1. Paid for connections or too connected to be good
 Paid for connections or too connected to be good? Social Networks and Executive and Non-Executive Director Compensation  
Joanne Horton, Yuval Millo London School of Economics George Serafeim Harvard Business School

2. Motivation: role of directors in current capitalism
Is there a demand for people that are better ‘connected’?
Most economic based research views social connections as a mechanism for rent extraction (Larcker et al. 2005; Barnea and Guedj 2007; Brown et al. 2008), following the managerial power paradigm of Bebchuk
Social capital and management theory suggests social connections can potentially be an important strategic tool for competitive advantage (Burt 2005)

3. Research Questions
Do social network measures exhibit an association with executive and non-executive directors’ compensation?
Do firms with better connected directors have better or worse future performance?

4. Directors: monitors and advisors
Corporate directors monitor management on behalf of the shareholders: such monitoring may reduce agency costs and improve firm performance (Fama and Jensen 1983)
Directors may provide resources to the firm. For example board members can serve as ‘boundary spanners’ (Zahra and Pearce 1989) providing access to communication channels with the external environment (Pfeffer and Salancik 1978) and thereby provide their firm with access to new information (Allen 1974, Burt 1979). These resources can reduce dependency between the firm and external contingencies (Pfeffer and Salancik 1978), enhance organizational legitimacy (Zahra and Pearce 1989), reduce uncertainty (Pfeffer 1972), lower transaction costs (Williamson 1984), and thereby impact firm decision making (Mizruchi 1996) and improve a company’s performance and solvency (Stokman et al. 1985, Mizruchi and Stearns 1988).

5. Predictions for director compensation
Directors receive higher compensation the better monitoring or advising services they provide
Since monitoring role is performed by NEDs better connected executives will have higher compensation
For NEDs the relation with pay is not clear. Better connected NEDs may or may not be better monitors
Better connected directors may create additional agency costs through leakage of proprietary information
Less connected directors may demand higher pay to compensate for inability to diversify their human capital risk

6. Network measures Directors Degree Closeness Dyadic Constraint
Normalized Aggregate
Directors Degree Closeness Dyadic Constraint A B C D E F G H I J K L
Network measures G A B C H E D F I K L J

7. Sample: companies and directors
All companies listed at the London Stock Exchange (LSE) and the Alternative Investment Market (AIM) between
4,278 firms, 31,495 directors and 111,114 directorship-year pairs
The sample includes 3,477 firms, 22,585 directors and 79,439 directorship-year pairs
Total compensation for each director is the sum of salary, bonus and other benefits (also examine stock options in section 5.3.1)

8. Results (in a nutshell)
Executive directors are rewarded for being socially connected
Non-executive directors are rewarded for having constrained social connections
The predicted component from high executive connectedness and low non-executive connectedness is positively associated to future firm performance

9. Research design: network measures and compensation
Compensation = f(Social networkit, Controlsijt)
Compensation is the compensation for director i, directorship j at year t
Social networkit is either the closeness or dyadic constraint measure for director i at year t.
We cluster standard errors at the company level to mitigate serial and cross-director correlation within a firm (we tried several alternative estimation methods in 5.3.5)

10. Association between compensation and closeness
Dependent variable: Compensation
CEO
CFO
Executive
Chairman
Non-Executive
Intercept
-86.71
-6.42
(-1.71)
(-3.41)
(-2.38)
(-3.28)
Closeness
354.47
148.57
319.05
-25.62
(3.26)
(2.73)
(2.63)
(-2.52)
(-3.03)
MTB
-0.34
0.16
-0.47
-0.37
-0.08
(-0.27)
(0.24)
(-0.36)
(-1.06)
(-1.85)
ROA
-51.90
-35.45
-74.43
-5.25
-2.10
(-4.90)
(-4.85)
(-3.84)
(-1.74)
(-1.95)
Firm size
98.12
58.91
75.09
15.82
2.82
(12.73)
(16.84)
(7.28)
(9.01)
Stock return
-17.23
-13.02
-15.41
-3.15
-1.25
(-3.96)
(-3.77)
Growth
-14.87
-7.11
-5.89
-3.52
-0.65
(-2.26)
(-0.68)
(-1.34)
Board size
45.57
31.34
45.30
31.95
8.34
(1.39)
(1.98)
(1.26)
(4.52)
Director tenure
60.66
66.31
55.26
26.21
4.29
(6.38)
(10.90)
(6.11)
(9.82)
Male
-27.81
13.30
64.86
15.49
1.52
(0.96)
(3.66)
(2.10)
Index, Industry and Year f.e.
Yes N
7,940
6,031
7,889
9,169
26,839
Adj-R squared
54.8%
62.9%
40.6%
29.1%
22.5%

11. Association between compensation and dyadic constraint
Dependent variable: Compensation
CEO
CFO
Executive
Chairman
Non-Executive
Intercept
-71.37
-70.61
-15.77
(-0.54)
(-1.03)
(-2.31)
(-4.77)
(-4.09)
Dyadic Constraint
-81.92
-62.23
74.41
53.93
9.86
(-1.85)
(-1.67)
(1.05)
(6.94)
(7.51)
MTB
-0.34
0.17
-0.44
-0.08
(-0.28)
(0.25)
(-0.34)
(-0.98)
(-1.93)
ROA
-51.11
-35.87
-74.98
-4.83
-2.02
(-4.79)
(-4.86)
(-3.86)
(-1.63)
(-1.87)
Firm size
99.52
59.09
77.99
16.11
2.84
(12.95)
(16.72)
(7.36)
(6.60)
(9.15)
Stock return
-17.77
-13.07
-15.76
-3.25
-1.29
(-4.08)
(-4.87)
(-3.45)
(-1.96)
Growth
-14.93
-7.41
-6.43
-3.86
-0.67
(-2.25)
(-1.77)
(-0.74)
(-1.48)
(-1.38)
Board size
23.86
11.62
81.57
42.53
10.22
(0.59)
(0.49)
(1.73)
(4.17)
(5.40)
Director tenure
59.34
66.04
54.49
24.69
4.23
(6.24)
(10.91)
(6.03)
(6.95)
(9.72)
Male
-27.59
13.47
65.99
13.55
1.56
(-0.53)
(0.96)
(3.73)
(1.04)
(2.14)
Index, Industry and Year f.e.
Yes N
7,940
6,031
7,889
9,169
26,839
Adj-R squared
54.8%
62.9%
40.4%
29.5%
22.8%

12. Within firm-year association between compensation and network measures
Dependent variable: Compensation
Executive directors
Non-Executive directors
Intercept
-58.96
-58.70
-60.96
-60.88
-23.37
-21.69
-24.78
-23.96
(-6.25)
(-6.23)
(-6.33)
(-6.31)
(-12.65)
(-11.86)
(-12.90)
(-12.56)
Closeness
(2.30)
(2.97)
(-8.54)
(-5.65)
Dyadic constraint
-99.93
38.63
32.35
(-1.41)
(-1.65)
(8.68)
(6.90)
Director tenure
78.34
78.47
78.61
78.63
18.44
18.54
18.31
18.34
(13.21)
(13.06)
(13.07)
(13.05)
(13.09)
(12.86)
(12.87)
Male
14.04
14.46
12.77
12.85
8.62
8.49
9.17
9.07
(1.69)
(1.74)
(1.53)
(1.54)
(5.11)
(5.02)
(5.34)
(5.28)
Busy Director
-21.77
-5.47
-6.18
-2.91
(-0.88)
(-0.21)
(-5.52)
(-2.71)
Firm x Year f.e.
Yes N
23,648
42,887
Adj R-squared
18.5%
18.2%
19.4%
18.8%
# of firms
2,181
3,032

13. Research design: network measures and firm performance (1)
We estimate the model for (estimation period) and we test the predictions in period (test period).
We also used as the test period and as the estimation period. The results were qualitatively similar to the ones reported here.
After estimating the predicted component for each directorship-year we average this variable by firm and then over the years used for estimation to construct a firm level variable
Intuition: if better connected executives are better advisors and less connected NEDs better monitors then future performance should increase in the predicted component of compensation

14. Research design: network measures and firm performance (2)
Then we include this firm-level variable in a regression where the dependent variable is future performance.
Performanceit = f(Pcompit-1, Performanceit-1, Controls
Where Performanceit is average stock return, MTB, ROA or sales growth for firm i over and Performanceit-1 is average stock return, MTB, ROA or sales growth for firm i over
By including lag performance we control for other potential omitted variables that can affect performance persistently and for the endogeneity of the director selection (firms with low or high performance may systematically choose directors with high social capital)

15. Association between predicted compensation from closeness and future performance
Dependent Variables:
Stock Return
MTB
ROA
Sales growth
Intercept
0.098
0.132
-0.709
-0.612
-0.092
-0.097
-0.002
0.075
(1.74)
(2.33)
(-1.79)
(-1.57)
(-2.09)
(-2.22)
(-0.02)
(0.50)
Lag Return
-0.016
-0.015
(-0.52)
(-0.50)
Lag MTB
0.557
(15.88)
(15.98)
Lag ROA
0.508
0.509
(13.54)
(13.52)
Lag 3-year Growth
0.282
0.284
(8.21)
(8.29)
Mean Predicted SN Comp
0.116
0.234
-0.014
0.196
(2.93)
(1.45)
(-1.35)
(2.89)
Sum Predicted SN Comp
0.021
-0.001
0.030
(2.84)
(2.38)
(-0.70)
(2.30)
Firm size
-0.003
0.062
0.064
0.013
0.014
0.018
0.015
(-0.21)
(-0.45)
(2.03)
(2.07)
(7.65)
(7.92)
(1.38)
(1.13)
0.032
(5.24)
(5.25)
Board size
-0.081
-0.099
-0.164
-0.253
-0.025
-0.028
-0.049
(-2.51)
(-2.88)
(-0.99)
(-1.47)
(-2.49)
(-2.35)
(-0.44)
(-0.75)
% Outsiders
-0.138
-0.135
0.231
0.242
-0.006
-0.153
-0.145
(-2.06)
(-2.00)
(0.68)
(0.71)
(-0.31)
(-1.32)
(-1.24)
% Busy directors
-0.056
-0.048
0.076
0.040
-0.007
-0.010
-0.219
-0.192
(-1.09)
(-0.95)
(0.35)
(0.19)
(-0.68)
(-1.99)
(-1.77)
Industy f.e.
Yes N
1340
1284
1300
1161
Adj-R squared
10.9%
10.6%
35.6%
35.8%
45.1%
45.0%
23.7%
23.3%

16. Association between predicted compensation from dyadic constraint and future performance
Dependent Variables:
Stock Return
MTB
ROA
Sales growth
Intercept
0.019
0.040
-0.731
-0.763
-0.098
-0.115
-0.191
-0.140
(0.29)
(0.66)
(-1.66)
(-1.72)
(-2.20)
(-2.66)
(-0.99)
(-0.77)
Lag Return
-0.023
-0.021
(-0.74)
(-0.70)
Lag MTB
0.558
(15.83)
(15.88)
Lag ROA
0.509
(13.51)
(13.55)
Lag 3-year Growth
0.287
0.286
(8.34)
Mean Predicted SN Comp
0.037
0.032
0.000
0.075
(2.37)
(0.44)
(0.10)
(2.73)
Sum Predicted SN Comp
0.011
0.014
0.002
0.020
(2.84)
(0.69)
(1.56)
(2.65)
Firm size
0.001
0.058
0.061
0.015
0.022
(-0.07)
(0.12)
(1.74)
(1.79)
(7.82)
(8.32)
(1.54)
(1.55)
(5.27)
(5.30)
Board size
-0.044
-0.065
-0.113
-0.133
-0.027
0.039
-0.002
(-1.33)
(-2.05)
(-0.69)
(-0.80)
(-2.6)
(-2.65)
(0.63)
(-0.02)
% Outsiders
-0.127
-0.124
0.241
0.250
-0.005
-0.119
(-1.87)
(-1.83)
(0.72)
(0.74)
(-0.27)
(-0.10)
(-1.00)
(-0.96)
% Busy directors
0.025
0.027
0.197
0.212
-0.012
-0.006
-0.063
-0.068
(0.54)
(0.56)
(0.94)
(1.01)
(-0.83)
(-0.44)
(-0.66)
Industy f.e.
Yes N
1340
1284
1300
1161
Adj-R squared
10.5%
10.7%
35.5%
45.0%
45.1%
23.6%
23.7%

17. Including stock option compensation
CEO
CFO
Executive
Chairman
Non-Executive
Dependent variable: Total compensation
Closeness
957.91
-75.14
(5.09)
(3.79)
(4.65)
(-1.29)
(-4.15)
Dyadic constraint
116.89
23.55
(-3.98)
(-3.20)
(-1.31)
(5.26)
(7.83)
Dependent variable: % Long-term incentives
0.83
0.68
0.71
(6.02)
(4.79)
(4.45)
-0.20
-0.25
-0.16
(-4.20)
(-3.82)
(-1.88)
Dependent Variable:
Stock Return
MTB
ROA
Sales growth
Mean Predicted Compensation (closeness)
0.034
0.038
0.000
0.090
(1.98)
(0.53)
(0.11)
(2.83)
Mean Predicted Compensation (dyadic constraint)
0.010
0.017
0.001
0.021
(2.16)
(0.85)
(2.50)
Sum Predicted Compensation (closeness)
0.008
0.016
0.015
(2.13)
(1.64)
(0.47)
(2.43)
Sum Predicted Compensation (dyadic constraint)
0.003
0.006
(2.32)
(0.29)
(1.61)
(2.64)

18. Implications
Better connected executive (non-executive) directors get higher (lower) compensation
Compensation that directors receive for the resources and monitoring they provide is associated with higher future stock market performance and sales growth
Market for directors works reasonably well without any regulatory intervention. Strict rules and guidelines about director selection and pay might distort the current equilibrium