The Effect of Managerial “Style” on the
Tone of Earnings Conference Calls
Angela Davis
Weili Ge
Dawn Matsumoto
Jenny Li Zhang
April 27, 2011
Research Questions
 Does a manager’s “style” impact the tone
expressed in conference calls?
• What is “style”?
• What is “tone”?
 Does the market react to the portion of tone that is
manager-specific?
2
We are a company known for being
conservative.
—Greg Maffei, Microsoft CFO 1997-2000
3
We are a company known for being
conservative.
—Greg Maffei, Microsoft CFO 1997-2000
I believe in being disciplined but aggressive.
—Chris Liddell,
Microsoft CFO 2005-2009
4
Motivation
 Recent interest in the use of language to convey (or
obscure) information to the capital markets
• Readability of annual report disclosures (Li 2008)
• Deceptive language in conference calls (Larcker and
Zakolyukina 2010)
• “Tone” in corporate disclosures (Davis et al. 2010; Demers and
Vega 2010; Frankel et al. 2009; Price et al. 2010)
 Overall conclusion: tone conveys information beyond
concurrent, quantifiable information
• Market reaction to tone, controlling for current performance
5
Motivation
 BUT: other factors likely influence tone
• Incentives to bias (Lang and Lundholm 2000)
• Unintentional bias
 Recent studies have shown manager-specific effects in
financial reporting choices:
• Dyreng et al. 2010 -- Tax avoidance behavior
• Bamber et al. 2010 & Yang 2010 – Disclosure behavior
• Ge et al. 2011 – Accounting choices
6
Contribution
 Evidence that manager-specific factors influence tone
• Separate from current performance, future performance, firm
and quarter effects
• Manager effects are large relative to other contexts
• Suggests tone is more than just a function of economic events
 Some evidence that the market reacts to the managerspecific component of tone for optimistic managers but
not pessimistic managers
• Consistent with the market identifying and discounting
manager-specific pessimism but not manager-specific
optimism
7
What is “style”?
Systematic choice made by a manager across situations
• “Upper echelons” theory (Hambrick and Mason 1984)
• Contrary to “neoclassical” view of the firm
Operationalization: Manager fixed effects controlling
for firm and time effects
Determinants of style
• Prior experiences
• Personality/disposition
8
What is “tone”?
 Optimism/pessimism expressed in corporate
disclosures
 Operationalization: Counts of words deemed
positive/negative
 Determinants of tone
• Positive/negative economic events (content)
• Manager choice of how to describe these events (language)
9
Does manager style impact tone?
 Style impacts choices more when manager discretion is
higher
• Hambrick 2007
• Ge et al. 2011
 Choice of language relatively unconstrained
• Not subject to GAAP, audits, SEC regulation
• Particularly in conference calls
10
Does manager style impact tone?
But style impacts choices more when optimal decision
is ambiguous (bounded rationality)
• Systematic over/under optimism can be costly
• Optimistic language increases probability of class action
lawsuits (Rogers et al. 2010)
 But bias may be unintentional
•
Dispositional optimism
 Prediction: Style impacts tone
11
Does the market react to style?
 Some evidence the market differentially prices tone
• Cross-sectional variation in “tone response coefficient” based
on firm-specific credibility measures (Demers and Vega 2010)
 Some evidence the market recognizes style
• Greater market reaction to forecasts of high forecast accuracy
managers (Yang 2010)
12
Does the market react to style?
 Difficult to identify “unwarranted” optimism ex ante
• Pricing of discretionary accruals (Xie 2001); pro forma
earnings (Doyle et al. 2003)
• Increased shareholder litigation consistent with markets
being misled (Rogers et al. 2010)
 Experimental evidence investors react to language
• Vivid vs. pallid language (Hales et al. 2011)
 Identifying “language style” more difficult than
“accuracy of forecast style”
13
Measures of Tone
 Frequency counts of positive vs. negative words
• Separate presentation from Q&A
• Use only comments in Q&A spoken by specific manager
 Three dictionaries used:
• TONE_D: Diction (Davis et al. 2010)
• TONE_H: Henry (2006, 2008); Henry and Leone (2009)
• TONE_LM: Loughran and McDonald (2009)
 TONE_i = (positive words – negative words) ÷ total words
14
Examples of Differences between Wordlists
Diction
Henry
L&M
Growth
Yes
Yes
Yes
Pleased
Yes
Yes
Yes
Proud
Yes
No
No
Thrilled
Yes
No
No
Bad
Yes
No
Yes
Excited
Yes
No
Yes
Great
Yes
No
Yes
Harsh
Yes
No
Yes
Achieve, Achieving
No
Yes
Yes
Opportunities
No
Yes
Yes
Exceeding
No
Yes
No
15
Excerpts from transcripts:
 “We’re excited about the accelerator, but we’re even more excited about
by Flash.” (John East, CEO, Actel 7/23/02)
 “During the call today, I’m going to focus my comments on the excitement,
the opportunities, the optimism and the commitment to achieving results
that are being created within this new enterprise.” (Bob Wood, CEO,
Chemtura 7/29/05)
 “I am very proud of the remarkable growth and progress Yahoo! has
demonstrated throughout this past year.” (Terry Semel, CEO, Yahoo
1/17/06)
 “We’re really thrilled by the way customers are responding to these stores
as they’re performing extremely well and they’re exceeding our sales
expectations.” (George Jones, CEO, Borders 5/27/08)
16
Sample Construction
 Identify CEOs/CFOs who have occupied the CEO/CFO
position in at least two companies for at least one year
in each firm between 2002 and 2009
 Gather conference call transcripts for firm quarters
between 2002 and 2009
 Eliminate managers who did not participate in at least
two quarterly conference calls at each firm
•
104 CEOs and CFOs in our sample (69 CFOs, 31 CEOs, 4
CEO/CFOs)
17
Sample Construction
 “Manager-firm matched sample” are firm-quarters of those managers
who move firms (for which we measure fixed effects)
 “Filler quarters” are firm-quarters for which we do not estimate a
manager fixed effect (because manager does not move firms)
2002
2003
2004
2005
2006
2007
2008
2009
CEO: Steve Odland
Autozone
Autozone
Autozone
Filler Qtrs Filler Qtrs
Filler Qtrs
Filler Qtrs
Filler Qtrs
Filler Qtrs
Filler Qtrs
Filler Qtrs
Office
Depot
Office
Depot
Office
Depot
Office
Depot
Office
Depot
Disentangle CFO-specific effects from firm-specific and time-specific
effects
18
Effect of Style on Tone – Research Design
(1) Relation btw tone and current and future performance:
TONEit   0  1MBE it   2 SURPit   3 LOSS it   4 RETURN it   5 ROAit
  6 ROAit 1   7 ROAit  2   8 ROAit 3   9 ROAit  4   it
Current Performance
Future Performance
(2) Manager-specific effect:
RESIDUALit   0  FIRM i  YEARt  QTR k  MANAGER j   it
19
Relation between Tone and Current and Future Performance
Table 4
Tone_D
Tone_H
Tone_LM
Intercept
1.44***
1.66***
0.43***
MBE
0.19***
0.33***
0.21***
SURP
-0.55
-1.36
3.13**
LOSS
-0.23***
-0.26***
-0.25***
RETURN
0.11*
0.29***
0.31***
ROA
0.50
3.23***
2.32***
ROAt+1
-0.63
3.49***
1.78***
ROAt+2
0.40
2.30***
1.51**
ROAt+3
0.23
1.52**
0.34
ROAt+4
0.15
-0.13
0.14
Adj R2
2.40%
9.3%
8.8%
20
Manager Effect on Residual Tone
Table 5, Panels A & B
Tone_D
Tone_H
Tone_LM
Base Adj R2
38%
37%
36%
Full Adj R2
44%
41%
41%
F-stat
4.72
3.39
4.05
p-value
0.001
0.001
0.001
5% level
29%
20%
22%
10% level
37%
31%
31%
Percent negative
46%
46%
42%
Significant effects:
21
Market Reaction to Tone – Research Design
Relation btw 3-day returns and tone:
CARit   0   1MBE it   2 SURPit   3 LOSS it   4 RETURN it
  5 ROAit   6 ROAit 1   7 ROAit  2   8 ROAit 3   9 ROAit  4
  10TONE FULLit   11MANAGERj   it
Current
Performance
Residual
Tone
Manager
Effect
Future
Performance
22
Market Reaction to Tone
Table 6, Panel B
Tone_D
Tone_H
Tone_LM
Intercept
-0.050***
-0.037***
-0.032***
MBE
0.039***
0.040***
0.038***
SURP
0.816***
0.799***
0.765***
LOSS
0.015**
0.014**
0.015*
-0.059***
-0.058***
-0.061***
ROA
0.054
0.051
0.31
ROAt+1
-0.049
-0.047
-0.070
ROAt+2
0.102
0.107
0.091
ROAt+3
0.146
0.141
0.131
ROAt+4
0.028
0.028
0.039
0.012***
0.005*
0.015***
RETURN
TONEFULL
23
Market Reaction to Manager Specific Tone
Table 6, Panel C
Tone_D
Tone_H
Tone_LM
Intercept
-0.049***
-0.036***
-0.032***
MBE
0.039***
0.040***
0.038***
SURP
0.827***
0.803***
0.802***
LOSS
0.015**
0.014*
0.014**
-0.059***
-0.058***
-0.061***
ROA
0.053
0.053
0.025
ROAt+1
-0.048
-0.044
-0.064
ROAt+2
0.103
0.108
0.094
ROAt+3
0.147
0.143
0.135
ROAt+4
0.027
0.029
0.038
TONE
0.011***
0.004
0.014***
0.001
0.001
0.006
RETURN
MANAGER
24
Market Reaction to Manager Specific Tone
Table 6, Panel D
Tone_D
Tone_H
Tone_LM
Intercept
-0.052***
-0.037***
-0.033***
MBE
0.038***
0.040***
0.038***
SURP
0.882***
0.803***
0.819***
LOSS
0.015**
0.014*
0.014**
-0.059***
-0.058***
-0.061***
ROA
0.043
0.049
0.023
ROAt+1
-0.046
-0.045
-0.062
ROAt+2
0.106
0.108
0.096
ROAt+3
0.152
0.142
0.137
ROAt+4
0.031
0.028
0.038
TONE
0.010**
0.004
0.013*
MANAGERPOS
0.012
0.004
0.011
MANAGERNEG
-0.011
-0.002
0.000
RETURN
25
Robustness checks
 Use only effects that are significant at the 10% level
• Significantly smaller sample (500 obs)
• No overall relation between returns and manager specific
component of tone
• Positive relation for optimistic managers but only using
Diction measure
Use decile ranks of manager effects
• Positive coefficient using L&M at 10% level
• Positive coefficient on top decile indicator for Diction (1%
level) and L&M (5% level)
26
Summary and Conclusion
 Manager “style” has a significant impact on tone of
presentation and Q&A
•
Impact is larger than “style” effects in other contexts
 Some evidence that the market prices the managerspecific component of tone for optimistic managers but
not pessimistic managers
•
BUT, need to check robustness of results
 Next steps:
•
•
Intraday trading data to measure price reactions
Increase sample size
27
28
Manager-firm matched sample (Table 2A)
Firm-quarters
Firms
Mgrs
Initial
3,326
415
206
No Transcripts
(658)
(93)
(46)
Fewer than 2 qtrs per firm
(836)
(128)
(56)
Manager Firm Matched
Sample
1,832
194
104
29
Sample Selection (Table 2B)
N of quarters in each
firm
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20 and above
Total
N of manager-firm
pairs
18
22
19
23
18
17
18
13
2
5
8
5
8
4
5
5
2
6
11
209
Percentage (%)
8.61
10.53
9.09
11
8.61
8.13
8.61
6.22
0.96
2.39
3.83
2.39
3.83
1.91
2.39
2.39
0.96
2.87
5.27
100
30
Frequency of Firms based on the number of different
Managers (Table 2C)
No. of different
Mgrs
Freq of firms
Percentage
No. of Mgr-firm
pairs
1
180
92.8
180
2
13
6.7
26
3
1
0.5
3
Total
194
100
209
31
Frequency of Managers based on the number of firm
changes (Table 2D)
No. of
changes
Freq of Mgrs
Percentage
No. of Mgr-firm
pairs
1
103
99
206
2
1
1
3
Total
104
100
209
32
Descriptive statistics (Table 3)
Variable
N
Min
Q1
Mean
Median
Q3
Max
Std. Dev.
ASSETS
4123
113.27
883
29,296
2,201
10,298
1,022,237
122,281
ROA
4123
-0.1238
0.00183
0.00863
0.00860
0.01893
0.08281
0.02581
ROAt+1
4120
-0.1231
0.00173
0.00853
0.00858
0.01874
0.08103
0.02565
ROAt+2
4111
-0.1271
0.00174
0.00823
0.00859
0.01869
0.07894
0.02596
ROAt+3
4093
-0.1271
0.00183
0.00821
0.00871
0.01874
0.07894
0.02596
ROAt+4
4018
-0.1231
0.00173
0.00828
0.00871
0.01892
0.07891
0.02561
SURP
3628
-0.07
-0.0004
-0.0004
0.0005
0.002
0.03
0.01
MBE
3630
0
0
0.72
1
1
1
0.45
LOSS
4123
0
0
0.21
0
0
1
0.40
RETURN
4235
-0.39
-0.10
0.003
-0.005
0.09
0.52
0.18
TONE_D
4390
-0.35
0.87
1.53
1.45
2.11
4.07
0.91
TONE_H
4390
-0.57
1.18
1.94
1.89
2.65
4.92
1.11
TONE_LM
4390
-1.51
0.00
0.58
0.52
1.12
2.92
0.87
TONE_DFULL
4290
0.67
1.51
1.88
1.86
2.23
3.23
0.54
TONE_HFULL
4290
0.35
1.31
1.78
1.74
2.20
3.60
0.67
TONE_LMFULL
4290
-1.04
-0.03
0.33
0.33
0.68
1.71
0.55
CAR
1624
-0.25
-0.03
0.007
0.006
0.05
0.28
0.08
33
Comparsion to Compustat (Table 3)
Our sample
Compustat
Difference in mean
(sample vs. Compustat)
Variable
Mean
Median
Mean
Median
ASSETS
29,296.46
2,200.90
3,531.07
202.97
25,765.39***
ROA
0.00863
0.00860
-0.03491
0.00249
0.0435***
ROAt+1
0.00853
0.00858
-0.03490
0.00249
0.0434***
ROAt+2
0.00823
0.00859
-0.03496
0.00248
0.0432***
ROAt+3
0.00821
0.00871
-0.03491
0.00246
0.0431***
ROAt+4
0.00828
0.00871
-0.03408
0.00249
0.0424***
SURP
-0.0004
0.0005
-0.003
0.0003
0.002***
MBE
0.72
1
0.65
1
0.07***
LOSS
0.21
0
0.42
0
-0.21***
RETURN
0.003
-0.005
0.022
-0.007
-0.02***
34
Correlation Matrix
TONE_D
TONE_H
TONE_LM
VCAR
MBE
SURP
LOSS
RETURN
ROAt+1
TONE_D
1.000
0.610
0.646
0.081
0.124
0.073
-0.154
0.054
0.057
TONE_H
0.614
1.000
0.682
0.100
0.191
0.106
-0.233
0.101
0.224
TONE_LM
0.655
0.684
1.000
0.114
0.183
0.136
-0.230
0.114
0.209
VCAR
0.068
0.085
0.098
1.000
0.284
0.341
-0.062
-0.055
0.066
MBE
0.123
0.194
0.183
0.242
1.000
0.777
-0.205
0.106
0.206
SURP
0.082
0.120
0.146
0.158
0.459
1.000
-0.145
0.162
0.102
LOSS
-0.148
-0.234
-0.233
-0.033
-0.205
-0.279
1.000
-0.111
-0.460
RETURN
0.049
0.099
0.112
-0.050
0.107
0.090
-0.092
1.000
0.153
ROAt+1
0.084
0.224
0.199
0.061
0.168
0.149
-0.389
0.161
1.000
35
Summary Statistics on Manager Effects (Table 5c)
Q1
Mean
Median
Q3
% Neg
EFFECT_TONE_D
-0.37
0.0002
0.05
0.35
46%
EFFECT_TONE_H
-0.29
0.030
0.03
0.37
46%
EFFECT_TONE_LM
-0.27
0.027
0.08
0.30
42%
36
Summary Statistics on Firm Effects (Untabulated)
Overall Tone
Q1
Mean
Median
Q3
EFFECT_TONE_D
0.16
0.63
0.59
1.04
EFFECT_TONE_H
0.01
0.47
0.44
0.86
EFFECT_TONE_LM
0.97
1.34
1.28
1.68
37
Relation between Tone and Current and Future Performance
CFO vs. CEO (Tone_D only)
Base
CEO interaction
Intercept
1.22***
0.58***
MBE
0.14***
0.14**
SURP
0.64
0.16
LOSS
-0.27***
0.18**
RETURN
0.12
-0.05
ROA
0.40
0.63
ROAt+1
-0.52
0.76
ROAt+2
-0.23
1.57
ROAt+3
0.40
-0.28
ROAt+4
-0.01
-0.59
38
Compute F-statistics
F-statistics =
(R2 – R2 *)/J
(1-R2)/(N-J-K)
For Tone_D: F-stat = (0.48656-0.412392)/99
= 4.72
(1-0.48656)/[3523-99-(177+7+3)]
39
40