Earnings Management and Overinvestment: Accrual-Based versus Real Activities

Daniel A. Cohen and Paul Zarowin

New York University

November, 2009

Purpose of paper
examine how both real and accrual-based earnings management activities affect firms investment activities

Motivation
dearth of evidence on how EM affects firms real activities Research on the consequences of EM has concentrated largely on announcement and post event stock market returns

Motivation, continued
earnings management may affect resource allocation by causing firms to make suboptimal investment decisions returns studies can only determine whether securities are mispriced, which causes redistribution (between different groups of shareholders), but cannot assess affects on real firm decisions Because they affect the size of the pie, and not just its distribution, real decisions are likely more costly than share price effects.

2 recent papers
Kedia and Philippon, 2008 McNichols and Stubben, 2008 Study real effects of EM

Kedia and Philippon, 2008

SEC mandated restatement sample overinvested (excess capexp) and over hired during the EM period subsequently underinvested and shed employment Their hypothesis: manipulating firms invest and hire excessively to pool with better performing firms, in order to avoid detection 6

McNichols and Stubben 2008

3 groups of firms that overstated earnings investigated by the SEC for accounting sued by their shareholders for accounting restated their financial statements Overinvested (excess capexp) during the misreporting period underinvested during post-event period Their hypothesis: overinvestment is caused by the misleading signals that the misstated information sends to both internal decision makers and external suppliers of capital.

2 recent papers, continued

Both studies focus exclusively on accrualbased earnings management Small, event-based samples Extreme cases of earnings management

Our contribution
1. First study to examine economic consequences of real EM 2. Large sample, not event-based, not extreme earnings management: results can be generalized to wide population of firms

Our findings
Both real and accrual EM firms overinvest in the years up to and including the period of high EM, and then underinvest Indicates that EM is associated with significant real effects Real EM firms overinvest more than accrual EM firms first evidence that real EM has important economic effects
10

Increased interest in real EM

Recent increased interest in EM thru real activities manipulation Gunny 2006 Roychowdhury 2006 Zang 2006

11

Increased interest in real EM, continued

Graham et al. (2005): managers prefer real earnings management activities compared to accrual-based earnings management. Real management activities can be indistinguishable from optimal business decisions. More difficult to detect.
12

Increased interest in real EM, continued

Cohen et al. (2008): managers have shifted away from accrual to real earnings management in the post SOX period. Perhaps because of the need to avoid detection following highly publicized accounting scandals.

13

Related Literature consequences of EM

focus on stock price effects related to EM EM around specific corporate events: IPOs, SEOs, management buyouts, stock repurchases, stock for stock acquisitions how ex-ante EM relates to observed post event abnormal stock returns Rangan (1998) and Teoh et al. for SEOs Teoh, Wong, and Rao (1998) for IPOs
14

Related Literature consequences of EM, contd

short-term capital market reactions around announcements of fraudulent reporting Foster (1979), Dechow, Sloan, and Sweeny (1996), Beneish (1997), and Palmrose, Richardson, and Scholz (2004) market reaction to disclosure of manipulation is on average negative, implying that investors were surprised and interpret these as negative news
15

Related literature: Studies of real earnings management

Graham et al.s survey (2005) - managers prefer real activities manipulation, over accruals manipulation, as a way to manage earnings. Several features of real earnings management: Involve current or future cash flows Cannot be made after the end of the fiscal period Tougher to be challenged by auditors Managers may favor real EM strategies Less regulatory scrutiny Earlier actions to safeguard against future potential accrual shortfalls

16

Real earnings management, contd

Roychowdhury (2006) - firms try to avoid losses 3 ways: (1) boosting sales through accelerating timing and/or generating additional unsustainable sales through increased price discounts or more lenient credit terms (2) overproducing and allocating more overhead to inventory and less to cost of goods sold (3) reducing aggregate discretionary expenses (R&D + advertising+ SG&A) to improve margins.
17

Real earnings management, contd

Zang (2006) - analyzes the tradeoffs between accrual manipulations and real earnings management. shows that real manipulation is positively correlated with the costs of accrual manipulation accrual and real manipulations are negatively correlated Conclusion: managers treat the two strategies as substitutes.
18

Empirical methodology Data and Initial Sample

COMPUSTAT from 1987 to 2006, to use SFAS No. 95 SCF to estimate accruals, (Collins and Hribar, 2002) all nonfinancial firms with available data necessary to calculate the discretionary accruals metrics and real EM proxies for our analysis 8 observations in each 2-digit SIC grouping per year 82,039 firm-year observations

19

SUSPECT Firms
Focus on SUSPECT firms firms likely to have managed earnings firms with annual earnings before extraordinary items (scaled by total assets) between 0 and 0.005 3,831 firm-year observations (Table 2, Panel B - 4.67% of initial sample)

20

Accrual-based Earnings Management Metric

modified cross-sectional Jones model (Jones 1991) as described in Dechow et al. (1995). (1)

21

Accrual model, contd

where, for fiscal year t and firm i, TA represents total accruals: TA it = EBXI it CFO it, EBXI is earnings before extraordinary items and disc. operations (annual Compustat 123) CFO is operating cash flows (from continuing operations) taken from the statement of cash flows (Compustat 308 Compustat 124), Assetit-1 is total assets (Compustat 6), REVit is change in revenues (Compustat 12), and PPEit is the gross value of property, plant and equipment (Compustat 7).
22

Accrual model, contd

cross-sectional model of discretionary accruals for each year we estimate the model for every 2-digit SIC code control for industry-wide changes in economic conditions that affect total accruals allows coefficients to vary across time
23

Accrual model, contd

coefficient estimates from equation (1) are used to estimate the firm-specific normal accruals (NA it) for our sample firms: (2)

discretionary accruals is the difference between total accruals and the fitted normal accruals, defined as: DAit = (TA it / Assetit-1) NAit. Accrual based earnings managers: top 10% of firm-year DA observations each year (8,204 firm-year observations)
24

Real Earnings Management Metrics

Based on Roychowdhury (2006), Zang (2006) and Gunny (2006) 3 metrics as proxies for real earnings management - abnormal levels of: CFO discretionary expenses production costs
25

3 real manipulation methods

1. Accelerate timing of sales by increased price discounts or more lenient credit terms. 2. Reporting of lower cost of goods sold through increased production. 3. Decreased discretionary expenses, such as advertising, R&D, and SG&A expenses.
26

Calculation of 3 metrics
1. sales > 0 and CFO < 0 INCR_SALES&DECR_CFO 2. COGS < 0 and inventory > 0 COGS_CUT&INV>0 3. 0 < EBDISXt < DISXt-1 can show profits by cutting discretionary expenditures below last years amount
27

Table 1- Panel A Sample statistics

comparison of SUSPECT vs nonSUSPECT firms SUSPECT firms are: smaller (in terms of assets, sales, MV) less profitable Invest more (cap and non-cap exp, %TA) greater growth in assets and employees. high growth -consistent with Kedia and Philippon and McNichols and Stubben. 28

Table 1, Panel A:
Comparison of Investment Activities and Earnings Management Strategies among Different Group of Firms: SUSPECT vs. NON-SUSPECT Firms
SUSPECT
MEAN

FIRMS
MEDIAN

NON-SUSPECT
MEAN MEDIAN

DIFF. in
MEAN (t-stat.) MEDIAN (z-stat.)

INVEST

0.231

0.184

0.143

0.094

0.088 (5.67)

0.090 (4.21) 0.075 (3.64) 0.058 (4.32) 0.011 (5.49) 0.015 (2.15) -3.319 (-6.81) -5.522 (-9.37) 88.03 (4.53) -116.68 (-8.17) -28.174 (-4.06) -0.021 (-4.67)

CAPEX

0.364

0.248

0.284

0.173

0.080 (4.37)

NONCAPEX

0.081

0.076

0.046

0.018

0.035 (3.04)

GROW

0.089

0.064

0.064

0.053

0.025 (4.51)

EMPL

0.051

0.042

0.034

0.027

0.017 (2.86)

Total Accruals ($Million) CFO ($Million)

-91.371

-9.643

-72.39

-6.324

-18.981 (-8.37)

85.694

7.742

152.312

13.264

-66.618 (-6.81)

Total Assets

1357.341

264.371

1423.231

176.341

-65.890 (-5.14)

MVE

763.491

86.374

1634.218

203.054

-870.727 (-11.37)

Sales

1432.697

267.153

1543.141

295.327

-110.444 (-7.68)

ROA

0.031

0.043

0.087

0.064

-0.056 (-3.16)

29

Table 1 - Panel B sample distribution

Firms using real earnings management: 2,000 2,600 observations (2.5% - 3.2% of initial sample)

30

Focus: 4 subgroups
intersection of SUSPECT and earnings management firms likely to have managed earnings we can identify the method of earnings management Thus, we can calculate the relation between the method and the extent of over- or under investment.
31

4 subgroups, contd
All 4 subgroups have similar number of observations (0.8% - 1.0% of sample) This gives us confidence that all four subgroups have comparable degrees of earnings management

32

Panel B: Sample Distribution

Number of Firm-Year Observations Overall Sample SUSPECT INCR_SALES&DECR_CFO COGS_CUT&INV>0 0 < EBDISXt < DISXt-1 DA SUSPECT & INCR_SALES&DECR_CFO SUSPECT & COGS_CUT&INV>0 SUSPECT & 0 < EBDISXt < DISXt-1 SUSPECT & DA 796 730 755 0.97% 0.89% 0.92% 82,039 3,831 2,174 2,609 2,043 8,204 673 Percentage (%) 100% 4.67% 2.65% 3.18% 2.49% 10% 0.82%

33

Table 2 - Investment behavior of SUSPECT firms

investment for years t-3 thru t+3 Investment relative to control groups of firms ranked by size and industry Results for total investment and its components, capital expenditures, and non-capital expenditures

34

Table 2, contd
column 1: total investment SUSPECT firms with overinvest during the period of upward management, and then subsequently underinvest Year 0: SUSPECT firms invest 1.9% more than comparable firms, as a % of TA year +1: invest 2.8% less relative investment decline of 4.7% (of TA) Columns 2, 3, and 4 (CAPEXP, non-CAPEXP, Employment) Consistent with Kedia and Philippon and McNichols and Stubben
35

Table 2:
Investments Activities partitioned by Alternative Earnings Management Strategies throughout Time among SUSPECT Firms

Year INVEST -3 -2 -1 0 1 2 3 0.001 0.002 0.014 0.019 -0.028 -0.012 -0.004 CAPEX 0.002 0.004 0.022 0.026 -0.032 -0.011 -0.003

SUSPECT FIRMS NONCAPEX 0.001 0.002 0.019 0.029 -0.026 -0.009 -0.002 GROW 0.002 0.003 0.017 0.013 -0.024 -0.007 -0.003 EMPL 0.001 0.002 0.006 0.005 -0.001 -0.002 0.002

36

Table 3 - Investment behavior for different EM strategies

investment for years t-3 thru t+3 Investment relative to control groups of firms ranked by size and industry Results for total investment and its components, capital expenditures, and non-capital expenditures

37

Table 3, contd
All strategies: overinvestment in the period up to and including EM year; subsequent underinvestment Panel A high DA firms overinvest during the period upto and including the EM year, and then subsequently underinvest Panels B D: same for real EM firms real EM firms have even greater overand underinvestment than high DA 38 firms

Table 3: Investment Activities for Different EM strategies

Year Panel A - Extreme decile of DA INVEST -3 -2 -1 0 1 2 3
Year

Panel B - INCR_SALES&DECR_CFO GROW 0.002 0.001 0.010 0.012 -0.018 -0.008 0.002 EMPL 0.001 0.002 0.003 0.005 -0.001 -0.004 0.001 INVEST 0.001 0.001 0.015 0.029 -0.013 -0.008 0.001 CAPEX 0.001 0.002 0.011 0.023 -0.017 -0.004 0.002 NONCAPEX 0.002 0.001 0.005 0.018 -0.009 -0.001 0.003 GROW 0.002 0.003 0.009 0.017 -0.026 -0.008 -0.002 EMPL 0.001 0.002 0.003 0.005 -0.004 0.003 0.001

CAPEX 0.003 0.002 0.006 0.012 -0.023 -0.015 -0.004

NONCAPEX 0.002 0.001 0.005 0.013 -0.019 -0.014 -0.003

0.002 0.002 0.013 0.015 -0.021 -0.016 -0.003

Panel C - DISX_CUT INVEST CAPEX 0.002 0.003 0.008 0.018 -0.021 -0.016 -0.002 NONCAPEX 0.001 0.002 0.003 0.017 -0.023 -0.013 -0.003 GROW 0.002 0.001 0.004 0.011 -0.014 -0.007 0.002 EMPL 0.002 0.003 0.006 0.009 -0.003 -0.002 0.001

Panel D - COGS_CUT&INV>0 INVEST 0.003 0.011 0.029 0.037 -0.046 -0.014 -0.007 CAPEX 0.002 0.003 0.017 0.035 -0.041 -0.016 -0.002 NONCAPEX 0.001 0.002 0.003 0.027 -0.029 -0.017 -0.002 GROW 0.003 0.005 0.012 0.019 -0.026 -0.011 -0.005 EMPL 0.001 0.002 0.004 0.003 -0.002 -0.003 0.002

-3 -2 -1 0 1 2 3

0.001 0.002 0.017 0.024 -0.029 -0.012 -0.008

39

Table 4 - Control for investment opportunities

investment is a function of investment opportunities and liquidity Omitted factors may be correlated with measures of earnings management relation between investment and EM may be due to omitted factors i.e., excess investment might really be optimal given the firms opportunity set
40

 Insert equation (10) here Model based on Biddle, Hilary, and Verdi (2008) Augmented by SUSPECT, EM, and SUSPECT*EM

Investment as a function of fundamentals and earnings management

41

Investment model
dependent variable: INVEST, CAPEX, or NONCAPEX, scaled by total assets (TA) LOG_ASSET is the log of total assets MKT_BK is MV of equity/BV of TA LEVERAGE is long term debt/MV of equity SLACK is cash/PPE AGE is current year - first year on CRSP OP_CYCLE is ln(AR/sales+inv/CGS)* 360 LOSS = 1 if NI before extra.items is negative and zero otherwise; TANGIBLE is PPE/TA DIVIDEND = 1 if the firm paid dividends and zero otherwise. 42

Table 4, Panel A Results of equation (10)

Overall, the coefficients on the fundamentals are consistent with expectations, and equation (10) explains about 25% of the cross-firm variation in investment.

43

Focus on SUSPECT*EM
focus on firms most likely to have managed earnings, but by different means coefficients on SUSPECT*EM capture the overinvestment for suspect firms differentiated by strategy enable us to compare the effects of different earnings management strategies on excess investment.
44

Results for SUSPECT*EM

Table 4, Panel A coefficients on all of the interaction terms are significantly positive shows that firms managing earnings by any of the strategies overinvest the coefficients on the real earnings management interactions are two to four times as great as the coefficient on the accrual interaction
45

Table 4, Panel A, contd

Firms increasing production to cut COGS have the greatest overinvestment: 14.2% of TA followed by firms cutting discretionary expenditures: 8.1% High DA firms - overinvestment of 3.1%

46

Table 4:
Relation between Investments and Alternative Earnings Management Strategies Panel A: Dependent variable is INVEST
EM = DA

EM = DISX_CUT

EM = COGS_CUT &INV>0

EM = INCR_SALES &DECR_CFO
0.085 (3.71) 2.498 (14.23) -5.617 (-9.61) -0.058 (-0.92) -0.071 (-7.12) -0.857 (-2.95) -3.639 (-7.63) 11.403 (9.23) -0.439 (-3.83) 0.064 (3.54) 0.027 (4.38) 0.063 (4.12) 0.253

LOG_ASSET MKT-BK LEVERAGE SLACK AGE OP_CYCLE LOSS TANGIBLE DIVIDEND SUSPECT EM SUSPECT*EM Adj. R2

0.083 (1.39) 2.408 (11.17) -5.638 (-13.07) -0.053 (-0.69) -0.071 (-8.26) -0.867 (-3.51) -3.613 (-12.18) 11.183 (12.79) -0.431 (-3.26) 0.065 (3.72) 0.019 (4.05) 0.031 (2.71) 0.263

0.082 (1.69) 2.506 (12.08) -5.572 (-12.35) -0.049 (-0.63) -0.074 (-7.05) -0.853 (-3.61) -3.704 (-11.94) 11.374 (14.06) -0.442 (-3.26) 0.062 (4.01) 0.025 (4.16) 0.081 (3.04) 0.243

0.078 (3.15) 2.574 (13.57) -5.664 (-8.17) -0.052 (-1.09) -0.069 (-5.67) -0.842 (-4.99) -3.657 (-8.64) 11.327 (9.84) -0.433 (-3.57) 0.064 (3.72) 0.038 (5.06) 0.142 (3.37) 0.284

47

Table 4, Panels B and C

Results due mainly to CAPEXP (Panel B) Little variation in NONCAPEXP (Panel C)

48

Panel B: Dependent variable is CAPEX

EM = DA

EM = DISX_CUT

EM = COGS_CUT &INV>0

EM = INCR_SALES &DECR_CFO
0.096 (3.26) 2.746 (9.34) -4.494 (-8.75) -0.036 (-0.79) -0.055 (-5.36) -0.723 (-3.16) -3.113 (-9.49) 10.647 (11.46) -0.431 (-3.98) 0.043 (4.18) 0.025 (3.81) 0.051 (3.45) 0.288

LOG_ASSET MKT-BK LEVERAGE SLACK AGE OP_CYCLE LOSS TANGIBLE DIVIDEND SUSPECT EM SUSPECT*EM Adj. R2

0.091 (2.15) 2.854 (8.16) -4.643 (-10.27) -0.033 (-0.57) -0.061 (-5.78) -0.741 (-4.78) -3.126 (-10.81) 10.718 (11.76) -0.327 (-3.57) 0.046 (3.23) 0.014 (2.84) 0.024 (2.98) 0.271

0.093 (2.14) 2.813 (7.34) -4.439 (-10.16) -0.035 (-0.72) -0.059 (-6.02) -0.709 (-3.94) -3.137 (-10.68) 10.824 (12.67) -0.329 (-3.56) 0.045 (3.51) 0.019 (2.98) 0.074 (3.26) 0.273

0.088 (2.96) 2.896 (10.67) -4.536 (-8.61) -0.031 (-0.67) -0.064 (-4.98) -0.712 (-4.11) -2.998 (-9.36) 10.735 (10.32) -0.332 (-3.69) 0.047 (3.96) 0.036 (4.67) 0.086 (2.83) 0.314

49

Panel C: Dependent variable is NONCAPEX

EM = DA

EM = DISX_CUT

EM = COGS_CUT &INV>0

EM = INCR_SALES &DECR_CFO
0.021 (-3.34) 1.627 (3.92) -2.334 (-3.89) -0.011 (-0.43) -0.063 (-4.68) 0.603 (3.28) -2.498 (-7.82) -9.351 (-6.81) -0.214 (-2.67) 0.013 (1.46) 0.015 (2.03) 0.007 (1.89) 0.259

LOG_ASSET MKT-BK LEVERAGE SLACK AGE OP_CYCLE LOSS TANGIBLE DIVIDEND SUSPECT EM SUSPECT*EM Adj. R2

0.026 (1.32) 1.554 (4.43) -2.402 (-3.76) -0.015 (-0.61) -0.066 (-5.23) 0.616 (3.16) -2.478 (-7.84) -9.286 (-7.14) -0.215 (-2.69) 0.008 (2.27) 0.014 (2.39) 0.008 (1.48) 0.248

0.024 (0.94) 1.539 (3.69) -2.417 (-4.10) -0.013 (-0.13) -0.067 (-5.16) 0.591 (3.29) -2.467 (-9.39) -9.258 (-6.34) -0.218 (-2.71) 0.011 (1.42) 0.018 (1.83) 0.004 (0.83) 0.217

0.018 (2.73) 1.631 (4.58) -2.411 (-4.56) -0.012 (-0.22) -0.064 (-4.36) 0.584 (4.32) -2.513 (-8.94) -9.239 (-7.03) -0.212 (-3.07) 0.009 (1.27) 0.024 (2.14) 0.006 (1.09) 0.221

50

Table 4, Panel D
Difference between coefficients on DA vs real earnings management proxies are highly statistically significant Summary: firms engaging in real earnings management overinvest more than firms engaging in accrual earnings management

51

Panel D: Investments (INVEST) and Different Earnings Management Strategies

EM = &DISX_CUT

EM = CUTS&INV>0

EM = INCR_SALES& DECR_CFO

SUSPECT DA EM

0.064 (3.81) 0.018 (4.27)

SUSPECT*DA
SUSPECT*EM

0.026 (4.59) 0.030 (2.86)

0.083 (3.72)

0.063 (3.86) 0.017 (4.12) 0.039 (5.27) 0.029 (3.01) 0.146 (3.64)

0.064 (3.79) 0.019 (4.22) 0.028 (4.52) 0.028 (3.16) 0.067 (4.96)

F-Tests for DA = EM and SUSPECT*DA = SUSPECT*EM

DA = EM SUSPECT*DA = SUSPECT*EM

12.37 16.92

13.21 27.48

11.98 17.35
52

Intertemporal relation between EM and investment

shows whether earnings management precedes the excess investment McNichols and Stubben: if earnings management causes overinvestment, then the overinvestment should be related to current and past earnings management estimate investment model (eqn (10)), augmented with measures of firms earnings management from t-2 thru t+1 53 do not include SUSPECT, which only

Table 5
current investments significantly related to both accrual-based and real current, past, and future earnings management coefficients on EMt, EMt-1, and EMt+1 > 0 relationship between investments and earnings management activities is the strongest in the concurrent year

54

Table 5, contd
relation between investments and EM in year t+1 is consistent with some portion of earnings management to offset poor returns from over investments made in past periods. effects of real EM on investment are greater than the effects of accrual EM earnings management additional evidence that firms that engage 55 in real EM experience greater

Table 5:
Investments and Earnings Management Strategies over Time

Panel A: Dependent variable is INVEST

EM = DA

EM = DISX_CUT

EM = COGS_CUT &INV>0

EM = INCR_SALES &DECR_CFO
0.079 (3.12) 2.487 (13.41) -5.628 (-9.17) -0.053 (-0.81) -0.073 (-7.38) -0.848 (-2.83) -3.645 (-7.86) 11.368 (9.46) -0.447 (-3.46) 0.003 (0.57) 0.014 (2.01) 0.027 (4.38) 0.014 (1.98) 0.272

LOG_ASSET MKT-BK LEVERAGE SLACK AGE OP_CYCLE LOSS TANGIBLE DIVIDEND EM t-2 EM t-1 EM t EMt+1 Adj. R2

0.083 (2.45) 2.473 (9.12) -5.401 (-10.24) -0.057 (-0.83) -0.076 (-5.29) -0.857 (-3.41) -3.504 (-9.79) 11.591 (11.34) -0.429 (-3.68) 0.005 (0.96) 0.009 (1.56) 0.018 (4.33) 0.009 (2.23) 0.262

0.078 (1.74) 2.541 (12.25) -5.613 (-11.69) -0.0451 (-0.68) -0.077 (-6.08) -0.862 (-3.43) -3.721 (-10.38) 11.238 (12.34) -0.448 (-3.81) 0.004 (0.68) 0.011 (1.74) 0.025 (4.16) 0.013 (2.19) 0.258

0.071 (3.29) 2.532 (13.07) -5.598 (-8.39) -0.049 (-1.02) -0.068 (-5.41) -0.859 (-4.84) -3.662 (-8.73) 11.289 (9.17) -0.441 (-3.12) 0.005 (0.84) 0.019 (1.87) 0.038 (5.06) 0.016 (2.31) 0.296

56

Table 5:
Investments and Earnings Management Strategies over Time

Panel B: Dependent variable is CAPEX

EM = DA

EM = DISX_CUT

EM = COGS_CUT &INV>0

EM = INCR_SALES &DECR_CFO
0.093 (3.01) 2.863 (9.67) -4.486 (-8.28) -0.034 (-0.83) -0.056 (-5.09) -0.728 (-3.21) -3.116 (-9.63) 10.656 (11.14) -0.329 (-3.37) 0.004 (0.72) 0.015 (2.24) 0.025 (3.81) 0.012 (1.88) 0.316

LOG_ASSET MKT-BK LEVERAGE SLACK AGE OP_CYCLE LOSS TANGIBLE DIVIDEND EM t-2 EMt-1 EM t EM t+1 Adj. R2

0.094 (2.58) 2.772 (8.28) -4.541 (-8.79) -0.037 (-0.93) -0.059 (-4.52) -0.719 (-4.16) -3.138 (-9.89) 10.674 (11.78) -0.34 (-3.12) 0.004 (0.68) 0.011 (2.89) 0.015 (4.39) 0.008 (2.79) 0.278

0.095 (2.37) 2.834 (7.12) -4.583 (-10.35) -0.034 (-0.65) -0.054 (-6.36) -0.721 (-3.34) -3.156 (-10.39) 10.786 (11.37) -0.326 (-3.42) 0.005 (0.72) 0.012 (1.86) 0.019 (2.98) 0.012 (2.58) 0.291

0.091 (2.84) 2.858 (10.29) -4.561 (-8.14) -0.032 (-0.62) -0.058 (-4.46) -0.719 (-4.38) -3.104 (-10.67) 10.763 (10.41) -0.334 (-3.76) 0.004 (0.42) 0.018 (1.93) 0.036 (4.67) 0.013 (2.26) 0.334

57

Table 5:
Investments and Earnings Management Strategies over Time

Panel C: Dependent variable is NONCAPEX

EM = DA

EM = DISX_CUT

EM = COGS_CUT &INV>0

EM = INCR_SALES &DECR_CFO
0.027 (3.17) 1.629 (3.74) -2.329 (-3.41) -0.013 (-0.38) -0.065 (-4.16) 0.611 (3.47) -2.493 (-7.54) -9.357 (-6.43) -0.218 (-2.76) 0.002 (0.43) 0.008 (1.53) 0.015 (2.03) 0.008 (1.57) 0.277

LOG_ASSET MKT-BK LEVERAGE SLACK AGE OP_CYCLE LOSS TANGIBLE DIVIDEND EM t-2 EM t-1 EM t EMt+1 Adj. R2

0.026 (2.75) 1.606 (4.81) -2.452 (-4.61) -0.013 (-0.63) -0.064 (-4.38) 0.614 (3.97) -2.497 (-7.56) -9.2401 (-6.42) -0.209 (-3.44) 0.003 (0.61) 0.007 (1.64) 0.009 (1.39) 0.004 (1.59) 0.253

0.026 (0.83) 1.536 (3.42) -2.412 (-4.16) -0.012 (-0.19) -0.061 (-5.25 0.587 (3.42) -2.456 (-8.61) -9.267 (-6.63) -0.215 (-2.74) 0.004 (0.51) 0.008 (1.26) 0.018 (1.83) 0.008 (1.42) 0.223

0.022 (2.36) 1.643 (4.15) -2.414 (-4.61) -0.011 (-0.26) -0.063 (-4.32) 0.586 (4.63) -2.521 (-8.14) -9.243 (-7.11) -0.216 (-3.23) 0.003 (0.21) 0.011 (1.78) 0.024 (2.14) 0.012 (1.98) 0.246

58

Intertemporal evidencesummary
evidence implies that EM leads or occurs contemporaneously with over investments smaller statistical significant association between investments and EM activities in the subsequent year is consistent with some portion of earnings management to offset poor returns from over investments made in past periods.
59

summary
first paper to examine the economic effects of real EM real EM firms overinvest and subsequently underinvest in the years surrounding the earnings management. excess investment associated with real EM is greater than excess investment associated with DA

60