.5520744 -.0388204 0489261 -.0009303 0245701 -.3706968 -.0122163 2008492 RM_2 xtgls uhat_prod_disexp size mtb netincome1_la suspect_changeinnetincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 233 uhat_prod_disexp Coef size mtb netincome1_la suspect_changeinnetincome1_la _cons 0158648 -.0044362 -.4761956 -.0248194 -.2978181 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 Std Err .0011144 0056003 0250117 0057565 022049 78 z 14.24 -0.79 -19.04 -4.31 -13.51 = = = = = = = P>|z| 0.000 0.428 0.000 0.000 0.000 672 233 2.88412 862.55 0.0000 [95% Conf Interval] 0136807 -.0154126 -.5252177 -.0361019 -.3410333 0180489 0065403 -.4271735 -.0135368 -.2546029 RM_3 xtgls uhat_cfo_prod_disexp size mtb netincome1_la suspect_changeinnetincome1_la, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 233 Number of obs Number of groups Obs per group: avg max Wald chi2(4) Prob > chi2 uhat_cfo_prod_disexp Coef size mtb netincome1_la suspect_changeinnetincome1_la _cons 0118916 -.0177358 -.9502728 -.0385825 -.1843703 79 Std Err .0042399 0134575 0859475 0129894 0871352 z 2.80 -1.32 -11.06 -2.97 -2.12 = = = = = = = P>|z| 0.005 0.188 0.000 0.003 0.034 661 233 2.83691 1918.65 0.0000 [95% Conf Interval] 0035817 -.044112 -1.118727 -.0640413 -.3551522 0202016 0086404 -.7818187 -.0131237 -.0135885 6.6 Result of model 5: Size effect on real earnings management 𝑌𝑡 = 𝛼0 + 𝛽1 𝑆𝐼𝑍𝐸𝑡−1 + 𝛽2 𝑀𝑇𝐵𝑡−1 + 𝛽3 𝑅𝑂𝐴𝑡 + 𝛽4 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑡 + 𝐷𝑆𝑖𝑧𝑒 + 𝐷𝑆𝑖𝑧𝑒 ∗ 𝑆𝑢𝑠𝑝𝑒𝑐𝑡 + 𝜀𝑡 (moded 5) xtgls uhat_cfo size mtb netincome1_la suspect_netincome1_la dummy_a d_a_suspect_netincome1 ,p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 301 uhat_cfo Coef size mtb netincome1_la suspect_netincome1_la dummy_a d_a_suspect_netincome1 _cons 0167437 -.0072444 6125457 0243582 -.0237372 -.0226501 -.382814 Number of obs Number of groups Obs per group: avg max Wald chi2(6) Prob > chi2 Std Err .0026811 0055093 031123 0085084 0089259 0120535 0508124 80 z 6.25 -1.31 19.68 2.86 -2.66 -1.88 -7.53 P>|z| 0.000 0.189 0.000 0.004 0.008 0.060 0.000 = = = = = = = 1001 301 3.325581 693.57 0.0000 [95% Conf Interval] 0114888 -.0180424 5515457 007682 -.0412316 -.0462746 -.4824045 0219986 0035536 6735456 0410343 -.0062428 0009743 -.2832236 xtgls uhat_prod size mtb netincome1_la suspect_netincome1_la dummy_a d_a_suspect_netincome1 ,p(h) c(p) force (note: 56 observations dropped because only obs in group) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic panel-specific AR(1) Estimated covariances = Estimated autocorrelations = Estimated coefficients = 178 178 uhat_prod Coef size mtb netincome1_la suspect_netincome1_la dummy_a d_a_suspect_netincome1 _cons 0365776 -.0331299 -1.055683 -.0804571 -.0501456 0191708 -.6511102 Number of obs Number of groups Obs per group: avg max Wald chi2(6) Prob > chi2 Std Err .0025474 0055041 0462687 006317 0034081 0101623 0488291 z 14.36 -6.02 -22.82 -12.74 -14.71 1.89 -13.33 81 P>|z| 0.000 0.000 0.000 0.000 0.000 0.059 0.000 = = = = = = = 621 178 3.488764 7091.83 0.0000 [95% Conf Interval] 0315847 -.0439178 -1.146368 -.0928381 -.0568253 -.0007469 -.7468134 0415704 -.022342 -.9649984 -.0680761 -.0434658 0390885 -.5554069 xtgls uhat_disexp size mtb netincome1_la suspect_netincome1_la dummy_a d_a_suspect_netincome1 ,p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 299 uhat_disexp Coef size mtb netincome1_la suspect_netincome1_la dummy_a d_a_suspect_netincome1 _cons 001463 006249 0747616 -.0072957 -.0054926 0097198 -.0416226 Number of obs Number of groups Obs per group: avg max Wald chi2(6) Prob > chi2 Std Err .0009115 0020695 0150435 002131 0021741 0033269 0182202 z 1.61 3.02 4.97 -3.42 -2.53 2.92 -2.28 82 P>|z| 0.108 0.003 0.000 0.001 0.012 0.003 0.022 = = = = = = = 1011 299 3.381271 109.25 0.0000 [95% Conf Interval] -.0003235 0021929 045277 -.0114723 -.0097537 0031992 -.0773335 0032495 0103052 1042462 -.0031191 -.0012314 0162403 -.0059117 𝑌𝑡 = 𝛼0 + 𝛽1 𝑆𝐼𝑍𝐸𝑡−1 + 𝛽2 𝑀𝑇𝐵𝑡−1 + 𝛽3 𝑅𝑂𝐴𝑡 + 𝛽4 𝑆𝑢𝑠𝑝𝑒𝑐𝑡2𝑡 + 𝐷𝑆𝑖𝑧𝑒 + 𝐷𝑆𝑖𝑧𝑒 ∗ 𝑆𝑢𝑠𝑝𝑒𝑐𝑡 + 𝜀𝑡 (moded 5) xtgls uhat_cfo size mtb netincome1_la suspect_changeinnetincome1_la dummy_a d_a_suspect_changeinnetincome1 ,p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 301 Number of obs Number of groups Obs per group: avg max Wald chi2(6) Prob > chi2 uhat_cfo Coef size mtb netincome1_la suspect_changeinnetincome1_la dummy_a d_a_suspect_changeinnetincome1 _cons 0164537 -.0085074 5856108 0234022 -.0242906 -.0192166 -.3737213 Std Err .0025276 0052147 0330563 0073573 0083369 0126 047837 z 6.51 -1.63 17.72 3.18 -2.91 -1.53 -7.81 = = = = = = = 1001 301 3.325581 677.41 0.0000 P>|z| 0.000 0.103 0.000 0.001 0.004 0.127 0.000 [95% Conf Interval] 0114996 -.018728 5208215 0089822 -.0406307 -.0439121 -.4674802 0214078 0017131 6504 0378222 -.0079506 0054789 -.2799625 xtgls uhat_prod size mtb netincome1_la suspect_changeinnetincome1_la dummy_a d_a_suspect_changeinnetincome1 ,p(h) c(p) force (note: 56 observations dropped because only obs in group) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic panel-specific AR(1) Estimated covariances = Estimated autocorrelations = Estimated coefficients = 178 178 Number of obs Number of groups Obs per group: avg max Wald chi2(6) Prob > chi2 uhat_prod Coef size mtb netincome1_la suspect_changeinnetincome1_la dummy_a d_a_suspect_changeinnetincome1 _cons 0341688 -.020663 -.9683161 -.0322851 -.0500402 -.0064495 -.6247292 Std Err .0028816 0053985 0533691 0023005 0060522 0133237 0556953 z 11.86 -3.83 -18.14 -14.03 -8.27 -0.48 -11.22 83 = = = = = = = 621 178 3.488764 8817.78 0.0000 P>|z| 0.000 0.000 0.000 0.000 0.000 0.628 0.000 [95% Conf Interval] 0285209 -.0312439 -1.072918 -.0367939 -.0619023 -.0325634 -.7338899 0398167 -.0100822 -.8637146 -.0277762 -.0381781 0196644 -.5155685 xtgls uhat_disexp size mtb netincome1_la suspect_changeinnetincome1_la dummy_a d_a_suspect_changeinnetincome1 ,p(h) c(p) force (note: 60 observations dropped because only obs in group) Cross-sectional time-series FGLS regression Coefficients: generalized least squares Panels: heteroskedastic Correlation: panel-specific AR(1) Estimated covariances = Estimated autocorrelations = Estimated coefficients = 239 239 Number of obs Number of groups Obs per group: avg max Wald chi2(6) Prob > chi2 uhat_disexp Coef size mtb netincome1_la suspect_changeinnetincome1_la dummy_a d_a_suspect_changeinnetincome1 _cons 002687 0076011 1184069 -.0051944 -.0048585 0077576 -.0688217 Std Err .0005914 0011805 0063456 0009761 0012876 0017172 0113518 z 4.54 6.44 18.66 -5.32 -3.77 4.52 -6.06 84 = 951 = 239 = = 3.979079 = = 1968.38 = 0.0000 P>|z| 0.000 0.000 0.000 0.000 0.000 0.000 0.000 [95% Conf Interval] 0015279 0052874 1059697 -.0071076 -.007382 0043919 -.0910709 0038461 0099147 1308441 -.0032813 -.0023349 0111233 -.0465725 6.7 Result of model The accociation between using real activites managmet to meet earnings benchmark and future performance ROA𝑖,𝑡+1 = 𝛼0 + 𝛽1 𝐵𝐸𝐴𝑇𝑖,𝑡 + 𝛽2 𝐽𝑈𝑆𝑇𝑀𝐼𝑆𝑆𝑖,𝑡 + 𝛽3 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑖,𝑡 + 𝛽4 𝑅𝑀𝑖,𝑡 + 𝛽5 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑖,𝑡 ∗ 𝑅𝑀𝑖,𝑡 + 𝛽6 𝑅𝑂𝐴𝑖,𝑡 + 𝛽8 𝑀𝑇𝐵𝑖,𝑡 + 𝛽7 𝑆𝐼𝑍𝐸𝑖,𝑡 (model 6) With RM = RM_CFO xtgls f.netincome1_la beat miss suspect_netincome1_la rm_cfo rm_cfo_suspect_netincome1_la netincome1_la mtb size, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 238 F.netincome1_la Coef beat miss suspect_netincome1_la rm_cfo rm_cfo_suspect_netincome1_la netincome1_la mtb size _cons -.0182027 -.0250454 -.0232451 0044936 -.0003114 5651399 0206887 -.0016338 0542093 Number of obs Number of groups Obs per group: avg max Wald chi2(8) Prob > chi2 Std Err .0067242 008361 0068105 0016629 0050084 0191064 0018778 0003526 0089776 z -2.71 -3.00 -3.41 2.70 -0.06 29.58 11.02 -4.63 6.04 85 = = = = = = = P>|z| 0.007 0.003 0.001 0.007 0.950 0.000 0.000 0.000 0.000 698 238 2.932773 5043.67 0.0000 [95% Conf Interval] -.0313819 -.0414328 -.0365934 0012344 -.0101276 527692 0170083 -.0023249 0366135 -.0050235 -.0086581 -.0098968 0077527 0095048 6025878 024369 -.0009427 0718051 ROA𝑖,𝑡+1 = 𝛼0 + 𝛽1 𝐵𝐸𝐴𝑇𝑖,𝑡 + 𝛽2 𝐽𝑈𝑆𝑇𝑀𝐼𝑆𝑆𝑖,𝑡 + 𝛽3 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑖,𝑡 + 𝛽4 𝑅𝑀𝑖,𝑡 + 𝛽5 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑖,𝑡 ∗ 𝑅𝑀𝑖,𝑡 + 𝛽6 𝑅𝑂𝐴𝑖,𝑡 + 𝛽8 𝑀𝑇𝐵𝑖,𝑡 + 𝛽7 𝑆𝐼𝑍𝐸𝑖,𝑡 (model 6) With RM = RM_DISEXP xtgls f.netincome1_la beat miss suspect_netincome1_la rm_disexp rm_disexp_suspect_netincome1_la netincome1_la mtb size, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 237 Number of obs Number of groups Obs per group: avg max Wald chi2(8) Prob > chi2 F.netincome1_la Coef beat miss suspect_netincome1_la rm_disexp rm_disexp_suspect_netincome1_la netincome1_la mtb size _cons -.0131044 -.0169412 -.0176892 -.002237 0079686 5686374 021608 -.0019936 0563477 Std Err .0015059 0027259 0020239 0015071 0058093 0204244 0017565 000404 0073385 z -8.70 -6.21 -8.74 -1.48 1.37 27.84 12.30 -4.93 7.68 86 = = = = = = = 702 237 2.962025 10432.04 0.0000 P>|z| 0.000 0.000 0.000 0.138 0.170 0.000 0.000 0.000 0.000 [95% Conf Interval] -.0160559 -.0222839 -.0216561 -.0051908 -.0034174 5286064 0181653 -.0027855 0419646 -.0101529 -.0115984 -.0137223 0007169 0193546 6086684 0250506 -.0012017 0707309 ROA𝑖,𝑡+1 = 𝛼0 + 𝛽1 𝐵𝐸𝐴𝑇𝑖,𝑡 + 𝛽2 𝐽𝑈𝑆𝑇𝑀𝐼𝑆𝑆𝑖,𝑡 + 𝛽3 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑖,𝑡 + 𝛽4 𝑅𝑀𝑖,𝑡 + 𝛽5 𝑆𝑢𝑠𝑝𝑒𝑐𝑡1𝑖,𝑡 ∗ 𝑅𝑀𝑖,𝑡 + 𝛽6 𝑅𝑂𝐴𝑖,𝑡 + 𝛽8 𝑀𝑇𝐵𝑖,𝑡 + 𝛽7 𝑆𝐼𝑍𝐸𝑖,𝑡 (model 6) With RM = RM_PROD xtgls f.netincome1_la beat miss suspect_netincome1_la rm_prod rm_prod_suspect_netincome1_la netincome1_la mtb size, p(h) Cross-sectional time-series FGLS regression Coefficients: Panels: Correlation: generalized least squares heteroskedastic no autocorrelation Estimated covariances = Estimated autocorrelations = Estimated coefficients = 188 F.netincome1_la Coef beat miss suspect_netincome1_la rm_prod rm_prod_suspect_netincome1_la netincome1_la mtb size _cons -.0128075 -.0191842 -.0173358 0039667 -.0099301 5748406 0224032 -.0023998 0661723 Number of obs Number of groups Obs per group: avg max Wald chi2(8) Prob > chi2 Std Err .0041197 0056264 004154 0019219 0051489 0103135 0005889 0000898 0043003 z -3.11 -3.41 -4.17 2.06 -1.93 55.74 38.05 -26.71 15.39 87 = = = = = = = P>|z| 0.002 0.001 0.000 0.039 0.054 0.000 0.000 0.000 0.000 453 188 2.409574 14755.85 0.0000 [95% Conf Interval] -.020882 -.0302118 -.0254774 0001998 -.0200217 5546264 0212491 -.0025759 057744 -.0047331 -.0081567 -.0091942 0077337 0001615 5950547 0235574 -.0022237 0746006 ... as a proxy measure real earnings manger They also create one more variable to measure earnings manager called real earnings management index to capture the total effect of real manipulation activities. .. management, finding of Graham et al (2005) survey indicate that for manage earnings, managers prefer applying real earning management to accrual-based management Managers prefer REM to accruals... working capital which have been used to manipulate earnings 2.5 Empirical research on real earnings management Even though accrual-based management has been commonly examined as management real earnings