The Effect of Managerial “Style” on the Tone of Earnings Conference Calls

Angela DavisWeili GeDawn MatsumotoJenny Li Zhang

April 27, 2011

22

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?

33

We are a company known for being conservative.

—Greg Maffei, Microsoft CFO 1997-2000

44

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

55

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

66

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

77

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 manager-specific 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

88

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

99

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)

1010

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

1111

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

1212

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)

1313

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”

1414

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

1515

Examples of Differences between WordlistsDiction 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 YesExceeding No Yes No

1616

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)

1717

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)

1818

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)

Disentangle CFO-specific effects from firm-specific and time-specific effects

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

1919

Effect of Style on Tone – Research Design

(1) Relation btw tone and current and future performance:

(2) Manager-specific effect:

Residual tone

itjktiit MANAGERQTRYEARFIRMRESIDUAL 0

Manager effect

ititititit

itititititit

ROAROAROAROAROARETURNLOSSSURPMBETONE

49382716

543210

Current Performance Future Performance

2020

Relation between Tone and Current and Future Performance Table 4

Tone_D Tone_H Tone_LMIntercept 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%

2121

Manager Effect on Residual ToneTable 5, Panels A & B

Tone_D Tone_H Tone_LMBase 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

Significant effects:

5% level 29% 20% 22%

10% level 37% 31% 31%

Percent negative 46% 46% 42%

2222

Market Reaction to Tone – Research Design

Relation btw 3-day returns and tone:

itjitFULL

ititititit

ititititit

MANAGERTONE

ROAROAROAROAROARETURNLOSSSURPMBECAR

1110

493827165

43210

Future PerformanceResidual

ToneManager

Effect

Current Performance

2323

Market Reaction to ToneTable 6, Panel B

Tone_D Tone_H Tone_LMIntercept -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*RETURN -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

TONEFULL 0.012*** 0.005* 0.015***

2424

Market Reaction to Manager Specific ToneTable 6, Panel C

Tone_D Tone_H Tone_LMIntercept -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**RETURN -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***MANAGER 0.001 0.001 0.006

2525

Market Reaction to Manager Specific ToneTable 6, Panel D

Tone_D Tone_H Tone_LMIntercept -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**RETURN -0.059*** -0.058*** -0.061***ROA 0.043 0.049 0.023ROAt+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 0.011MANAGERNEG -0.011 -0.002 0.000

26

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)

2727

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 manager-specific 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

2828

2929

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

3030

Sample Selection (Table 2B)N of quarters in each

firm N of manager-firm

pairs Percentage (%) 2 18 8.61 3 22 10.53 4 19 9.09 5 23 11 6 18 8.61 7 17 8.13 8 18 8.61 9 13 6.22

10 2 0.96 11 5 2.39 12 8 3.83 13 5 2.39 14 8 3.83 15 4 1.91 16 5 2.39 17 5 2.39 18 2 0.96 19 6 2.87

20 and above 11 5.27 Total 209 100

3131

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

3232

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

3333

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

3434

Comparsion to Compustat (Table 3) 

Our sample CompustatDifference 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***

3535

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

3636

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%

3737

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

3838

Relation between Tone and Current and Future Performance CFO vs. CEO (Tone_D only)

Base CEO interactionIntercept 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

3939

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)]

4040