It s Not Who You Know It s Who Knows You: Employee Social Capital and Firm Performance

(1)

It’s Not Who You Know—It’s Who Knows You:

Employee Social Capital and Firm Performance

DuckKi Cho

Peking University

Michael Hertzel

Arizona State University

Lyungmae Choi

City University of Hong Kong

Jessie Jiaxu Wang

(2)

Social Capital and Corporations

• A firm’s social capital consists of the

relationships

that a firm and its

employees have built with external stakeholders (Servaes, Tamayo 2017).

• One view: social capital is a societal characteristic (Putnam 1993)

–

regions with more social capital enjoy better economic outcomes (La Porta et

al. 1997); local firms have better access to capital (Kuchler et al. 2017)

• Another view: social capital is an individual asset embedded in networks,

enables access to resources and information (Lin 2002)

–

a person’s social capital: social skills, charisma, the size of their Rolodex

➢

Q: How does social capital embodied in employees’ connections

contributes to firm performance and valuation?

(3)

This Paper: employee social capital and firm performance

• Measure

: to measure firm-level social capital derived from its employees’

connections, we use unique data from a professional networking app.

• Findings

: firms with more employee social capital perform better:

– the positive effect arises from employees being valued/remembered by others.

– connections by all levels of employees matter, not just the top.

• Causal evidence

: we exploit a negative shock to professional networks, in

the context of an anti-graft act.

(4)

Outline

•

• Data

Employee social capital and firm performance

• Causal evidence from the Anti-Graft Act

(5)

Data: a professional networking app

• A unique

cultural institution to track networks

– Exchanging business cards is an important ritual for building professional connections in Asia.

– App-users upload the cards collected to keep track of their professional network.

• Novel data from a professional networking app,

Remember.

‒ No. 1 business card management app in Korea

‒ 140 million cards uploaded, 2.5 million users by Dec 2018

(6)

Data: a professional networking app

• A unique

cultural institution to track networks

– Exchanging business cards is an important ritual for building professional connections in Asia.

– App-users upload the cards collected to keep track of their professional network.

• Novel data from a professional networking app,

Remember.

‒ No. 1 business card management app in Korea

‒ 140 million cards uploaded, 2.5 million users by Dec 2018

(7)

Data: business card exchange network

No. of connections 12,391,177

No. of employees 2,363,295

No. of employees who are app-users 411,039 No. of employees in public firms 1,016,023 No. of employees in public firms who are app-users 119,423

No. of firms in KIS 126,987

No. of public firms in OSIRIS Industrials 1,866

• Unit of observation is a

connection pair

consisting of the app-user and external

contact whose card is uploaded.

Employee_from Firm_from Job_from Employee_to Firm_to Job_to

A 1 Staff C 2 Staff A 1 Staff D 2 President A 1 Staff E 3 Manager E 3 Manager A 1 Staff E 3 Manager B 1 Manager A B C D E Firm 1 Firm 2 Firm 3 F

(8)

Data: directed nature of the connections

• While card exchanges are mutual, uploading cards to the app is not necessarily

mutual; our connection-level data are

directed

-> infer the

value of connections

.

A B C D E Firm 1 Firm 2 Firm 3 F

(9)

Data: directed nature of the connections

• While card exchanges are mutual, uploading cards to the app is not necessarily

mutual; our connection-level data are

directed

-> infer the

value of connections

.

• In-degree: number of links pointing inward

(who uploads your card: “who knows you”)

(10)

Data: directed nature of the connections

• While card exchanges are mutual, uploading cards to the app is not necessarily

mutual; our connection-level data are

directed

-> infer the

value of connections

.

• In-degree: number of links pointing inward

(who uploads your card, “who knows you”)

• Out-degree: number of links going outward

(whose cards you upload, “who you know”)

• Total degree = In-degree + Out-degree

• Firm-level ESC construction

ESC in-degree: average In-degree across firm’s employees in the network. ESC out-degree: average Out-degree across app-user employees of a firm.

(11)

• Korean public firms from OSIRIS Industrials, 2014—2018

• 5340 firm-year observations, 1553 unique firms

Sample

Mean Median SD ESC total degree _6.84 _5.32 _5.84 ESC in-degree _3.68 _3.14 _2.39 ESC out-degree _31.0 _24.2 _26.8 Tobin’s q _1.46 _1.11 _1.10

ROA _0.04 _0.04 _0.09

(12)

Employee Social Capital and Performance: total degree

Table 3 ESC total degree (total connections)

Y = Tobin’s q ROA Sales

Growth

ln(1+ESC) 0.084 0.008** 0.038***

(0.053) (0.004) (0.012)

Controls Y Y Y

FE Ind × Year Ind × Year Ind × Year

Obs. 5,340 5,340 5,340

Adj. R2 _0.248 _0.148 _0.035

Yi,t = 𝛽0 +

𝜷1

× ln 1+ESCi,t−1 + 𝛾

′

Xi,t−1 + 𝛼j,t + 𝜀i,t, t = 2015—2018

• Lagged controls: R&D, book leverage, total assets, stock return volatility, firm age, emp

(13)

Directions of Connections: in-degree vs. out-degree

Table 4 ESC in-degree (Who Knows You) ESC out-degree (Who You Know) Y = Tobin’s q ROA Sales

Growth Tobin’s q ROA

Sales Growth ln(1+ESC) 0.330*** 0.021*** 0.098*** 0.042 0.004* 0.004

(0.090) (0.007) (0.024) (0.030) (0.002) (0.007)

Controls Y Y Y Y Y Y

FE Ind × Year Ind × Year Ind × Year Ind × Year Ind × Year Ind × Year Obs. 5,340 5,340 5,340 4,994 4,994 4,994 Adj. R2 _0.252 _0.150 _0.038 _0.252 _0.142 _0.035

• Positive effect from in-degree: employees being remembered by others

• One std dev (3.1) ↑in ESC in-degree -> ROA ↑ 0.9 ppts (mean = 4.3 ppts)

(14)

Directions of Connections: in-degree vs. out-degree

• Why is “Who Knows You” more valuable than “Who You Know”?

‒ Social capital literature: networks provide goodwill, resources, information

‒ Our results: whether employees can mobilize these benefits for their employer depends on if contacts value them

‒ If employees having broad network (out-degree) expands outside job opportunities (Gortmaker, Jeffers, Lee 2020), benefits do not accrue to current employer.

• Robustness

‒ App-users vs. Non-users

‒ Reciprocal connections

‒ Efforts by sales personnel Employee skill levels

(15)

Whose Connections are Valuable?

Y = Tobin’s q ROA Sales

Sales Growth

Executives Non-Executive Employees

ln(1+ESC) 0.190*** 0.013*** 0.050*** 0.207** 0.032*** 0.090*** (0.056) (0.004) (0.013) (0.100) (0.008) (0.025)

Non-Executive Managers Rank-and-File Employees

ln(1+ESC) 0.159* 0.029*** 0.068*** 0.136 0.024*** 0.084*** (0.089) (0.007) (0.022) (0.108) (0.007) (0.022)

• ESC of non-executive employees has larger economic significance on ROA, Sales Growth, ESC of executives has bigger impact on Tobin’s q

• One std dev ↑ in ESC of executives, non-executive managers, rank and file → ROA ↑ 0.7, 1.3, 0.8 ppts

(16)

Establishing a Causal Relation: the Anti-Graft Act

• A shock: the 2016 enactment of the Improper Solicitation and Graft Act

– Illegal for employees in the media & public sector to accept gifts; limits meal expenses

• Identification: a negative shock to ESC with the media & public sector

– Business precautions: fewer social events and meetings with employees of media & public, restricting firms’ ability to leverage their employee social capital

– Aggressive enforcement: imposing penalties such as imprisonment

(17)

Causal Effect of Employee Social Capital on Performance

Yi,t = 𝛽0 + 𝛽1 × Act Exposurei + 𝜷2 × Act Exposurei × Postt + 𝛾′ Xi,t−1 + 𝛼j,t + 𝜀i,t Act Exposure = ESC Act/ESC in 2015; Post = 1 for 2016—2018

Table 6 Panel A Y = Tobin’s q

Act Exposure 6.578*** 6.640***

(1.273) (1.272) Act Exposure × Post -4.930*** -4.726***

(1.132) (1.052)

Controls Y Y

FE Ind × Year Ind × Year

Including 2016 N Y

Obs. 3,778 5,101

Adj. R2 _0.242 _0.245

• One std dev ↑in Exposure → Tobin’s q ↑ 17.5% before the Act, only 4.4% after.

(18)

Causal Effect of Employee Social Capital: dynamic trend

Yi,t = 𝛽0 + 𝛽1 × Act Exposurei + ෍

𝑡=2015 2018

𝜷t × Act Exposurei × dt + 𝛾′ Xi,t−1 + 𝛼j,t + 𝜀i,t

(19)

Causal Effect of Employee Social Capital: matched sample

Matched sample Treated

(Obs. = 635) Control (Obs. = 635) R&D 0.021 0.023 Book leverage 0.107 0.109 ln(1+Asset) 12.347 12.304 Volatility 0.142 0.148 Age 29.191 30.710 ln(1+Emp) 5.572 5.565 Y = Tobin’s q Act Exposure 6.507*** 6.531*** (1.356) (1.353) Act Exposure × Post -4.651*** -4.409***

(1.232) (1.140)

Controls Y Y

Including 2016 N Y

Obs. 3,541 4,811

(20)

Causal Effect of Employee Social Capital: subsamples

Table 6 Panel C Y = Tobin’s q

Act Exposure 8.350*** 8.190*** 6.362*** (1.535) (2.232) (1.363) Act Exposure × Post -6.211*** -6.376*** -4.760***

(1.407) (2.046) (1.196)

Subsample drop broader media and public sectors

drop supplier and customer industries of media & public

drop observations with zero exposure

Controls Y Y Y

Obs. 3,464 2,686 3,344

Adj. R2 _0.251 _0.222 _0.234

(21)

Table 6 Panel D Y = Tobin’s q

Act Exposure 8.081*** 5.532*** (1.527) (1.601)

Act Exposure × Post -5.390*** -4.218*** (1.306) (1.482)

Entertainment Expense 0.531*** 0.175 (0.189) (0.262)

Entertainment Expense × Post -0.322** -0.153 (0.163) (0.223)

Act Exposure × Entertainment Expense 10.183* (5.925)

Act Exposure × Post × Entertainment Expense -5.114 (5.216)

Controls Y Y

Entertainment expense captures excess expenses that are viewed as graft and bribery (Cai, Fang, and Xu 2011; Kang, Kim and Kim 2020)

-> Negative effect on firm performance is due to a reduction in the value of ESC, not bribery.

(22)

Stock Market Reaction to the Court Ruling on the Act

Table 7 Below Median Above Median Above − Below Correlation Coefficient

CAR (CAPM-adjusted)

[-3, 3] 0.41% -0.61%** -1.02%** -0.076**

CAR (size-adjusted)

[-3, 3] 0.52% -0.43%** -0.95%** -0.065** The announcement date is July 28, 2016 when the Act is ruled constitutional.

(23)

Benefit of Media Connections

Act Exposure = ESC₂₀₁₅𝑀𝑒𝑑𝑖𝑎/ESC

Media & firm value: Tetlock 2007; Dougal et al 2012; Gurun, Butler 2012; Ahern, Sosyura 2014 Media connections promote news coverage of the firm and coverage with a positive tone.

Table 8 Y = Tobin’s q ln(1+Media Coverage) Positive Media Coverage Ratio Act Exposure 8.016*** 4.495*** 0.437**

(1.591) (1.564) (0.180) Act Exposure × Post -5.655*** -2.991** -0.305* (1.398) (1.445) (0.172)

Controls Y Y Y

Obs. 3,778 3,775 3,775

Adj. R2 _0.242 _0.343 _0.164

(24)

Benefit of Public Sector Connections

Act Exposure = ESC₂₀₁₅𝑃𝑢𝑏𝑙𝑖𝑐/ESC

Schoenherr (2019, JF): Korean public officers allocate procurement contracts to firms with a connected CEO.

Public sector connections help firms obtain more government contracts.

• One std dev ↑in ESC 𝑃𝑢𝑏𝑙𝑖𝑐/ESC → contract volume ↑ 6.8% before the Act, only 3.4% after. Table 8 Y = Tobin’s q ln(1+Proc. Contracts) ln(1+Amt of Proc.

Contracts) Act Exposure 6.181** 3.756*** 19.837***

(2.414) (1.111) (5.295) Act Exposure × Post -4.782** -1.878** -9.700**

(1.981) (0.839) (4.443)

Controls Y Y Y

Obs. 3,778 3,775 3,775

(25)

Connections with Investment Banking Industry

Examine the information sharing channel using detailed data on bond issuance and networks Issuers with more ESCI−bank have better information sharing with underwriters and face lower at-issue bond spreads (Leland and Pyle 1977)

• One std dev ↑in ESC𝐼−𝑏𝑎𝑛𝑘 → Bond spread reduces by 7.3 basis points.

• Effect is mainly driven by ESCI−bank of issuer’s rank-and-file employees.

• A likely motivation for i-banks to remember issuer’s rank-and-file employees is to acquire information in the due diligence investigation.

• Effect stronger for innovative issuers where information is likely more asymmetric. Table 9 Y = Tobin’s q ROA Sales

Growth

At-issue Bond Spreads ln(1+ESC I−bank) 1.461*** 0.051*** 0.124*** -0.454**

(0.241) (0.014) (0.038) (0.216)

Controls Yes Yes Yes Yes

FE Ind × Year Ind × Year Ind × Year Ind

Obs. 5,340 5,340 5,340 480

(26)

Conclusions

• We provide causal evidence that a firm’s employee social capital is valuable to

performance.

– Unique data: to measure firm-level employee social capital

– Direction of connections matter: who knows you > who you know

– Connections by all levels of employees matter: top, lower-level manager, rank-and-file

(27)

(28)

Data: employee-level connections

Table 1 Panel B N Mean Median

App-users

In-degree 119,423 26.329 11

Out-degree 119,423 56.916 17

Total degree 119,423 83.244 30

Non-app-users

In-degree = Total degree 896,600 4.820 2

All public firm employees (app-users + non-app-users)

In-degree 1,016,023 7.348 2

Total degree 1,016,023 14.038 2

In-degree by employee job levels

Executives 98,864 12.909 2

Non-executive managers 581,094 8.198 2 Rank and file employees 336,065 4.242 2

(29)

App-users vs. Non-app-users

• Results are not driven by app-users being more tech savvy and work for productive firms. Table 4 Panel B ESC in-degree

of non-app-user employees

ESC out-degree to app-users Y = Tobin’s q ROA Sales

Sales Growth ln(1+ESC) 0.427*** 0.029*** 0.135*** 0.089* 0.005* 0.006

(0.110) (0.009) (0.029) (0.047) (0.003) (0.010)

(30)

Nonreciprocal In-degree vs. Nonreciprocal Out-degree

Table 4 Panel B ESC in-degree nonreciprocal

ESC out-degree nonreciprocal Y = Tobin’s q ROA Sales

Sales Growth ln(1+ESC) 0.517*** 0.031*** 0.128*** 0.026 0.004* 0.005

(0.115) (0.009) (0.029) (0.025) (0.002) (0.006)

(31)

Alternative Measures of Employee Social Capital

Table IA.2 Panel A ESC in-degree (Who Knows You) ESC out-degree (Who You Know) Y = Tobin’s q ROA Sales

Sales Growth ln(1+ESC: Excl. 0.389*** 0.020*** 0.093*** 0.050* 0.003 0.002 Sales) (0.084) (0.007) (0.024) (0.028) (0.002) (0.006) ln(1+ESC: Single 0.361*** 0.018** 0.102*** -0.025 -0.002 0.006 Count) (0.093) (0.007) (0.022) (0.028) (0.002) (0.007) ln(1+ESC: Sum) 0.251*** 0.016*** 0.067*** -0.004 0.002 0.007 (0.070) (0.006) (0.017) (0.022) (0.002) (0.005) Controls Y Y Y Y Y Y

FE Ind×Year Ind×Year Ind×Year Ind×Year Ind×Year Ind×Year

• ESC: Excl. Sales excludes customer-facing employees who perform sales functions

(32)

Subsample Analysis

Table IA.2 Panel B ESC in-degree (Who Knows You) ESC out-degree (Who You Know) Y = Tobin’s q ROA Sales

Sales Growth

[Excluding Top 20 Companies Most Wanted by University Students]

ln(1+ESC) 0.329*** 0.021*** 0.083*** 0.043 0.004* 0.003 (0.090) (0.008) (0.021) (0.030) (0.002) (0.007)

[Excluding Financial Sector]

ln(1+ESC) 0.325*** 0.020*** 0.100*** 0.042 0.004* 0.004 (0.092) (0.008) (0.024) (0.031) (0.002) (0.007)

[Excluding Top 3% Companies based on Total Assets]

ln(1+ESC) 0.350*** 0.020*** 0.079*** 0.044 0.004* 0.003 (0.091) (0.008) (0.022) (0.030) (0.002) (0.007)

FE Ind×Year Ind×Year Ind×Year Ind×Year Ind×Year Ind×Year

• Follow Cohen, Frazzini, and Malloy (2010) to drop firms popular among skilled employees

(33)

Propensity Score Matching

Table IA.2 Panel C Tobin’s q ROA Sales Growth Number of Matches Above-Median – Below-Median 0.203*** 0.014*** 0.065*** 2,456

(ESC in-degree) (0.047) (0.004) (0.016)

Above-Median – Below-Median 0.025 0.005 -0.002 2,237 (ESC out-degree) (0.047) (0.004) (0.015)

(34)

Cross-Sectional Analysis

Table IA.3 Tobin’s q ROA Sales

Sales Growth

[By labor intensity]

Above Median Below Median

ln(1+ESC) 0.438*** 0.037*** 0.077*** 0.197* 0.005 0.103*** (0.140) (0.013) (0.028) (0.110) (0.008) (0.038)

[By organization capital]

Above Median Below Median

ln(1+ESC) 0.421*** 0.032*** 0.074* 0.190** 0.012 0.131*** (0.146) (0.011) (0.041) (0.095) (0.008) (0.030)

[By Industry Cluster]

Outside Industry Clusters Within Industry Clusters ln(1+ESC) 0.420*** 0.020** 0.134*** 0.207* 0.024** 0.049

(0.136) (0.010) (0.038) (0.123) (0.012) (0.033)

• Effects are stronger on Tobin’s q and ROA for firms with higher

− labor intensity (EMP/AT): reliance on labor in production

(35)

The Act and the Formation of Employee Connections

Before the Act: 2015

After the Act: 2018

• ESC Act/ESC decreases by 7.7% relative to sample mean.

(36)

Placebo Test: randomization of exposure to the Act

(37)

Causal Effect of Employee Social Capital: interact controls

Table IA.7 Y = Tobin’s q

Act Exposure 7.380*** 7.380***

(1.319) (1.318) Act Exposure × Post -5.847*** -5.544***

(1.175) (1.100)

Controls × Post Y Y

Controls Y Y

Including 2016 N Y

Obs. 3,778 5,101

(38)

Firm Employees in the Network App-user Employees in the Network ESC in-degree ESC out-degree ESC total degree 1 2 1 1 3 2.5 2 2 0 1 - 1 3 1 1 1 2 3