Managers Self-Inclusive Language in Conference Calls: Multi-Method - - PowerPoint PPT Presentation

Managers’ Self-Inclusive Language in Conference Calls: Multi-Method Evidence

Zhenhua Chen Serena Loftus Tulane University March 2017

• How do investors react to managers’ SIL?
• SIL: self-inclusive language

– Statements that explicitly include the speaker – Two types:

• Individual: I, me, mine
• Collective: We, us, ours

Research Question

Managers and SIL

I

Managers and SIL

We

Managers and SIL

SEL

Managers and SIL

SEL

?

Motivation SIL

• Role in impression management
• SIL processed automatically in brain

– Suggests investors cannot “unwind” SIL (Kimbrough and Wang 2014; Cianci and Kaplan 2010)

• Managers have some latitude to substitute SIL

– Can use SIL vs. SEL or collective vs. individual SIL

Examples of SIL in impression management

More collective SIL after 9/11 (Pennebaker and Lay 2003)

Examples of SIL in impression management “As silly as it sounds, pronouns matter. Whenever possible...substitute ‘we’ for ‘I.’”

– Sheryl Sandberg, Chief Operating Officer of Facebook in her book Lean In

Contribution

• SIL can play a role in impression management

– Complements Cho, Roberts and Patten 2010; García Osma and Guillamón-Saorín 2011; Merkl-Davies, Brennan and McLeay 2011; Kimbrough and Wang 2014

• Function language impacts investors

– Complements research focused on content language (Li 2008; Davis, Pigor and Sedor 2012; Hales, Kuang and Venkataraman 2011; Loughran and McDonald 2011)

• Traditional attribution theories may not apply to managers

– Aerts 2005; Merkle-Davis and Brennan 2011; Clatworthy and Jones 2003

• Multiple research methods to complement findings

– Address concerns by Koonce and Mercer 2005; Li 2011; answers call from Bloomfield, Nelson and Soltes 2016

• Growing research investigating managers’ language

Institutional Background: Language

Two Types of Language Content Language Used to convey content or meaning (e.g., nouns, adjectives, verbs) Function Language Is the “…linguistic glue used to hold content words together.” (Groom and Pennebaker 2002) (e.g., pronouns, articles, conjunctions, etc.)

Real-World Example

“Through our expansion program, we're positioning ourselves for the surge in prescription use that's already underway. And we anticipate getting a bigger slice of that growing pie. The demographics isn't the only thing we have going for us. Our financial position is stronger than it's ever been. We're generating strong cash flow from operations. We have virtually no debt. We have nearly $300 million in the bank, and $3 billion in owned real estate. We're growing in a down economy, and our superior real estate gives us a dramatic competitive

• edge. Bottom line, we're excited about what we can accomplish during fiscal

2003.” – Rick Hans, Director of Finance, Walgreen’s (1/3/2003; $0.01 + ES)

Real-World Example

“Through our expansion program, we're positioning ourselves for the surge in prescription use that's already underway. And we anticipate getting a bigger slice of that growing pie. The demographics isn't the only thing we have going for us. Our financial position is stronger than it's ever been. We're generating strong cash flow from operations. We have virtually no debt. We have nearly $300 million in the bank, and $3 billion in owned real estate. We're growing in a down economy, and our superior real estate gives us a dramatic competitive

• edge. Bottom line, we're excited about what we can accomplish during fiscal

2003.” – Rick Hans, Director of Finance, Walgreen’s (1/3/2003; $0.01 + ES)

– Readability (FOG) score: 11.5 – Affect: 4%

Real-World Example

“Through our expansion program, we're positioning ourselves for the surge in prescription use that's already underway. And we anticipate getting a bigger slice of that growing pie. The demographics isn't the only thing we have going for us. Our financial position is stronger than it's ever been. We're generating strong cash flow from operations. We have virtually no debt. We have nearly $300 million in the bank, and $3 billion in owned real estate. We're growing in a down economy, and our superior real estate gives us a dramatic competitive

• edge. Bottom line, we're excited about what we can accomplish during fiscal

2003.” – Rick Hans, Director of Finance, Walgreen’s (1/3/2003; $0.01 + ES)

– 54% Function Language – 15% Collective SIL

• Archival analysis of over 50,000 earnings conference calls
• SIL approximately 6.7% of words spoken
• Substantial variation in SIL:

– 3.51% at 1% – 9.81% at 99%

• Support for substitution of individual/collective SIL

– Negative correlation coefficient (-0.12; p = 0.01)

• Overall, SIL material portion of managers’ remarks

Institutional Background: Use of SIL

H1

SIL Perceptions of Controllability Investment

• Explicit attributions have been shown to increase perceptions of

managers’ controllability – e.g., Elliott, Hodge and Sedor 2011

• Our innovation: SIL increases perceptions of controllability

– Even when not in attributions – Gives the impression the manager is associated with outcomes

+ +

H1

SIL Perceptions of Controllability Investment

• Normally, perceptions of control moderated by outcomes

– Crant and Bateman 1993

• For CEOs, high controllability always positive regardless of news

– Siegel and Brockner 2005; Lee, Peterson and Tiedens 2004

• High controllability for negative events creates impression CEO is more in

control of future events

– Lee and Robinson 2000; Salancik and Meindl 1984

+ +

H1

SIL Perceptions of Controllability Investment

H1a: Investors will react more positively to disclosures containing high SIL than low SIL (regardless of news).

+ +

• Use experiment to test H1: strong causal evidence
• 2 x 2 between-subjects design with manipulated variables:

– Language (pooled): SIL, SEL – News: Good, Bad

• 491 Amazon.com Mechanical Turks complete study

– Appropriate proxies for RQ – 236 participants fail at least one comprehension check question – 3 comprehension check questions: identify news, identify name of CEO, identify correct EPS forecast

• Most of the comprehension check failures are clustered by participant and occur on the

news question (statistically more in the bad news condition)

– 255 participants included in final sample for hypothesis testing

• Participants earn $1.80 flat wage

Research Design

(I am/We are/Webtex is) [pleased/disappointed] to announce second-quarter earnings for the 2016 fiscal year of $6 per share, which consists of revenue of $182,795 and net income of $39,510. (I/We/Management) had expected earnings per share of [$5/$7], so our (Webtex’s) earnings are [higher/lower] than (my/our/management’s) forecast by $1 per share. (My/Our/Management’s) earnings forecast was based on Webtex’s net income from the second-quarter of the last fiscal year, which means that Webtex’s current earnings are [higher/lower] than second-quarter earnings for the 2015 fiscal year. As (I/we/Webtex’s managers) look ahead to the next quarter, (I am/we are/ management is) very optimistic about Webtex’s future. (I am/We are/ Management is) currently forecasting earnings-per-share of $6.75 for the next quarter, which is the third quarter of the 2016 fiscal year. (I/We/Webtex’s managers) want to thank you for your continued support of

• Webtex. This concludes (my/our/management’s) comments for the quarter.

Manipulation (Recorded by Voice Actor)

Experimental Flow (2-stage design)

Receive information about a fictional firm Webtex Listen to conference call excerpt containing language and news manipulation Read letter to shareholder containing fully-crossed 1 x 2 Language manipulation Answer dependent variable questions Answer supplemental questions Answer demographic questions Within-Subjects Follow-up Experiment

Results (Figure 1 Panel A)

20 30 40 50 60 70 80 Good News Bad News Likelihood SEL SIL

SIL > SEL: F=3.53; p < 0.03, one-tailed

Panel B: Analysis of variance Dependent Source of Sum of Mean Variable Variation Squares d.f. Square F-Statistic p-value Likelihood SIL 1598.72 1 1598.72 3.53 0.06 News 28950.53 1 28950.53 63.97 <0.01 SIL x News 122.71 1 122.71 0.27 0.60 Error 113591.3 251 452.44 Investment SIL 17216633 1 17216633 3.74 0.05 News 77634350 1 77634350 16.88 <0.01 SIL x News 2639716 1 2639716 0.57 0.45 Error 1.16E+09 251 4599968

Results (Table 2)

Results (Table 2)

Panel A: Descriptive Statistics [standard deviations] Language News n Likelihood Investment SEL Good 42 67.10 [21.47] 3,144.07 [2,469.16] Bad 31 26.81 [16.91] 1,057.74 [1,018.46] SIL (I + We) Good 98 74.47 [20.57] 3,909.32 [2,530.71] Bad 84 31.08 [23.30] 1,368.71 [ 1,743.39] Within SIL: I Good 55 76.42 [19.95] 4,117.25 [2,586.02] Bad 45 35.91 [23.83] 1,542.84 [1,963.10] We Good 43 71.98 [21.31] 3,643.35 [2,462.54] Bad 39 25.51 [21.66] 1,167.79 [1,448.84]

Exploratory Analysis of SIL Type (Figure 1 Panel B)

20 30 40 50 60 70 80 Good News Bad News Likelihood SEL I We

Panel C: Multiple Regression Analysis - Dependent Variable is Likelihood (standard errors) Column (1) (2) (3) (4) (5) (6) SIL 7.37

++

(3.92) SIL x News

• 3.10

(5.95) I 6.85 * 4.44 9.32 ** (3.65) (4.41) (4.29) I x News 2.97 5.96

• 0.22

(5.43) (6.47) (6.50) We

• 0.40

4.88 (3.91) (4.48) We x News

• 6.28
• 6.18

(5.74) (6.69) News

• 40.29 ***
• 43.48 ***
• 40.18 ***
• 46.46 ***
• 40.29 ***
• 40.29 ***

(5.04) (3.40) (3.27) (4.79) (4.89) (4.95) Intercept 67.10 *** 69.56 *** 72.38 *** 71.98 *** 67.10 *** 67.10 *** (3.28) (2.28) (2.17) (3.30) (3.19) (3.23) Observat ions 255 255 255 182 155 173 Adj R- squared 0.50 0.51 0.49 0.50 0.52 0.49 Included ? SEL Yes Yes Yes No Yes Yes I Yes Yes Yes Yes No Yes We Yes Yes Yes Yes Yes No

Experimental Flow (2-stage design)

Receive information about a fictional firm Webtex Listen to conference call excerpt containing language and news manipulation Read letter to shareholder containing fully-crossed 1 x 2 Language manipulation Answer dependent variable questions Answer supplemental questions Answer demographic questions Within-Subjects Follow-up Experiment

Results from Within-Subjects Experiment (Table 3)

Experiment 1 Alternative Sign if Language Language n prefer "We" Preference T-ratio vs. 0 I We 52 + 16.81 [29.93] 4.05 [p < 0.01] SEL 48 N/A 3.17 [34.39] 0.64 [p = 0.53] We I 43

• 27.53 [27.25]

6.62 [p < 0.01] SEL 38

• 19.37 [28.41]

4.20 [p < 0.01] SEL I 52 N/A

• 6.89 [34.69]

1.19 [p = 0.24] We 48 + 18.94 [30.55] 3.72 [p < 0.01]

Insights:

• Strong explicit preference for collective SIL over individual SIL & SEL
• Inconsistent with revealed preference for individual SIL
• Suggests investors may be unaware of the influence of SIL

• Support that investors’ react positively to SIL over SEL

– Support for H1

• Positive reaction to SIL driven primarily by positive reaction to individual

SIL

• Revealed preference for individual SIL inconsistent with explicit preference

for collective SIL

Summary of Experiment Findings

• Purpose of the Archival Sample:

– Provide real-world evidence of managers’ use of SIL – Triangulate experiment findings, where possible

• Approximately 50,000 conference call transcripts from 2002-2007
• Only CEO comments from both MD and QA
• Use LIWC to measure SIL

Supplemental Archival Analysis

Table 1 (Archival Sample Descriptive Statistics)

Panel A: Sample Construction Number of quarterly conference call transcripts on StreetEvents database from 2002 to 2007 62,280 Less: Transcripts with no CEO participation in the calls (10,277) Less: Transcripts with no sufficient COMPUSTAT or CRSP data (1,569) Final sample of firm-quarter observations 50,434

Table 1 (Archival Sample Descriptive Statistics)

Insights:

• SIL 6.67% of managers’ remarks
• Large variation in SIL
• More We than in daily speech

Panel B: Descriptive Statistics of the Archival Sample (n= 50,434) Variable Mean Std Dev P1 P25 Median P75 P99 Daily Speech PRONOUN 8.16 1.46 4.45 7.20 8.18 9.12 11.82 13.63 SIL 6.67 1.27 3.51 5.83 6.68 7.52 9.81 7.39 I 1.24 0.59 0.21 0.81 1.15 1.57 3.21 6.30 WE 5.42 1.19 2.54 4.62 5.40 6.21 8.47 1.09 LENGTH 3.05 1.66 0.29 1.82 2.80 4.01 8.54 ROA 0.01 0.04

• 0.19

0.00 0.02 0.03 0.11 LNMVE 6.77 1.66 2.97 5.64 6.70 7.83 11.03 BM 0.48 0.35

• 0.32

0.26 0.43 0.64 1.92 FIRMAGE 15.69 15.48 0.00 6.00 10.00 20.00 78.00

Table 1 (Archival Sample Descriptive Statistics)

Insights:

• Negative correlation between I and We

Panel C: Pearson Correlations Variable SIL I WE ROA LNMVE BM FIRMAGE I 0.36*

• WE

0.88*

• 0.12*
• ROA

0.11* 0.06* 0.08*

• LNMVE

0.03* 0.12*

• 0.02*

0.41*

• BM

0.04*

• 0.01

0.04*

• 0.10*
• 0.23*
• FIRMAGE

0.01 0.08*

• 0.03*

0.14* 0.35*

• 0.00
• LENGTH
• 0.07*

0.12*

• 0.12*

0.05* 0.15*

• 0.09*

0.03*

Table 4: Determinants of Managers’ SIL

Insights: SIL positively associated with firm performance Managers at big firms use more SIL

SIL I WE (1) (2) (3) ROA 0.07*** 0.02*** 0.06*** (0.01) (0.01) (0.01) LNMVE 0.01 0.08***

• 0.03***

(0.01) (0.01) (0.01) BM 0.03*** 0.02** 0.02** (0.01) (0.01) (0.01) FIRMAGE

• 0.00

0.06***

• 0.03**

(0.01) (0.01) (0.01) LENGTH

• 0.07***

0.11***

• 0.12***

(0.01) (0.01) (0.01) Intercept 5.50*** 1.83*** 4.92*** (0.17) (0.13) (0.15) Industry FE Yes Yes Yes Year FE Yes Yes Yes S.E. Clustering Firm & Quarter Firm & Quarter Firm & Quarter N 50434 50434 50434 Adjusted R2 0.070 0.039 0.086

Predicted Sign (1) (2) (3) (4) IvsWE + 0.22*** 0.23*** (0.04) (0.07) IvsWE x MBE ?

• 0.01

(0.08) I ? 0.00 (0.07) I x MBE ? 0.15** (0.07) WE ?

• 0.39***

(0.08) WE x MBE ? 0.25*** (0.08) Industry Fixed Effect Yes Yes Yes Yes Time Fixed Effects Yes Yes Yes Yes S.E. Clustering Firm & Quarter Firm & Quarter Firm & Quarter Firm & Quarter N 42559 42559 42559 42559 Adjusted R2 0.106 0.106 0.105 0.106

Table 5: SIL & Mkt Reactions

• SIL influences investors’ reactions to disclosures
• Investors’ reactions to disclosures are more positive when SIL is present, regardless
• f news

– Consistent with positive reactions to increased perceptions of managers’ controllability – Primarily driven by positive reaction to individual SIL regardless of news

• Investors may be unaware of the impact of SIL on their decisions
• While investors reactions to SIL are not affected by news, managers use more SIL

when news is positive

• Overall, tiny words can influence investors’ reactions

Conclusion

Thank you!