Slides - Old Dominion University

advertisement
Is the Outcome of a Securities
Class Action a Reliable Signal of
Accounting Irregularity?
Nana Y. Amoah
Old Dominion University
Alex P. Tang
Morgan State University
Feb. 2011
Motivations
• Dismissal of several class actions raises questions
about the appropriateness of securities litigation
as a signal of accounting irregularity.
• 65.9% (1,396) of 2119 class actions over the
period 1996-2005 were dismissed (Simmons and
Ryan, 2007).
• Fich and Shivdasani (2007) note that it cannot be
confirmed that fraud has occurred as a result of a
securities lawsuit.
• Effect of PSLRA on securities lawsuit outcome.
Motivations
• Proponents of litigation reform argue that only
settled lawsuits are a credible signal of accounting
irregularity.
• Critics argue that the litigation process is widely
abused and settlements are excessive and do not
reflect the wrongdoing.
• No clear evidence of fraud in settlements.
• Debate on effectiveness of securities litigation as
a governance mechanism (Rogers and Buskirk,
2009).
Figure 1: Timeline of Filing and Resolution of
Securities Lawsuit
Restating Firm
Announces
Irregularity
Or Error
(Irreg=1/0)
Defendants File Motion
to Dismiss the Lawsuit
(Dismiss=1/0)
Securities Lawsuit is
Settled
(Settlement=S)
TIME
Restatement Triggers
Securities Lawsuit
(Litig = 1/0)
Class Certification
and Discovery
Accounting Irregularity and Litigation
Outcome (Related Research & Hypotheses)
• Negative association between fraud and restatement
announcement returns (Palmrose, Richardson and
Scholz, 2004; Lev, Ryan and Wu, 2007).
• Restatements due to accounting irregularity are
more likely to trigger securities lawsuits.
• Positive association between accounting irregularity
and restatement-induced securities lawsuit (Hennes,
Leone and Miller, 2008; Amoah and Tang, 2010).
• Accounting irregularity is defined similarly to
Hennes, Leone and Miller (2008).
Accounting Irregularity and Litigation
Outcome (Related Research & Hypotheses)
• Under the PSLRA, meritorious securities lawsuits
are less likely to be dismissed (Choi and Thompson,
2006).
• Restatement-induced lawsuit centered on
accounting irregularity provides an inference of
scienter and should be more likely to be settled.
• Accounting irregularity is associated with greater
shareholder losses, thus, settlement amount should
be positively associated with accounting
irregularity.
Hypotheses
(Accounting Irregularity and Litigation Outcome)
• H1: Ceteris paribus, settled restatement-induced
lawsuits are positively associated with accounting
irregularity.
• H2: Ceteris paribus, the settlement amount of a
restatement-induced lawsuit is positively associated
with accounting irregularity.
CEO Turnover and Litigation Outcome
(Related Research & Hypotheses)
• Likelihood of CEO turnover increases with securities
litigation (Niehaus and Roth,1999; Collins, et. al,
2008).
• SEC and DOJ enforcement action positively
associated with termination of executive officers
implicated in the fraud (Karpoff, Lee and Martin,
2008) .
• CEOs have access to material non-public adverse
information and should be culpable when financial
statements are false and misleading.
Hypothesis
(CEO Turnover and Litigation Outcome)
• Thus, when restating firms settle securities
lawsuits, it is likely that CEOs are culpable in the
alleged fraud.
• H3: Ceteris paribus, settled restatement-induced
securities lawsuit is positively associated with
CEO turnover.
Sample Selection Procedure
Description
Restatements from 1997-2005 in GAO (2002)
and GAO (2007) databases
Number
2309
Deleting repeated restatements
(322)
Missing Compustat and CRSP data
(860)
Technical restatements, income increasing
restatements and missing restatement and
litigation data
(432)
Restating sample
695
Litigation firms
185
Distribution of Restatements by fiscal year
Year
1997 1998 1999 2000 2001 2002 2003 2004 2005 Total
Restating firms
31
26
35
43
36
93
96
130
205
695
Litigation firms
18
19
23
22
20
42
15
16
10
185
Settlement firms
Dismissed
Lawsuits (%)
Settled Lawsuits
(%)
14
14
17
17
14
31
11
8
7
133
22% 26% 26% 23% 30% 26%
27%
50%
30%
28%
78% 74% 74% 77% 70% 74%
73%
50%
70%
72%
Distribution of Litigations by fiscal year
Year
1997 1998 1999 2000 2001 2002 2003
2004
2005 Total
Litigation firms
15
20
22
24
19
40
15
16
14
185
Settlement firms
13
14
16
17
14
31
13
8
7
133
Descriptive Statistics of Continuous Variables (Restating Firms)
Variable
R(-1,1)
Mean
-0.067
Median
-0.027
Std. Dev.
0.155
t-stat
(p-value)
<0.001
Wilcoxon
(p-value)
<0.001
Lev
0.241
0.204
0.230
<0.001
<0.001
LnA
6.289
6.165
2.157
<0.001
<0.001
ROA
0.025
0.056
0.179
<0.001
<0.001
Frequency of Binary Variables (Restating Firms)
Variable
Number of firms
Percentage
Irregularity
177
26%
Revenue
213
31%
Litig
185
27%
Dismiss
52
8%
Settle
133
19%
Chi-Sq
(p-value)
167.31
(<0.001)
104.12
(<0.001)
151.98
(<0.001)
502.56
(<0.001)
264.81
(<0.001)
Univariate Results - Irregularity and No-Irregularity Restating Subsamples
Binary Variables
Variable
Revenue
Pred.
Sign
+
Litig
+
Dismiss
?
Settle
+
Irregularity= 1
Observations (N=177)
(percentage)
92
(51.98%)
88
(49.72%)
13
(7.34%)
75
(42.37%)
No-Irregularity= 0
Observations (N=518)
(percentage)
121
(23.36%)
97
(18.73%)
39
7.53%)
58
(11.20%)
Chi-Square
(p-value)
50.83a
(0.008)
64.87a
(<0.001)
0.01
(0.936)
82.86a
(<0.001)
Continuous Variables
Pred.
Sign
-
Irregularity= 1
(N=177)
Mean
-0.103
No-Irregularity= 0
(N=518)
Mean
-0.055
Lev
+/-
0.249
0.239
LnA
+/-
6.811
6.086
+/-
0.040
0.020
Variable
R(-1,1)
ROA
Difference
in Mean
(p-value)
-0.048a
(0.001)
0.010
(0.638)
0.725a
(<0.001)
0.020
(0.147)
Descriptive Statistics of Continuous Variables (Litigation Firms)
Variable
R(-1,1)
Mean
-0.197
Median
-0.154
Std. Dev.
0.204
t-stat
p-value
<0.001
Wilcoxon
p-value
<0.001
LnS
16.864
16.730
1.673
<0.001
<0.001
Lev
0.266
0.257
0.207
<0.001
<0.001
LnA
7.622
7.437
2.020
<0.001
<0.001
ROA
0.075
0.082
0.097
<0.001
<0.001
Frequency of Binary Variables (Litigation Firms)
Variable
Number of firms
Percentage
Irregularity
88
48%
Revenue
103
56%
Dismiss
52
28%
Settle
133
72%
Chi-Sq
(p-value)
0.27
(0.605)
2.89
(0.089)
35.85
(<0.001)
35.85
(<0.001)
Univariate Results - Irregularity and No-Irregularity Litigation Subsamples
Binary Variables
Variable
Revenue
Pred.
Sign
+
Dismiss
-
Settle
+
Irregularity= 1
Observations (N=88)
(percentage)
54
(61.36%)
13
(14.77%)
75
(85.23%)
No-Irregularity= 0
Observations (N=97)
(percentage)
49
(51.58%)
38
(40.0%)
57
(60.0%)
Chi-Square
(p-value)
1.78
(0.182)
14.46
(0.936)
14.46a
(<0.001)
Continuous Variables
-
Irregularity= 1
(N=88)
Mean
-0.185
No-Irregularity= 0
(N=97)
Mean
-0.207
Lev
+/-
0.251
0.279
LnA
+/-
7.666
7.582
ROA
+/-
0.079
0.072
LnS
+
17.098
16.458
Variable
R(-1,1)
Pred.
Sign
Difference in
Mean
(p-value)
0.022
(0.465)
-0.028
(0.374)
0.084
(0.780)
0.392
(0.599)
0.640
(0.111)
Analysis of Accounting Irregularity
IRREG   0  1 R( 1,1)   2 Re v   3 Litig   4 Dismiss   5 LnS   6Settle  7 ROA   8 Lev
  9 LnA....(1)
Logistic Regressions of Accounting Irregularity
(Restating Sample)
Variable
Logit
Coeff.
Effect
Size
Logit
Coeff.
Effect
Size
Intercept
(?)
-1.852a
(37.01)
-
-1.873a
(36.98)
-
0.177
(0.08)
19.3%
R(-1,1) (+)
Model
Logit
Coeff.
-1.854a
(36.23)
Effect
Size
Logit
Coeff.
Effect
Size
Logit
Coeff.
Effect
Size
-
-1.866a
(35.77)
-
-2.278a
(46.38)
-
0.512
(0.62)
66.9%
0.595
(0.80)
1.000a
(23.81)
81.3%
Revenue (+)
Litig (+)
1.330a 278.2%
(42.18)
1.327a
(31.36)
276.9%
Dismiss (?)
Settle (+)
ROA (+/-)
Lev (+/-)
LnA (+/-)
ModChiSq
(P-value)
H-L ChiSq
(P-value)
-0.182
(0.10)
-0.132
(0.10)
0.058
(1.36)
57.49a
(<0.01)
4.25
(0.83)
-16.7%
-12.3%
5.9%
-0.305
(0.28)
-0.031
(0.01)
0.060
(1.44)
54.00a
(<0.01)
9.25
(0.32)
171.9%
-26.3%
-3.1%
6.1%
0.357
(1.03)
1.675a
(55.28)
-0.128
(0.05)
-0.143
0.058
(1.37)
71.84a
(<0.01)
5.18
(0.74)
42.9%
433.9%
-12.0%
-13.4%
6.0%
0.334
(0.81)
1.758a
(44.33)
-0.257
(0.19)
-0.037
(0.01)
0.060
(1.40)
69.83a
(<0.01)
12.42
(0.13)
39.7%
480%
-22.6%
-3.7%
6.1%
0.030
(0.01)
1.381a
(24.75)
-0.055
(0.01)
-0.003
(0.00)
0.083
(2.58)
93.23a
(<0.01)
12.62
(0.13)
3.1%
297.7%
-5.3%
-0.3%
8.6%
Logistic Regressions of Accounting Irregularity
(Litigation Sample)
Model
Variable
Logit
Coeff.
Effect
Size
Logit
Coeff.
Effect
Size
Logit
Coeff.
Effect
Size
Logit
Coeff.
Effect
Size
Logit
Coeff.
Effect
Size
Intercept (?)
-1.692b
(5.05)
-
-1.656b
(5.10)
-
-1.392b
(3.85)
-
-1.748b
(5.28)
-
-3.941c
(3.74)
-
R(-1,1) (+)
1.022
(1.64)
178.0%
1.022
(1.65)
177.7%
1.048
(1.71)
185.3%
1.288
(2.15)
262.7%
Revenue (+)
0.427
(1.66)
53.2%
0.441
(1.76)
55.4%
0.213
(0.28)
23.7%
0.278b
(4.88)
32.1%
-1.277
(1.56)
-72.1%
0.343
(1.11)
40.9%
LnS (+)
Settle (+)
1.412a
(13.35)
310.6%
ROA (+/-)
1.292a
(12.00)
1.124
(0.47)
-0.913
(1.18)
264.1%
0.073
(0.72)
7.6%
207.6%
-59.9%
1.481a
(14.70)
0.841
(0.25)
-0.720
(0.70)
339.9%
0.064
(0.55)
6.6%
131.9%
-51.3%
1.430a
(13.52)
0.994
(0.35)
-0.676
(0.60)
317.9%
0.080
(0.84)
8.3%
170.2%
Lev (+/-)
-0.716
(0.68)
-51.1%
-49.1%
LnA (+/-)
0.086
(0.99)
9.0%
Model
ChiSq
(P-value)
H-L ChiSq
(P-value)
19.56a
(<0.01)
11.71
17.05a
(<0.01)
6.61
18.14a
(<0.01)
4.12
19.90a
(<0.01)
3.91
8.36c
(0.079)
4.05
(0.17)
(0.58)
(0.85)
(0.87)
(0.85)
Analysis of CEO Turnover
CEO1   0  1 R( 1,1)   2 Re v   3 Litig   4 Dismiss   5Settle  6 ROA   7 Lev
  8 LnA....(2i)
CEO2   0  1 R( 1,1)   2 Re v   3 Litig   4 Dism iss   5Settle  6 ROA   7 Lev
  8 LnA....(2ii)
CEO3   0  1 R( 1,1)   2 Re v   3 Litig   4 Dismiss   5Settle  6 ROA   7 Lev
  8 LnA....(2iii)
Logistic Regressions of CEO Turnover
CEO1
Logit
Coeff.
CEO1
Effect
Size
CEO1
Logit
Coeff.
CEO1
Effect
Size
CEO2
Logit
Coeff.
CEO2
Effect
Size
CEO3
Logit
Coeff.
CEO3
Effect
Size
-3.104a
(12.90)
-
-2.574a
(8.39)
-
-1.947a
(7.28)
-
-1.541b
(4.37)
-
R(-1,1) (+)
0.965
(0.95)
162.5%
1.298
(1.63)
266.2%
-0.162
(0.04)
-14.9%
0.561
(0.45)
75.2%
Revenue (+)
0.450
(1.44)
56.9%
0.243
(0.38)
27.5%
0.454
(2.05)
57.4%
0.297
(0.87)
34.6%
Litig (+)
1.642a
(12.22)
416.5%
0.851
(1.87)
2.173a
(18.25)
-3.283
(2.40)
0.061
(0.00)
0.011
(0.01)
134.2%
0.155
(0.11)
1.206a
(9.03)
-2.600
(2.18)
0.324
(0.19)
0.053
(0.34)
16.8%
1.039b
(5.42)
1.844a
(19.58)
0.612
(0.12)
0.189
(0.06)
-0.054
(0.33)
182.6%
Variable
Intercept (?)
Dismiss (?)
Settle (+)
ROA (+)
Lev (+)
LnA (+/-)
0.158
(0.03)
0.041
(0.14)
17.1%
4.2%
778.3%
-96.2%
6.3%
1.1%
234.1%
-92.6%
38.2%
5.4%
Model ChiSq
(P-value)
H-L ChiSq
23.38a
(<0.01)
3.34
33.35a
(<0.01)
10.17
29.68a
(<0.01)
6.59
35.22a
(<0.01)
3.83
(P-value)
(0.91)
(0.25)
(0.58)
(0.87)
532.4%
84.4%
20.8%
-5.2%
Logistic Regressions of SEC Enforcement Action
Variable
Intercept
(?)
R(-1,1) (+)
Revenue
(+)
Litig (+)
Logit Effect
Coeff.
Size
a
-3.181
(39.27)
0.725 106.5%
(0.64)
0.630b 87.7%
(4.32)
Logit
Coeff.
-2.869a
(34.79)
0.953
(1.09)
Effect
Size
-
Model
Logit
Coeff.
-3.114a
(38.97)
Effect
Size
-
159.5%
0.512c
(2.82)
Settle (+)
LnA (+/-)
ModChiSq
(P-value)
H-L ChiSq
(P-value)
15.4%
Logit
Coeff.
-12.186a
(13.47)
0.190
(0.03)
0.233
(0.18)
Effect
Size
21.0%
26.2%
1.284a 261.0%
(11.81)
Dismiss (?)
Lev (+/-)
Effect
Size
-
66.9%
LnS (+)
ROA (+/-)
Logit
Coeff.
-12.191
(13.46)
0.143
(0.02)
0.147 15.8%
(0.02)
-0.001 -0.00%
(0.00)
0.059
6.0%
(0.60)
35.28a
(<0.01)
6.61
(0.58)
0.836
(2.63)
1.875a
(24.47)
0.162
(0.02)
-0.002
(0.00)
0.031
(0.17)
36.06a
(<0.01)
6.52
(0.59)
130.6%
552.3%
17.6%
-0.2%
3.2%
0.706
(2.11)
1.391a
(15.16)
0.295
(0.07)
-0.00
(0.00)
0.050
(0.44)
37.32a
(<0.01)
5.21
(0.73)
0.590a
(8.11)
80.4%
0.575a
(7.54)
77.8%
-1.786
(0.32)
-2.549c
(2.97)
0.212
(1.67)
18.96a
(<0.01)
10.58
(0.23)
-83.2%
-1.701
(0.30)
-2.487c
(2.81)
0.222
(1.77)
19.13a
(<0.01)
6.30
(0.61)
-81.7%
102.5%
301.9%
34.3%
-0.2%
5.1%
-92.2%
23.7%
-91.7%
24.9%
Results
• Settled lawsuit positively associated with accounting
irregularity consistent with our first hypothesis (H1).
• Positive association between settled lawsuit and CEO
turnover (Supports H3).
• Weak evidence of a positive association between
accounting irregularity and settlement amount.
• Results for settled lawsuit are robust with SEC
enforcement action as the fraud measure.
• Strong evidence of a positive association between
settlement amount and SEC enforcement action.
Conclusions
• Settled lawsuit is an appropriate signal of
accounting irregularity.
• CEOs of settlement firms are implicated in the
alleged fraud that triggered the lawsuit.
• Findings are contrary to Beneish (1999) who find
that SEC enforcement action is not effective in
discouraging management fraud.
Contributions
• First study that examines the relation between
accounting irregularity and securities litigation
outcome.
• Highlights the importance of distinguishing
between dismissed and settled lawsuits in litigation
studies.
• Contributes to the debate on whether securities
litigation is an effective governance mechanism.
• Extends studies linking restatement and
restatement-induced securities lawsuit to CEO
turnover (such as Desai, Hogan, and Wilkins, 2006
& Collins, Reitenga and Sanchez, 2008).
Download