Dou Et Al (2015) - Should Independent Directors Have... - FM
Dou Et Al (2015) - Should Independent Directors Have... - FM
Dou Et Al (2015) - Should Independent Directors Have... - FM
Shareholder advisory firms and regulators increasingly view lengthy experience as a negative
attribute for independent directors. Independent directors with extended tenures, subsequently
referred to as experienced directors, are seen as ineffective in fulfilling the roles of monitoring
management and setting firm strategy. The Financial Reporting Council in the United Kingdom
does not consider a director who has been on the board for longer than nine years to be an
independent director.1 The implication is that directors become entrenched and aligned with
managers after an extended period and are, therefore, unable to monitor them adequately.
In the United States, the National Association of Corporate Directors (NACD), an advisory
organization that publishes best practice procedures in boardrooms, recommends tenure limits of
10 to 15 years when evaluation procedures are not in place (NACD, 2005). Director term limits
have also been seen as a way to bring fresh thinking and ideas onto the board, avoiding stagnation
in strategic decision making (Young, 2011). A survey by Heidrick and Struggles (2007) polling
2000 of the largest US firms found that 22% of the 660 respondents imposed term restrictions
on directors, more than doubling the 9% in 2001 when the survey was previously conducted.
Despite concerns that experienced directors exacerbate the manager-shareholder agency
problem (Jensen and Meckling, 1976), nearly 60% of the Standard & Poors 1500 (S&P 1500)
firms in our sample from 1998 to 2013 have an independent director who has been on the
board for longer than 15 years. According to a recent analysis by Guaranteed Minimum Income
(GMI) Ratings for the Wall Street Journal, among Russell 3000 companies, nearly 34% of
We would like to thank Renee Adams for her extensive guidance on this paper. We would also like to thank an anonymous
referee, Raghu Rau (Editor), Ron Masulis, Peter Pham, Daniel Metzger, Fariborz Moshirian, Bohui Zhang, Sarah Azzi,
Rasha Ashraf, Matthias Thul, Jing Xu, and participants at the 2013 FIRS Conference and the 2013 FMA Annual Meeting
for their valuable suggestions.
Ying Dou is a Ph.D. Candidate in the School of Banking and Finance, University of New South Wales in Sydney, New
South Wales, Australia. Sidharth Sahgal is the Managing Director at Macrovue Pty Ltd. Emma Jincheng Zhang is a Ph.D.
Candidate in the School of Banking and Finance, University of New South Wales in Sydney, New South Wales, Australia.
1
UK Corporate Governance Code (June 2010). Publicly listed companies on the London Stock Exchange are required to
state how they have complied with this code.
584
the total independent directors have had a tenure of 10 years or longer.2 There are at least three
reasons why having experienced directors on the board may actually be optimal for firms. First,
given their long tenure, most experienced directors have worked with multiple CEOs, which
should help them better assess the ability of the current chief executive officer (CEO). Only
10% of directors with tenure longer than 15 years in our sample were hired during the current
CEOs term. In addition, over the course of their term, experienced directors will have built up
significant financial stakes in the company, aligning their interests with that of the shareholders.
A 1% increase in underlying stock increases wealth by $13,318 for a median director with a
tenure greater than 15 years, but only $4,382 for those with a shorter tenure. Moreover, just as a
long, successful tenure is seen as increasing a CEOs bargaining power (Hermalin and Weisbach,
1998), longer tenure periods should also buttress the position of the directors, helping them to
balance the CEOs influence when it is time to make decisions in the boardroom.
In this paper, we examine whether tenure limits are an optimal solution to the potential agency
problems caused by experienced directors using an unbalanced panel data set of S&P 1500
firms from 1998 to 2013. To isolate the contribution of experienced directors, we define an
experienced director as one with more than 15 years of experience. We then use the proportion
of these experienced directors on the board as the key explanatory variable in regressions where
firm outcomes including CEO compensation, CEO turnover-performance sensitivity, earnings
restatements, acquisition decisions, and acquisition performance are the dependent variables.
Using an aggregate measure, like the median or average, masks the distribution of tenure
among directors. The choice of using a fixed number to define an experienced director is
crucial to the analysis as it is hard to glean any insight regarding the heterogeneous effects
of experienced directors without a concrete definition of experienced. The decision to use
15 years as the threshold is supported by our director-level piecewise linear regression analysis,
though we obtain quantitatively similar results when we use 12 years as the threshold. Also,
as mentioned previously, the tenure limit recommended by NACD (2005) is 10 to 15 years.
Therefore, we choose the upper bound of this range to examine whether independent directors
with tenures beyond this range are really as problematic as the recommendation implies.
The efficacy of different forms of board structure has been studied extensively in the financial
literature. However, despite widespread recommendations and legislation on term limits, there
is limited empirical evidence as to how director tenure impacts board decision making. One
exception is a study by Vafeas (2003), which uses one year of data from 1994 and splits
directors according to their relative tenure within the firm. Another set of related studies (Core,
Holthausen, and Larcker, 1999; Landier et al., 2012; Coles, Daniel, and Naveen, 2014;) considers
whether directors who were hired after the current CEO began their term are coopted by these
CEOs. A longer tenure could be a proxy for the fact that the director was hired before the
current CEO. However, the results in this paper are not driven solely by the hiring effect. This
is demonstrated by controlling for the proportion of directors hired after the current CEO was
appointedr and also by rerunning the analysis in a subsample of firms where the CEOs tenure
is greater than that of the senior-most independent director.
The analysis consists of five main parts. First, we use director-level data to examine whether
experienced directors commitments to their roles change as tenure increases. We find that far
from shirking their responsibilities, these directors are significantly less likely to miss board
meetings. Even though participation in a monitoring committee requires extra effort in terms
of time and responsibility, experienced directors are significantly more likely to be members of
2
The 40-Year Club: Americas Longest-Serving Directors, Wall Street Journal, July 16, 2013, http://online.
wsj.com/news/articles/SB10001424127887323664204578607924055967366.
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
585
the major board monitoring committees. This membership could be attributed to experienced
directors having deeper knowledge of the firm and its operations relative to newer directors, but
it nevertheless signals a continued commitment to their role.
Next, we examine the determinants of a higher proportion of experienced directors on the
board. Our results suggest that across firms in the same industry, smaller, older, more stable, and
successful firms with older and longer tenured CEOs are more likely to have experienced directors. However, when examining within-firm variations using firm fixed-effects, many of these
relationships change. For example, a larger proportion of experienced directors is now positively
related with firm size, which is intuitive since firms tend to increase in size over the sample period.
Third, we examine the role of experienced directors in three different firm monitoring outcomes
using panel regressions on firm-level data. Concerns regarding endogeneity and selection bias do
not easily allow for causal inference from such regressions. To alleviate some of these concerns, we
use firm and director fixed-effects along with an extensive set of control variables in our analysis.
When examining total CEO pay, our results indicate that an increase of one experienced director on the compensation committee decreases the average pay of the CEO by about 3.2%. CEO
turnover is also shown to be more sensitive to stock performance when there is a higher proportion
of experienced directors. We also find that the presence of one additional experienced director
on the audit committee can reduce the probability of intentional misreporting by approximately
25% relative to those firms where there are no experienced directors on the audit committee.
In addition, we determine whether the presence of experienced directors affects strategic
decisions in the firm. We find that firms with a higher proportion of experienced directors make
fewer acquisitions, thereby limiting the empire building aspirations of the CEO (Roll, 1986).
An examination of whether the quality of acquisitions made is higher, using announcement
day abnormal stock returns as a proxy for quality, also indicates that a higher proportion of
experienced directors on the board is related to increased abnormal returns.
Finally, we conduct several robustness checks to see whether our results are driven by certain
alternative explanations. The concerns we consider include self-selection concerns that directors
are only willing to stay at better firms, and that firms are only willing to retain better directors over
time. We also consider the possibility that a directors tenure is merely a proxy for their general
experience and does not have to be accumulated by staying at the same firm for an extensive
period. Overall, our results suggest that none of these concerns can fully explain our findings.
This paper falls into the category of research that examines the impact of heterogeneity
among directors on board decision making. Initial studies concerning board structure focused
on the number of independent directors on the board. Studies suggest that CEO dismissal is
more sensitive to performance in firms with a higher number of independent directors (e.g.,
Weisbach, 1988; Kaplan and Minton, 2012), but independent boards have not been able to
curtail excessive CEO pay. Guthrie, Sokolowsky, and Wan (2012) find that the requirement
to have all independent directors on the compensation committee actually increased CEO pay
after the passage of the Sarbanes-Oxley Act (SOX). Given the lack of clear evidence regarding
the benefits of independent directors, research has focused on their heterogeneity. Studies have
examined the effect of directors who are foreign (Masulis, Wang, and Xie, 2012), bankers (Guner,
Malmendier, and Tate, 2008), female (Adams and Ferreira, 2009), venture capitalists (Baker
and Gompers, 2003), CEOs in other firms (Fahlenbrach, Low, and Stulz, 2010), and politically
connected (Goldman, Rocholl, and So, 2009). Following these studies on director heterogeneity,
this paper uses extended tenure as a distinguishing attribute among independent directors.
This paper makes four contributions. First, we directly address the question of implementing
director term limits, which has been suggested as a way to infuse fresh blood onto a board and
avoid complacency. Our results indicate that such a policy would be short-sighted as experienced
586
I. Data Analysis
A. Data Description
For detailed information on director attributes, we use the RiskMetrics (formerly IRCC)
database, which provides data from 1996 to 2013. Only data after calendar year 1999 are
included because there is incomplete information on committee membership and leadership
prior to that year. The year variable is adjusted to denote the fiscal year rather the year of the
Annual Meeting. This allows for matching between the RiskMetrics data and the accounting
data from Compustat. The final sample includes data for fiscal years 1998-2013. The Centre
for Research in Security Prices (CRSP) is used for stock returns and Execucomp for information
on CEO pay and attributes like age and tenure. We use data from Execucomp (the CEOANN
flag) to identify CEOs. In the final sample, firms are only included when they have data listed
in each of the RiskMetrics, Compustat, CRSP, and Execucomp databases. There are a total of
19,020 firm-year observations, 159,580 director-year observations, and 139,963 independent
director-year observations after dropping observations without complete matching data in all of
the databases. An independent director is one who is marked as independent by RiskMetrics.
We make a few adjustments to the data to correct errors in the databases. Director tenure is set
to missing if it is greater than 90 or less than zero. In addition, if a directors age is less than 21,
it is set to be missing. Approximately 10% of the observations do not include CEO age. Another
10% do not include CEO tenure in Execucomp. We manually set the CEO age and tenure values
for these observations if publicly available. We also set director ownership to missing if the
shares in the company are dual class. The year the stock of a firm is listed is used as the firm age.
B. Data Summary
Table I contains a summary of the independent director data. Each observation is representative
of one director-firm-year. As such, each director could be represented multiple times for different
firms and years. The sample is divided into four groups to better understand how the characteristics of directors evolve as tenure increases. The first group has tenure less than or equal to five
years (L5), the second group has tenure between 6 and 15 years (B6-15), the third group (L15)
includes all observations in the first two groups, and the fourth group (G15) has tenure greater
than 15 years. In addition, we also maintain a subsample of the G15 directors who are not former
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
587
Attributes
Count
Age
Tenure
Female
CEO
Poor Attendance
Former Employee
Number Boards
Hired by CEO
Three most
frequent
primary titles
(A)
(B)
(C)
(D)
(E)
(F)
(G)
All
T5
5 < T 15
T 15
T > 15
T > 15
(F) (D)
139,963
61.801
7.723
14.012%
11.769%
1.441%
0.259%
0.956
38.487%
Retired
Finl Svce
Executive
54,260
58.328
2.160
16.152%
15.557%
1.675%
0.116%
0.960
62.551%
Retired
Executive
CEO
65,889
62.842
8.491
13.829%
10.493%
1.345%
0.235%
1.007
27.176%
Retired
Finl Svce
Other
120,149
60.804
5.632
14.878%
12.813%
1.494%
0.181%
0.986
43.151%
Retired
Executive
Finl Svce
19,814
67.839
20.402
8.762%
5.348%
1.120%
0.732%
0.774
10.205%
Retired
Finl Svce
Other
18,398
67.834
7.030
20.380
14.748
8.816% 6.062%
5.391% 7.422%
1.124% 0.370%
0.000% 0.181%
0.778
0.208
10.229% 32.922%
Retired
Finl Svce
Other
44.027%
40.462%
46.598%
46.647%
6.185%
42.663%
39.753%
42.774%
35.792%
40.042%
42.495%
39.911%
42.422%
2.863%
6.630%
1.264
1.190
1.264
1.263
0.073
Committee Membership
Compensation
41.331% 36.133%
committee
Audit committee
42.387% 42.908%
Nomination
36.741% 30.982%
committee
Compensation +
1.201
1.100
Audit +
Nomination
Committee Leadership
Compensation
10.039% 5.352%
committee
Audit committee
10.202% 7.311%
Equity Ownership
% Ownership
mean
% Ownership
median
Dollar sensitivity
mean
Dollar sensitivity
median
9.378%
14.046%
14.083%
4.705%
12.079%
9.926%
11.875%
11.861%
1.935%
0.205%
0.199%
0.197%
0.198%
0.249%
0.250%
0.052%
0.074%
0.063%
0.079%
0.072%
0.088%
0.088%
0.016%
68,660
34,286
53,821
44,999
210,525
209,903
164904
5,109
1,977
7,799
4,382
13,318
13,233
8851
12.694%
588
employees. Since we argue that directors with longer tenure tend to be better at monitoring and
advising, one concern is that there is a high proportion of former employees who have better understanding of the firm within the long-tenure directors. In this case, our results are not capturing the
effect of having a longer tenure. To eliminate this concern, in all subsequent tests, we do not consider experienced directors who used to be employees of the firm. Finally, in Column (G), we report
the differences in means for each variable between directors with tenures lower than or equal to 15
years and directors with tenures greater than 15 years who are not former employees in the firm.
While some attributes in Table I show the expected correlation, like the positive correlation
between age and tenure, other attributes offer more interesting insights. Approximately 63% of
the directors in the L5 group were hired during the term of the current CEO, which decreases
to 27% for directors in the B6-15 group and to only 10% for directors in the G15 group. This
statistic weakens the claim that directors with extended tenures are entrenched with the current
CEO of the firm. On the contrary, experienced directors have worked with multiple CEOs, giving
them the opportunity to improve their monitoring skills. In addition, the evidence in Column (G)
suggests that when compared with directors with short tenures, experienced directors are less
likely to be female and tend to have fewer directorships. The differences for all characteristics
are significant at the 1% level.
To further investigate any differences between directors among different tenure groups, we
also report the three most frequently observed primary titles for each group. This information
is available in RiskMetrics for years 1998-2001 and 2007-2013. The data suggest that for all
independent directors including the experienced directors, the most common title is always
Retired. One difference between short-tenured and long-tenured directors is that while
Executive and CEO are the second and third most common titles for directors with short
tenures, they are much less common for directors with long tenures. In fact, the second most
frequently observed title for long-tenured directors is Financial Service. Given this difference,
we include control variables for these characteristics in all subsequent director-level analyses.
There is also a sizable difference between equity ownership and, consequently, monetary
incentives among directors as tenure progresses. Director ownership includes both stocks and
options, which can be exercised within 60 days of the annual meeting. Ownership increases
from 0.063% in the L5 group to 0.088% in the G15 group. Although the stakes are small in
percentage terms, the dollar amounts are high. We calculate dollar sensitivity to a 1% change in
the underlying stock price by multiplying the total number of shares and options by 1% of the
fiscal year-end price. Directors in the group G15 have a sensitivity of $13,318 to a 1% change
in the stock price, while directors with less experience have approximately a third of the dollar
sensitivity at $4,382. These differences are, once again, significant at the 1% level. A change of
approximately 20% in the stock price can lead to a change in wealth of $315,460 for the median
experienced director, which is a sizable effect.
In terms of board governance, poor attendance, an indicator variable that is set to one if the
director does not attend 75% of the board meetings, decreases from 1.68% in the L5 group to
1.12% in the G15 group. Membership in the compensation committee increases from 36% in
the L5 group to approximately 47% in the G15 group. While a similar increase is seen in the
nomination committee, there is actually a small decrease in the likelihood of membership for G15
directors on the audit committee relative to L5 directors. This is probably due to the influence of
Sarbanes-Oxley which mandated the presence of at least one director with financial expertise on
the audit committee. This regulation led firms to bring in new directors with financial expertise
onto the board to meet this requirement.
Panel A in Table II reports firm-level data from Compustat, RiskMetrics, and CRSP used
in the sample. The size of the firms, on average, is large at approximately $15.87 billion. Size
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
589
9.570
0.739
0.286
0.384
0.331
0.592
0.740
0.860
9.000
0.778
0.000
0.333
0.250
1.000
1.000
1.000
2.890
0.157
0.452
0.362
0.370
0.491
0.439
0.347
15,873
7.778
3,292
8,784
1.881
0.224
0.025
0.614
0.099
1.315
0.188
0.928
0.146
25.418
2,024
7.613
774
1,786
1.438
0.208
0.000
1.000
0.090
0.894
0.075
0.613
0.067
22.000
84,215
1.700
10,568
27,274
1.425
0.185
0.054
0.487
0.075
1.507
0.968
15.977
0.646
15.030
0.132
0.137
0.036
0.119
0.127
0.132
0.025
0.079
0.103
0.089
0.041
0.476
(Continued)
590
CEO
CEO Tenure
CEO Age
CEO Total Compensation (thousands)
CEO Total Compensation Inflation Adjusted (thousands)
Log (CEO Total Compensation Inflation Adjusted)
CEO Ownership
CEO Dollar Sensitivity (hundred thousands)
M&A
GIM
Staggered Board
Mean
Median
Std Deviation
7.824
56.229
4,794
4,616
7.802
0.032
1,362
5.000
56.000
2,534
2,368
7.771
0.031
240
7.822
7.235
10,151
10,668
1.196
0.030
12,272
9.298
0.553
9.000
1.000
2.726
0.497
PROP-G12
PROP-G9
PROP-L5
0.142
0.171
0.141
0.175
0.225
0.262
0.221
0.273
0.352
0.395
0.342
0.410
0.388
0.336
0.388
0.322
and other firm-level characteristics are comparable to other studies including Coles, Daniel,
and Naveen (2008), Faleye, Hoitash, and Hoitash (2011), and Masulis and Mobbs (2011), which
use the same database for their analysis. Panel B in Table II provides the composition of boards
and committees according to the proportion of directors with specified tenures. On average,
the proportion of independent directors with terms of 15, 12, and 9 years to the total number of
independent directors is 0.14, 0.23, and 0.35, respectively.
II. Results
A. Board-Level Governance
1. Committee Membership
Board committees are delegated with important monitoring tasks, such as setting CEO compensation, choosing and reviewing auditors, and nominating new directors. Participation in these committees means attending special meetings and spending more time on board duties. Faleye et al.
(2011) find that participation by a majority of independent directors in two or more committees
improves the quality of the boards monitoring. Given the extra commitment required for committee membership, we construe participation in a committee as a signal of engagement with the firm.
Table III contains the results for ordinary least square (OLS) regressions where the dependent
variable is an indicator variable that takes the value of one if the director is a member of any
of the three major board committees (i.e., audit, compensation, and nomination). Among all of
the independent directors, we construct five indicator variables indicating whether the directors
tenure is lower than or equal to five years, between 6 and 10 years, between 11 and 15 years,
between 16 and 20 years, or greater than 20 years. In Columns (1) to (5) of Table III, we regress
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
591
(2)
(3)
(4)
(5)
(7)
0.048
(0.000)
T6-10
(6)
0.055
(0.000)
0.041
(0.693)
T11-15
0.012
(0.000)
T16-20
0.034
(0.000)
T > 20
0.030
(0.000)
0.016
(0.199)
0.023
(0.000)
0.051
(0.000)
0.051
(0.000)
0.034
(0.000)
0.023
(0.000)
0.002
(0.613)
0.016
(0.000)
0.002
(0.451)
0.019
(0.000)
0.182
(0.000)
0.038
(0.000)
0.035
(0.000)
0.023
(0.000)
0.003
(0.506)
0.017
(0.000)
0.002
(0.467)
0.015
(0.000)
0.174
(0.000)
Tenure
0.034
(0.000)
Female Director
0.036
(0.000)
Busy Director
0.023
(0.000)
Executive
0.004
(0.408)
Largest
0.018
(0.000)
Retired
0.002
(0.463)
Financial Service
0.014
(0.001)
Board Size
0.175
(0.000)
Age_L65
0.047
(0.000)
0.034
(0.000)
0.023
(0.000)
0.000
(0.985)
0.016
(0.000)
0.003
(0.374)
0.018
(0.000)
0.178
(0.000)
0.047
(0.000)
0.035
(0.000)
0.024
(0.000)
0.001
(0.788)
0.017
(0.000)
0.002
(0.456)
0.018
(0.000)
0.181
(0.000)
0.046
(0.000)
0.035
(0.000)
0.024
(0.000)
0.001
(0.835)
0.017
(0.000)
0.002
(0.520)
0.017
(0.000)
0.182
(0.000)
0.002
(0.000)
0.035
(0.000)
0.036
(0.000)
0.024
(0.000)
0.002
(0.644)
0.019
(0.000)
0.002
(0.493)
0.015
(0.000)
0.179
(0.000)
(Continued)
592
% Independent
CEO-Chairman
CEO Age
CEO Tenure
CEO Ownership
Firm Size
Firm Age
Tobins q
R&D
Observations
Adjusted R2
Fixed-effects
(1)
(2)
(3)
(4)
(5)
(6)
(7)
0.272
(0.000)
0.037
(0.000)
0.064
(0.075)
0.015
(0.670)
0.133
(0.000)
0.013
(0.000)
0.005
(0.000)
0.000
(0.865)
0.071
(0.191)
126,207
0.112
0.277
(0.000)
0.034
(0.000)
0.063
(0.082)
0.014
(0.692)
0.134
(0.000)
0.014
(0.000)
0.005
(0.000)
0.000
(0.841)
0.074
(0.171)
126,207
0.111
0.279
(0.000)
0.035
(0.000)
0.063
(0.085)
0.014
(0.703)
0.130
(0.000)
0.014
(0.000)
0.005
(0.000)
0.000
(0.951)
0.068
(0.212)
126,207
0.109
0.280
(0.000)
0.034
(0.000)
0.064
(0.076)
0.015
(0.671)
0.132
(0.000)
0.013
(0.000)
0.005
(0.000)
0.000
(0.970)
0.065
(0.229)
126,207
0.109
Firm, Year
0.279
(0.000)
0.034
(0.000)
0.062
(0.089)
0.013
(0.719)
0.132
(0.000)
0.014
(0.000)
0.005
(0.000)
0.000
(0.960)
0.069
(0.203)
126,207
0.109
0.270
(0.000)
0.035
(0.000)
0.063
(0.080)
0.014
(0.692)
0.135
(0.000)
0.013
(0.000)
0.005
(0.000)
0.000
(0.871)
0.074
(0.175)
126,207
0.113
0.279
(0.000)
0.037
(0.000)
0.065
(0.070)
0.016
(0.643)
0.129
(0.000)
0.013
(0.000)
0.006
(0.000)
0.000
(0.925)
0.064
(0.238)
126,207
0.111
the dependent variable on these five indicator variables separately. In Column (6), we pool these
variables together, treating directors whose tenures are between 6 and 10 years as the omitted
group. Finally, in Column (7), we use tenure directly as our key independent variable. All of the
regressions include firm and year fixed-effects.
The results suggest that tenure has a distinct impact on committee memberships. Directors
with tenures of less than or equal to five years are significantly less likely to become committee
members, while directors whose tenures are between 11 and 20 years are more likely to become
committee members. The discrepancy between directors with short and long tenures is a
sensible result, given that new directors tend to be less knowledgeable of the firms operations.
Furthermore, once we pool the indicator variables together, the active committee involvement
effect disappears for directors with tenure between 11 and 15 years and only remains significant
for directors with tenure between 16 and 20 years. Interestingly, the coefficient for T > 20
appears significantly negative, suggesting that once a directors tenure exceeds 20 years,
their involvement in committees begins to diminish.3 Nevertheless, as Column (7) indicates,
when we regress the dependent variable on director tenure directly, we observe a significantly
positive relation indicating that experienced directors overall play an active role in the corporate
governance of the firm through committee memberships.
3
In untabulated results, we find that the effect of tenure remains significantly positive until it reaches 27 years.
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
593
Among the control variables, the results suggest that older directors and female directors are
more likely to be committee members. Even though busy directors tend to have time constraints,
they are also more likely to be committee members. To allow for the possibility that directors
involvement in board committees is also affected by the importance of the firm relative to their
other directorships, we control for whether the director is an executive at another firm, and
whether the firm is the largest (measured by total assets) in the directors directorship portfolio.4
The results suggest that directors are more likely to be a committee member at firms that are the
largest among all of their board seats.
As previously mentioned, we also control for whether the primary title of the director is Retired
or Financial Service. Because this information is only available in RiskMetrics for a subset
of years, we make an adjustment to these two variables to avoid losing observations. For each
director in the database, once their primary title contains Retired (Financial Service) in a given
year, we assume this director is retired (has experience in financial service) in all subsequent
years where the information is missing. The results suggest that whether a director is retired
does not affect their committee memberships, while having experience in financial service
significantly increases the likelihood that the director is a committee member. In untabulated
results, we find that this positive relation mainly exists in the audit and compensation committees,
but not in the nomination committee.
2. Attendance
Attendance at board meetings is integral to the monitoring and advising duties of a director.
Firms are required to report to the Securities and Exchange Commission (SEC) whether directors
attended less than 75% of the meetings. In cases where the director has not attended 75% of
the meetings, an indicator variable called Poor Attendance is set to one. In the sample of all
independent directors, the mean of Poor Attendance is only 1.4% suggesting that missing of
meetings is relatively rare. In this section, we examine whether attendance at board meetings
changes for directors with different tenures.
We include control variables for individual director attributes similar to the previous section
on committee membership. Adams and Ferreira (2008) show directors are more likely to attend
meetings when meeting fees are higher. Therefore, information on meeting fees and the number
of meetings from Execucomp are included as control variables. There is no data on these two
variables from the year 2007 onward, so analysis is restricted to the years 1998-2006, leaving a
total of 65,938 director-year observations.5
Table IV includes the results for OLS regressions where the dependent variable is Poor Attendance. Consistent with the evidence in Table III which suggests that new directors are less likely
to be committee members, the evidence here indicates that these directors are also more likely
to miss board meetings. In addition, even though Table III suggests that directors whose tenures
are between 16 and 20 years have more committee memberships and, as such, a higher workload,
Table IV implies that they are less likely to miss board meetings. The coefficient for these
directors in Column (4) is 0.6%. Given the mean of Poor Attendance is 1.4%, G15 directors
are 42.9% less likely to miss more than 75% of board meetings. Overall, the results indicate that
director commitment to attending board meetings does not wane as their tenure increases.
The control variables in this table, in general, have less predictive power than those in
Table III. Nevertheless, the results suggest that directors are more likely to miss board meetings
4
5
This indicator variable is also equal to one if the director has only one directorship.
As a robustness check, we perform our analysis using data for all years, without the meeting fees and number of meetings
variables, and the results are quantitatively similar.
594
0.061
(0.000)
T6-10
(2)
(3)
(4)
(5)
0.028
(0.026)
0.021
(0.236)
T11-15
(6)
T16-20
0.015
(0.589)
0.003
(0.885)
0.036
(0.091)
0.003
(0.923)
0.023
(0.107)
0.022
(0.224)
0.075
(0.003)
0.067
(0.009)
0.019
(0.290)
0.006
(0.935)
0.053
(0.390)
0.159
(0.002)
0.007
(0.655)
0.025
(0.177)
0.075
(0.003)
0.060
(0.021)
0.018
(0.329)
0.004
(0.958)
0.047
(0.447)
0.149
(0.004)
0.054
(0.009)
T > 20
Tenure
Age_L65
0.008
(0.580)
Female Director 0.024
(0.187)
Busy Director
0.075
(0.003)
Executive
0.060
(0.020)
Largest
0.018
(0.313)
Retired
0.004
(0.954)
Financial Service 0.047
(0.440)
Board Size
0.149
(0.004)
0.024
(0.090)
0.021
(0.243)
0.075
(0.003)
0.066
(0.010)
0.020
(0.259)
0.006
(0.929)
0.053
(0.386)
0.155
(0.003)
(7)
0.057
(0.000)
0.023
(0.115)
0.022
(0.222)
0.075
(0.003)
0.067
(0.010)
0.019
(0.280)
0.005
(0.938)
0.053
(0.392)
0.158
(0.002)
0.021
(0.144)
0.023
(0.211)
0.075
(0.003)
0.067
(0.010)
0.019
(0.298)
0.005
(0.942)
0.052
(0.400)
0.158
(0.002)
0.002
(0.073)
0.015
(0.317)
0.024
(0.194)
0.075
(0.003)
0.065
(0.013)
0.017
(0.330)
0.004
(0.949)
0.050
(0.414)
0.156
(0.002)
(Continued)
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
595
(2)
(3)
(4)
(5)
(6)
(7)
0.304
(0.009)
0.070
(0.011)
0.896
(0.613)
2.903
(0.117)
0.080
(0.635)
0.048
(0.052)
0.034
(0.000)
0.003
(0.368)
0.260
(0.327)
0.027
(0.014)
2.682
(0.255)
65,938
0.040
0.307
(0.008)
0.070
(0.011)
0.871
(0.622)
2.849
(0.124)
0.084
(0.619)
0.048
(0.055)
0.035
(0.000)
0.003
(0.387)
0.259
(0.330)
0.027
(0.014)
2.671
(0.257)
65,938
0.040
0.309
(0.008)
0.070
(0.011)
0.883
(0.618)
2.850
(0.124)
0.076
(0.654)
0.048
(0.056)
0.035
(0.000)
0.003
(0.389)
0.256
(0.336)
0.027
(0.013)
2.678
(0.255)
65,938
0.040
Firm, Year
0.310
(0.008)
0.070
(0.011)
0.866
(0.624)
2.826
(0.127)
0.082
(0.629)
0.048
(0.055)
0.035
(0.000)
0.003
(0.385)
0.257
(0.333)
0.027
(0.014)
2.667
(0.257)
65,938
0.040
0.302
(0.010)
0.074
(0.007)
0.935
(0.597)
3.039
(0.101)
0.086
(0.614)
0.048
(0.053)
0.033
(0.000)
0.003
(0.369)
0.260
(0.328)
0.027
(0.015)
2.645
(0.261)
65,938
0.040
0.310
(0.008)
0.072
(0.009)
0.876
(0.621)
2.860
(0.123)
0.084
(0.621)
0.047
(0.057)
0.034
(0.000)
0.003
(0.387)
0.255
(0.337)
0.027
(0.014)
2.656
(0.259)
65,938
0.040
when they are busy or when they are executives at other firms. They are also more likely to
miss board meetings when there are more directors on the board or when the proportion of
independent directors is higher. Finally, consistent with the findings in Adams and Ferreira
(2008), the results suggest that attendance tends to increase with meeting fees.
B. Determinants of Experienced Directors
1. Firm-Level Analysis
This subsection analyzes the factors that drive the proportion of experienced directors to be
higher in certain firms. The evidence in Tables III and IV suggests that directors with tenure
greater than 15 years exhibit a significant improvement in terms of participating in committee
activities and attending board meetings. Therefore, hereafter, we consider these directors as
the experienced directors.
Panel A of Table V provides the results for regressions where the dependent variables are the
proportion of experienced independent directors. We examine the determinants of the proportion
of experienced directors on the compensation committee and the audit committee, in addition to
the proportion of experienced directors on the entire board as they are also used as key explanatory
596
PROPCC-G15
PROPAC-G15
PROPG15
PROPCC-G15
PROPAC-G15
(1)
(2)
(3)
(4)
(5)
(6)
0.069
(0.000)
0.129
(0.000)
0.002
(0.748)
0.072
(0.006)
0.070
(0.000)
0.019
(0.714)
0.022
(0.000)
0.004
(0.000)
0.006
(0.024)
0.072
(0.318)
0.348
(0.000)
0.088
(0.000)
0.174
(0.000)
0.014
(0.119)
0.038
(0.463)
0.116
(0.007)
0.047
(0.413)
0.034
(0.000)
0.006
(0.000)
0.006
(0.152)
0.133
(0.302)
0.606
(0.000)
0.033
(0.107)
0.171
(0.000)
0.003
(0.760)
0.155
(0.006)
0.129
(0.000)
0.081
(0.152)
0.029
(0.000)
0.005
(0.000)
0.006
(0.141)
0.182
(0.115)
0.336
(0.008)
0.021
(0.026)
0.053
(0.000)
0.002
(0.604)
0.047
(0.009)
0.034
(0.174)
0.030
(0.362)
0.012
(0.003)
0.003
(0.000)
0.000
(0.913)
0.030
(0.628)
0.107
(0.025)
0.014
(0.248)
0.006
(0.750)
0.000
(0.960)
0.004
(0.895)
0.047
(0.162)
0.042
(0.320)
0.007
(0.214)
0.002
(0.011)
0.001
(0.655)
0.067
(0.380)
0.065
(0.303)
0.011
(0.382)
0.022
(0.183)
0.005
(0.261)
0.072
(0.000)
0.016
(0.544)
0.094
(0.014)
0.015
(0.003)
0.001
(0.347)
0.003
(0.104)
0.022
(0.761)
0.085
(0.163)
(Continued)
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
597
PROPCC-G15
PROPAC-G15
PROPG15
PROPCC-G15
PROPAC-G15
(1)
(2)
(3)
(4)
(5)
(6)
0.063
(0.079)
16,495
0.093
0.015
(0.815)
16,291
0.043
0.148
(0.014)
16,423
0.037
0.062
(0.035)
16,557
0.051
(0.196)
16,353
0.034
(0.382)
16,484
Industry,
Year
Industry,
Year
Industry,
Year
0.578
Firm,
Year
0.488
Firm,
Year
0.475
Firm,
Year
Age_L65
Female Director
Busy Director
Executive
Largest
Retired
Financial Service
CEO-Chairman
CEO Age
CEO Tenure
CEO Ownership
Board Size
%Independent
Firm Size
Firm Age
Tobins q
R&D
(1)
(2)
(3)
(4)
0.921
(0.008)
1.227
(0.000)
1.105
(0.023)
1.332
(0.000)
0.923
(0.050)
1.381
(0.000)
0.992
(0.914)
0.945
(0.066)
1.171
(0.001)
0.995
(0.913)
1.262
(0.000)
0.916
(0.034)
1.303
(0.000)
0.960
(0.566)
1.619
(0.000)
4.133
(0.049)
0.984
(0.000)
0.612
(0.261)
1.196
(0.069)
3.260
(0.000)
1.153
(0.000)
0.986
(0.000)
0.970
(0.085)
4.378
(0.001)
0.948
(0.086)
1.210
(0.000)
0.994
(0.896)
1.255
(0.000)
0.893
(0.006)
1.311
(0.000)
0.973
(0.704)
1.686
(0.000)
2.817
(0.256)
0.982
(0.000)
0.576
(0.197)
1.051
(0.587)
3.090
(0.000)
1.112
(0.000)
0.990
(0.000)
0.987
(0.411)
3.877
(0.000)
0.952
(0.110)
1.151
(0.003)
1.011
(0.804)
1.225
(0.001)
0.876
(0.001)
1.183
(0.006)
0.654
(0.000)
1.707
(0.000)
2.878
(0.145)
0.981
(0.000)
0.683
(0.365)
1.048
(0.610)
2.869
(0.000)
1.111
(0.000)
0.990
(0.000)
0.982
(0.267)
3.975
(0.000)
(Continued)
598
(2)
(3)
1.008
(0.038)
0.985
(0.425)
Industry Volatility
Industry Return
Post-Sox
Chi2
Observations
Fixed-effects
700.971
120,334
Industry, Year
1041.352
120,334
Industry, Year
839.454
120,596
Year
(4)
0.997
(0.331)
1.070
(0.160)
0.676
(0.000)
663.821
120,596
variables in future analyses. Columns (1) to (3) use the Tobit procedure and include industry and
year fixed-effects where differences across firms, but within the same industry, are examined.
Columns (4) to (6) use OLS and examine within-firm variations using firm and year fixed-effects.
Overall, the results suggest that the proportion of experienced directors increases with CEO
age. Similarly, the coefficient for CEO tenure is positively significant in regressions with industry
fixed-effects, indicating that relative to other firms, directors and CEOs with longer tenures tend
to coexist. However, the coefficient for CEO tenure is not significant in regressions with firm
fixed-effects indicating that a higher proportion of experienced directors is not correlated with an
increase in the tenure of the CEO within the firm. This is an intuitive result given that almost 90%
of experienced directors no longer work with the CEO who was present when they were hired.
In terms of firm characteristics, the coefficient for firm size is negatively significant at the
1% level in regressions with industry fixed-effects, suggesting that relatively smaller firms are
much more likely to have experienced directors. However, when firm fixed-effects are used,
the coefficient for firm size is positively significant indicating that the proportion of experienced
directors is increasing as a firm grows larger. Publicly listed firms (in the Compustat database)
have grown in size over time. The positive relation between firm size and the proportion
of experienced directors is reflective of this fact. As expected, firm age is highly positively
significant at the 1% level in regressions using both industry and firm fixed-effects.
In terms of return on asset (ROA), the positive coefficient on five-year average ROA and the
negative coefficient on the standard deviation of ROA (calculated using annual ROA over five
years) in regressions with industry fixed-effects demonstrate the desire of experienced directors
to be at relatively more stable and profitable firms. However, in Column (4), the coefficients of
both ROA and its standard deviation show the opposite sign when firm fixed-effects are used.
This negative relation between firm performance and the proportion of experienced directors
can, to some extent, be explained by the findings in Fahlenbrach, Low, and Stulz (2013) that
outside directors have incentives to resign from firms that they anticipate will perform poorly.
Their findings imply that boards tend to experience a reduction in size during periods of poor
performance as some directors may choose to leave the firm, while few directors will be willing
to join the firm. In this case, if experienced directors, due to their long service, do not abandon the
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
599
firm like some other directors do, then this might explain the relative increase in the proportion
of experienced directors when the ROA is low.
Overall, the results confirm that firms that are smaller and older, and that have a more stable
and higher level of accounting profitability, are more likely to have experienced directors on
their boards. However, when firm fixed-effects are taken into account, it is not clear whether
experienced directors have a direct role in improving or reducing profitability.
2. Director-Level Analysis
Next, we analyze characteristics that make a director more likely to become an experienced
director. Following recent studies, such as Oyer (2008), Do, Nguyen, and Rau (2014), and
Fahlenbrach et al. (2013), we employ Cox Proportional Hazards regressions to model the tenure
of each independent director as a function of their personal attributes and certain firm-level and
macroeconomic characteristics. In our sample, each directorship is traced until either the director
leaves the firm (the event) or the firm leaves the sample. One advantage of using Cox models is
that they allow for a possibility that the event (i.e., director turnover) might never occur during
the observed period.
The results are reported in Panel B of Table V. We report hazard ratios rather than the raw
coefficients as it is easier to infer the economic significance of the explanatory variables. In
Column (1), we include only directors personal attributes. In Column (2), we add firm and
board characteristics. Since Do et al. (2014) find that macroeconomic factors can also influence
a directors career, in Column (3), we bring in the value-weighted stock returns over the past
three years at the industry level, as well as the industry return volatility. Given the potential
influence on the director labor market brought by SOX (Chen and Moers, 2014), in Column (4),
we replace year fixed-effects with an indicator variable that separates the pre-SOX and post-SOX
period.
Because our events are when directors leave the firm, variables whose hazard ratios are
greater than one are factors that negatively correlate with a directors probability of becoming
experienced. Overall, our results suggest that directors above 65, female directors, busy directors,
directors who are executives at other firms, and retired directors are less likely to stay very long.
However, if the director only has one directorship, or if the directorship of interest is the largest
among all their directorships, the director is less likely to leave.
Among the firm and board characteristics, the results suggest that directors are less likely
to become experienced at larger, younger, and more complex firms, and firms in which the
CEO has more power (captured by the CEO-Chairman duality) or the proportion of independent
directors is higher. Furthermore, there is some evidence that directors are less likely to become
experienced in volatile industries. Finally, the coefficient for the indicator variable Post-SOX is
significantly negative, implying that directors are less likely to depart after SOX.
C. Board Monitoring
1. CEO Compensation
CEO compensation has shown a sharp increase since the 1970s (Frydman and Saks, 2010).
There are two competing theories to explain this rise in executive pay. The first explanation is
that executive pay is an optimal contracting solution to the agency problems faced by a firm. As
the size and complexity of firms have increased, so has CEO compensation (Gabaix and Landier,
2008; Tervio, 2008). The second explanation is that high CEO pay is a form of rent extraction
by powerful CEOs and they are able to exert managerial power and influence on the boards
600
decisions (Bebchuk and Fried, 2003). In this section, we examine whether the concern that
experienced directors amplify agency problems is justified by examining the relation between
the proportion of experienced directors on the board and total CEO pay.
Data on CEO compensation are from Execucomp and are for the fiscal years 1998-2013.
Total pay includes salary, bonus, equity, and long-term incentive pay. There are a total of 19,020
firm-year observations with matching CRSP, Compustat, RiskMetrics, and Execucomp data.
Observations for the year 1998 are excluded since compensation for the CEO is determined by the
board from the previous year, and board structure data are only available from 1998. Execucomp
changed its reporting of CEO option valuation in 2006, which does not allow for consistent withinfirm and across-firm comparisons of total compensation. To allow for an accurate comparison,
we calculate CEO option compensation using a consistent methodology as described in detail in
the Appendix. Compensation is also adjusted for inflation using the consumer price index (CPI)
deflator for the year 2003. The mean total annual compensation for the entire sample is $4.62
million, while the median is $2.37 million (in 2003 dollars). There is a large difference in the mean
and the median, indicating, as is widely known, that the distribution of compensation is skewed.
Hence, log transformed compensation is used as the dependent variable. Since remuneration
decisions are made by the compensation committee, the ratio of experienced directors on the
compensation committee to the total number of directors on the compensation committee is used
as the key explanatory variable. The variable PROP-CC-G15 is used to denote the proportion of
independent directors with tenures greater than 15 years on the compensation committee.
Table VI provides the results of regressions where log transformed compensation is the
dependent variable. In Column (1), when firm and year fixed-effects are used, the coefficient
for PROP-CC-G15 is 0.095 and is significant at the 5% level. The average size of the
compensation committee is three members and the addition of one experienced director will raise
the proportion of experienced directors by 33%. The results in Column (1) indicate that a change
from no experienced director on the compensation committee to one such director will reduce
CEO compensation by 3.2% or $147,712 which is an economically significant effect, especially
given that we are examining within-firm variations. To distinguish the effect of experienced
directors from those who are not hired during the term of the CEO, Column (2) includes the
subsample of firm-year observations where CEO tenure is greater than or equal to that of all the
independent directors. In this subsample, all of the directors will have the potential to have been
influenced by the CEO. The results remain significant in this subsample.
In Column (3), the regression does not treat independent directors hired by the current CEO as
experienced. Doing so should produce stronger results as directors hired prior to the current CEO
are unlikely to be coopted by the current CEO. Consistent with this conjecture, the coefficient
for PROP-G15-CC in Column (3) is both economically and statistically more significant than
that in Column (1). In Column (4), the regression uses industry and year fixed-effects and also
controls for the proportion of directors hired during the term of the CEO on the compensation
committee. The coefficient for experienced directors is also significant at the 5% level.
These regressions have not controlled for the possibility of selection bias that may arise
when firms retain better directors. If this was indeed the case, results suggesting that
experienced directors help mitigate agency conflict would not be surprising. To address this
concern of selection bias, Column (5) uses director fixed-effects that allow the examination of
within-director variations as tenure changes. One complication that arises when using director
fixed-effects is a change in the methodology of setting director identification numbers by
RiskMetrics in the year 2003. To address this issue, we match directors within firm using name
(first, middle, and last) and director age to ensure that a consistent ID is used through the tenure
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
601
(2)
0.095
(0.020)
0.055
(0.442)
0.226
(0.028)
0.029
(0.598)
0.003
(0.507)
0.004
(0.586)
(3)
0.770
(0.072)
0.028
(0.892)
0.009
(0.985)
0.065
(0.724)
0.014
(0.357)
(4)
0.174
(0.014)
0.018
(0.823)
0.241
(0.024)
0.041
(0.456)
0.002
(0.618)
0.007
(0.410)
(5)
0.100
(0.020)
0.169
(0.121)
0.493
(0.000)
0.048
(0.343)
0.004
(0.312)
0.004
(0.781)
0.067
0.180
0.068 0.043
(0.006)
(0.640)
(0.013)
(0.000)
CEO Age
0.064
0.237 0.086
0.478
(0.734)
(0.164)
(0.603)
(0.486)
CEO Tenure
0.536
0.278
0.149
0.380
(0.329)
(0.151)
(0.493)
(0.181)
CEO Ownership
0.714 0.990 0.781 1.586
(0.078)
(0.029)
(0.068)
(0.011)
0.149 0.188
One Year Stock Return 0.153 0.113
(0.000)
(0.119)
(0.000)
(0.000)
Volatility
0.054
0.063
0.043
0.017
(0.136)
(0.672)
(0.262)
(0.615)
CEO-Chairman
0.036
(0.085)
Board Size
0.088
(0.008)
%Outside
0.153
(0.032)
Busy Director
0.015
(0.328)
Director Age
0.012
(0.021)
CEO-Hire
0.018
(0.172)
G15
0.005
(0.777)
CC
0.012
(0.276)
CEO-Chairman
0.020
(0.194)
CEO Age
0.780
(0.084)
CEO Tenure
0.504
(0.000)
CEO Ownership
0.450
(0.004)
One Year Stock Return 0.145
(0.000)
Volatility
0.120
(0.000)
G15 CC
(Continued)
602
(2)
0.227
(0.000)
0.036
(0.000)
0.378
(0.424)
16,622
0.673
Firm, Year
(3)
0.298
(0.013)
0.048
(0.002)
0.168
(0.889)
2,581
0.536
Firm, Year
(4)
0.240
(0.000)
0.033
(0.000)
0.664
(0.192)
16,622
0.609
Firm, Year
(5)
0.372
(0.000)
0.002
(0.168)
1.653
(0.000)
16,560
0.369
Industry, Year
Firm Size
Firm Age
R&D
0.316
(0.000)
0.000
(0.758)
1.474
(0.000)
94,012
0.593
Director, Year
of a director.6 The key explanatory variable in this regression is an interaction variable between
G15 and being a compensation committee member. This variable is significant at the 10% level
with a coefficient of 0.036 indicating that the presence of an experienced director on the
compensation committee is correlated with lower pay for the CEO.
2. CEO Turnover
Boards are tasked with monitoring the leadership of a firm and initiating changes in case of
poor performance (Adams, Hermalin, and Weisbach, 2010). The literature indicates that board
characteristics play an important role in the decision to dismiss a CEO. The factors linked to
a higher probability of CEO dismissal include higher equity-based compensation and higher
stock ownership for directors (Ertugrul and Krishnan, 2010), splitting CEO and chairman
duties (Goyal and Park, 2002), a majority of independent directors serving on at least two of
the three monitoring committees (Faleye et al., 2011), smaller boards (Yermack, 1996), higher
independence of the board (Weisbach, 1988), and the presence of female directors on the board
(Adams and Ferreira, 2009). In this section, we examine whether boards with a higher proportion
of experienced directors increase CEO turnover sensitivity to performance.
The dependent variable in this section is an indicator for CEO dismissals that takes a value
of one when the CEO is in their last year and zero otherwise. The indicator variable also has a
value of zero where the reason for CEO leaving has been marked as Deceased or Retired
by Execucomp. The turnover sample ends in the year 2013. There are 1,939 dismissals in the
sample of 17,536 firm-year observations, yielding an unconditional probability of 11.06%. The
intentions of the board and the CEO are not always clear, making the process of identifying
CEO dismissal extremely prone to error. Thus, similar to Jenter and Lewellen (2010) and
Adams and Ferreira (2009), all CEO dismissals are included in the sample. To identify the
effect of experienced directors on CEO departures in the case of poor performance, we include
the interaction term between stock returns and the proportion of experienced directors as an
independent variable. Stock returns are calculated as the previous years firm stock returns
minus the previous years return on the value weighted CRSP market index. Other than this
6
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
603
additional interaction term, the control variables are the same as in the previous section on
compensation.
Results for the regressions on CEO turnover are in Table VII. In Column (1), with firm
and year fixed-effects, the interaction term between stock returns and PROP-G15 is negatively
significant at the 5% level. This negative coefficient for the interaction term indicates that there
is a greater likelihood of CEO turnover when the firm performs poorly and there are experienced
directors on the board. There is also a positive coefficient for the interaction term between stock
returns and % CEO-Hire, demonstrating that the directors who were hired after the CEO are less
likely to dismiss a CEO even when the firm performs poorly. In contrast to experienced directors,
directors hired after the CEO exacerbate concerns about agency problems on the board. The two
effects are distinct since the interaction terms using both sets of variables are significant.
In order to further examine whether experience is just another way of looking at directors who
were not hired prior to the CEO, we examine the subset of observations where the CEO has a
longer tenure than any of the independent directors in Column (2). Within-firm variation in CEO
turnover and a corresponding change in the proportion of experienced directors is not possible
given the sample size. As such, we use industry and year fixed-effects in Column (2). The
coefficient of the interaction between PROP-G15 and stock returns is similar in magnitude to the
coefficient in Column (1), but the significance drops to the 10% level. In Column (3), we do not
consider directors as experienced if they are hired by the current CEO. Once again, the results
are stronger in this column. In Column (4), the regression includes all of the observations with
industry and year fixed-effects. The coefficient for the interaction variable is again significant at
the 5% level. Finally, in Column (5), we examine the within-director variation by using director
fixed-effects. The interaction term G15 CC in this column is significant at the 1% level. Overall,
the results provide little evidence of the ability of the CEO to entrench himself in the face of
poor performance if there are experienced directors on the board.
3. Earnings Restatement
Although earnings manipulation can lead to large losses in reputation for firm management,
financial incentives sometimes outweigh these concerns as managers seek to maximize
performance-linked payouts or retain their jobs by manipulating accounting figures in the
financial statements. The audit committee is responsible for appointing and evaluating auditors
and providing oversight concerning the integrity and compliance of company financial statements
to reduce the likelihood of such earnings manipulation, which can result in large shareholder
losses. Previous research has found that board independence and the presence of a financial
expert can reduce (Klein, 2002; Agrawal and Chadha, 2005) and the presence of foreign
directors on the audit committee can increase (Masulis et al., 2012) the probability of financial
misreporting. In a similar vein, we examine the propensity of firms to misreport earnings when
experienced directors are on the audit committee.
Data on accounting restatements are from the US Government Accountability Office (GAO)
database, which released reports in 2003 and 2007 containing a list of financial statements that
were incorrect on their release date. The reason for the inaccuracy could stem from either an
error (unintentional reporting) or irregularities (intentional misreporting) and it is important to
distinguish between the two effects (Hennes, Leone, and Miller, 2008). In addition, these reports
released by the GAO list the date when the correction was made to the financial statements,
but do not list the reporting period when the original infraction occurred. To address these two
concerns in the data, we use modified versions of these reports obtained from Masulis et al.
(2012) for the report released in 2007 (restatements from 2003 to 2006) and Burns and Kedia
604
0.079
(0.022)
0.030
(0.098)
0.031
(0.304)
0.101
(0.007)
0.067
(0.004)
0.003
(0.040)
0.339
(0.000)
0.107
(0.000)
0.233
(0.000)
0.077
(0.355)
0.118
(0.000)
0.129
(0.213)
0.021
(0.056)
0.028
(0.009)
0.197
(0.271)
(2)
0.066
(0.086)
0.082
(0.012)
0.108
(0.047)
0.051
(0.384)
0.001
(0.494)
0.066
(0.096)
0.315
(0.000)
0.060
(0.240)
0.033
(0.145)
0.086
(0.313)
0.027
(0.077)
0.004
(0.635)
0.103
(0.557)
(3)
(4)
0.108
(0.006)
0.022
(0.232)
0.031
(0.298)
0.104
(0.006)
0.068
(0.004)
0.004
(0.010)
0.339
(0.000)
0.073
(0.012)
0.232
(0.000)
0.080
(0.342)
0.126
(0.000)
0.137
(0.189)
0.016
(0.160)
0.028
(0.009)
0.190
(0.294)
0.079
(0.012)
0.028
(0.107)
0.026
(0.062)
0.024
(0.254)
0.050
(0.007)
0.001
(0.352)
0.146
(0.000)
0.016
(0.300)
0.141
(0.000)
0.091
(0.119)
0.022
(0.207)
0.114
(0.011)
0.023
(0.013)
0.006
(0.012)
0.010
(0.889)
(5)
G15 Ret
CEO-Hire Ret
Board Size
%Independent
Busy Director
Director Age
CEO-Hire
Tenure 15
CEO-Chairman
CEO Age
CEO Tenure
CEO Ownership
Stock Return
Firm Size
R&D
0.021
(0.006)
0.012
(0.027)
0.041
(0.000)
0.174
(0.000)
0.001
(0.847)
0.003
(0.165)
0.191
(0.000)
0.024
(0.000)
0.063
(0.000)
0.254
(0.000)
0.123
(0.031)
0.056
(0.164)
0.025
(0.000)
0.000
(0.965)
0.175
(0.005)
(Continued)
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
605
Tobins q
Firm Age
Observations
Adjusted R2
Fixed-effects
(1)
(2)
(3)
(4)
0.003
(0.428)
0.004
(0.005)
14,539
0.104
Firm, Year
0.002
(0.718)
0.001
(0.110)
2,266
0.190
Industry, Year
0.004
(0.375)
0.004
(0.006)
14,539
0.103
Firm, Year
0.006
(0.014)
0.001
(0.000)
14,487
0.066
Industry, Year
(5)
Tobins q
Firm Age
0.001
(0.482)
0.001
(0.000)
81,755
0.043
Director, Year
(2006) for the report released in 2003 (restatements from 1997 to 2002).7 These modified reports
include the misreported years and quarters and also use the Hennes et al. (2008) methodology
to classify restatements as errors or irregularities.
We use the proportion of experienced directors on the audit committee (PROP-AC-G15) as the
key independent variable. Table VIII contains the results for logistic regressions that use industry
and year fixed-effects. There is limited within-firm variation of earnings restatements. As such,
we do not use firm fixed-effects in the analysis. The control variables are similar to those used
in previous sections. The dependent variable in Column (1) is an indicator variable that takes
a value of one when a firm makes an earnings restatement and zero otherwise. The coefficient
for PROP-AC-G15 is not significant in this regression. In Column (2), the dependent variable
is an indicator variable that takes a value of one when the earnings restatement is classified
as an irregularity and the restatement is more likely to be due to a lapse in board monitoring.
In contrast to the results on all earning restatements, the results in Column (2) indicate that
the coefficient for PROP-AC-G15 is 1.36 and is negatively significant at the 5% level. The
marginal effect is 0.011 suggesting that adding one director with tenure greater than 15 years
on the audit committee, where none existed, will decrease the probability of an irregular earnings
restatement by about 0.36% (0.33 0.11). This is an economically significant effect as the
unconditional probability of an irregular earnings restatement is just 1.5%. Thus, the presence
of an experienced director can reduce the occurrence by about 25%.
Columns (3) and (4) repeat the analysis from Columns (1) and (2), but use the subset of firms
where the tenure of the CEO is greater than that of the independent directors. These regressions
continue to confirm that the presence of experienced directors on the audit committee is less
likely to be associated with intentional misreporting, even in cases where the CEO and the
experienced directors have had a chance to build a strong relationship over long tenures. Finally,
in Columns (5) and (6), we do not consider a director as experienced if they are hired by the
current CEO. Our key independent variable PROP-AC-G15 remains significantly negative in
the Irregularities sample. Overall, the results provide strong support for the hypothesis that the
presence of experienced directors on the board alleviates agency problems as these firms have
a lower propensity to misreport earnings intentionally.
7
We thank Natasha Burns, Simi Kedia, Ron Masulis, Cong Wang, and Fei Xie for providing us with the modified earnings
data.
606
Rest.
Irrg.
Rest.
Irrg.
(1)
(2)
(3)
(4)
(5)
(6)
0.199
(0.542)
0.016
(0.952)
0.090
(0.803)
0.006
(0.984)
0.001
(0.920)
0.063
(0.741)
0.039
(0.749)
16.180
(0.073)
6.073
(0.575)
0.522
(0.589)
0.123
(0.014)
0.001
(0.897)
0.014
(0.787)
0.491
(0.674)
10,344
0.053
1.356
(0.030)
0.196
(0.646)
0.969
(0.078)
0.168
(0.624)
0.016
(0.497)
0.052
(0.865)
0.111
(0.562)
35.464
(0.024)
29.501
(0.116)
5.148
(0.036)
0.296
(0.000)
0.000
(0.992)
0.003
(0.960)
1.393
(0.308)
9,157
0.098
0.229
(0.719)
0.003
(0.992)
0.463
(0.478)
0.425
(0.401)
0.018
(0.379)
3.260
(0.044)
0.361
(0.660)
0.165
(0.881)
1.059
(0.126)
0.021
(0.682)
0.298
(0.381)
0.230
(0.419)
0.128
(0.753)
0.005
(0.985)
0.017
(0.231)
0.133
(0.533)
0.012
(0.929)
14.073
(0.156)
3.618
(0.739)
0.464
(0.672)
0.128
(0.021)
0.001
(0.798)
0.169
(0.033)
0.008
(0.996)
10,344
0.059
1.339
(0.098)
0.083
(0.895)
0.656
(0.392)
0.124
(0.792)
0.004
(0.898)
0.049
(0.912)
0.130
(0.678)
34.681
(0.133)
40.417
(0.066)
11.955
(0.034)
0.329
(0.003)
0.003
(0.776)
0.283
(0.174)
0.869
(0.643)
9,157
0.114
Irrg.
0.261
0.571
(0.322)
(0.263)
18.619
59.525
(0.313)
(0.250)
9.168
47.235
(0.516)
(0.116)
0.295
8.964
(0.826)
(0.082)
0.194
0.169
(0.057)
(0.453)
0.006
0.032
(0.582)
(0.073)
0.017
0.193
(0.889)
(0.549)
0.936
1.272
(0.582)
(0.603)
1,915
1,251
0.113
0.161
Industry, Year
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
607
D. Board Advising
Besides monitoring management, directors serve as advisors, playing a key role in important
strategic decisions like making and choosing acquisitions (Adams et al., 2010). Experienced
directors develop expertise in the industry and firm on whose board they serve through their
long tenures exposing them to the companys strategy, finances, and competitive environment.
Recent research has explored the conflict between the advising and monitoring of management
that arises because independent directors are dependent on the CEO (Adams and Ferreira, 2007)
or inside directors (Harris and Raviv, 2008) for firm-specific information. The first best solution
to the trade-off between monitoring and advising in Adams and Ferreira (2007) is the sharing
of information by managers, so that the boards can both monitor management effectively and
give quality advice. An alternative may be the presence of experienced directors who have built
firm-specific knowledge and may not be totally reliant on management for information. Their
reduced dependence on management as a source of information would allow these directors to
fulfill their roles as both monitors and advisors.
In this section, two aspects of this advisory role are examined. First is the decision to pursue
acquisitions. Roll (1986) posits that takeovers are a sign of management hubris. We test the hypothesis that directors with long tenures may build relationships with the CEO and facilitate this
hubris to retain their positions and ensure reelection (Bebchuk and Fried, 2003). Next, we examine
the hypothesis that experienced directors can provide advisory benefits to the firm by testing
whether the presence of experienced directors on the board is associated with higher quality
acquisitions.
The Securities Data Commission (SDC) database is used to obtain data on deals and the criteria
in Masulis, Wang, and Xie (2007) is followed to identify acquisitions. Transactions where the
acquisition is completed and the acquirer controls less than 50% of the targets shares prior to the
announcement and owns 100% of the targets shares after the transaction are included. Deals must
be larger than $1 million and at least 1% of the acquirers market value of equity, as measured on
the 11th trading day prior to the announcement date. In addition to these conditions, deals where
the acquirer is making multiple deals on the same day are excluded (Faleye et al., 2011). After
including only those observations that have information on director attributes in RiskMetrics,
stock data from CRSP, and accounting data from Compustat, the sample contains 3,599
acquisitions.
In addition to the control variables from the previous sections, firm-level accounting variables,
like free cash flow and leverage, which have been used in the merger and acquisitions literature,
are included. The dependent variable is an indicator variable that is set to one when a firm
decides to pursue one or more acquisitions within the fiscal year. In Table IX, Column (1)
uses firm and year fixed-effects. The coefficient for PROP-G15 is negatively significant at the
10% level with a coefficient of -0.05. Using an average board size of nine, an increase in one
experienced director will lead to a 0.56% decrease in the probability of a merger. In Column (4),
industry and year fixed-effects are used and the coefficient for PROP-G15 remains negative and
significant at the 10% level. The importance of tenure can again be seen in Column (5), where
director and year fixed-effects are used. The coefficient for the PROP-G15 variable is significant
at the 10% level indicating that the likelihood of acquisitions is lower even when examining
within-director variation. Overall, the results imply that director term limits would rid the board
of a counterbalance to the CEO in the board room.
608
(2)
0.046
(0.090)
0.032
(0.246)
0.057
(0.154)
0.006
(0.840)
0.001
(0.414)
0.003
(0.846)
0.008
(0.437)
0.012
(0.748)
0.002
(0.906)
0.010
(0.929)
0.056
(0.000)
0.009
(0.000)
0.199
(0.389)
0.001
(0.868)
0.255
(0.000)
0.203
(0.000)
(3)
0.163
(0.061)
0.058
(0.511)
0.099
(0.376)
0.102
(0.073)
0.004
(0.390)
0.040
(0.175)
0.045
(0.374)
0.010
(0.793)
0.194
(0.213)
0.074
(0.150)
0.008
(0.173)
1.025
(0.091)
0.001
(0.933)
0.166
(0.408)
0.427
(0.002)
(4)
0.057
(0.041)
0.034
(0.217)
0.057
(0.152)
0.002
(0.944)
0.001
(0.435)
0.002
(0.902)
0.007
(0.464)
0.013
(0.729)
0.011
(0.549)
0.010
(0.930)
0.055
(0.000)
0.010
(0.000)
0.203
(0.359)
0.000
(0.953)
0.253
(0.000)
0.211
(0.000)
(5)
0.039
(0.066)
0.022
(0.252)
0.004
(0.916)
0.018
(0.416)
0.001
(0.551)
0.026
(0.037)
0.002
(0.794)
0.060
(0.002)
0.004
(0.874)
0.182
(0.009)
0.005
(0.102)
0.001
(0.020)
0.032
(0.787)
0.010
(0.003)
0.079
(0.178)
0.041
(0.102)
G15
Board Size
%Outside
Busy Director
Director Age
CEO-Hire
CEO-Chairman
CEO Age
CEO Tenure
CEO Ownership
Firm Size
Firm Age
R&D
Tobins q
Free Cash Flow
Leverage
0.008
(0.094)
0.066
(0.000)
0.063
(0.004)
0.010
(0.080)
0.001
(0.804)
0.003
(0.385)
0.005
(0.361)
0.063
(0.002)
0.065
(0.001)
0.021
(0.591)
0.001
(0.697)
0.001
(0.000)
0.212
(0.000)
0.001
(0.192)
0.160
(0.000)
0.164
(0.000)
(Continued)
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
609
Cash
Observations
Adjusted R2
Fixed-effects
(1)
(2)
(3)
(4)
0.007
(0.253)
15,478
0.172
Firm, Year
0.073
(0.301)
2,419
0.185
Firm, Year
0.006
(0.348)
15,478
0.216
Firm, Year
0.006
(0.008)
15,421
0.048
Industry, Year
(5)
Cash
0.002
(0.209)
104,168
0.156
Director, Year
2. Acquirer Returns
This section examines whether experienced directors assist firms in making more profitable
acquisitions, using a commonly employed event study methodology that utilizes cumulative
abnormal returns (CAR). Acquisitions are chosen under the same methodology as in the previous
section. Firm-level accounting data and board structure data from the year prior to the acquisition
is used. CAR for bidding firms is calculated as in Masulis et al. (2007). First, the market model is
estimated using daily stock returns data from CRSP for the 200-day period from Event Day 210
to Event Day 11. Then, the CAR are calculated over five days, two days before and two days after
the announcement day. In this sample, the median CAR is 0.257% and the mean CAR is 0.208%.
In addition to the control variables from the previous sections, variables deemed important in
previous studies on mergers and acquisitions (M&A) are also included, such as deal size (Moeller,
Schlingemann, and Stulz, 2004), percentage of payments made in cash (Travlos, 1987), different
industry (Morck, Shleifer, and Vishny, 1990), private company (Chang, 1998), and takeover defenses (Masulis et al., 2007). RiskMetrics data on two other corporate governance mechanisms
related to management entrenchment and the market for corporate control are also used. The first
variable (Staggered Board) denotes whether a firm elects only a fraction of its directors to the board
during annual elections. The second variable (GIM as in Gompers, Ishii, and Metrick, 2003) counts
the incidence of 24 governance rules as a proxy for shareholder rights. The GIM index is only updated every other year, so we assume the value does not change unless it is updated in the database.
Table X provides the results for OLS regressions where the dependent variable is the five-day
CAR, multiplied by 100. Since this is an event study with small or no within-firm variation, we
only use industry and year fixed-effects for all of the regressions. The coefficient for PROP-G15 in
Column (1) is significant at the 1% level with a magnitude of 1.84. In Column (2), the GIM index
is used as a control variable, while in Column (3), the Staggered Board variable is used as a control
variable. The coefficient for PROP-G15 continues to remain significant at the 1% level in these
regressions with a magnitude similar to that of Column (1). Furthermore, in Columns (4) to (6), we
again repeat the regressions, but do not treat directors hired by the current CEO as experienced. The
results in these three columns remain significant, but the level of significance appears to be slightly
lower than that in Columns (1) to (3). Overall, the evidence presented here suggests that experienced directors can provide significant advisory benefits to firms as their presence on the board
can lead to significantly better quality acquisitions as measured by announcement day returns.
610
PROP-G15
Board Size
% Independent
Busy Board
Mean Dir Age
% CEO-Hire
CEO-Chairman
CEO Age
CEO Tenure
CEO Ownership
Firm Size
Firm Age
R&D
Tobins q
Free Cash Flow
Leverage
(1)
(2)
(3)
(4)
(5)
(6)
1.841
(0.003)
0.381
(0.509)
0.617
(0.480)
0.636
(0.360)
0.006
(0.836)
0.148
(0.697)
0.236
(0.352)
0.274
(0.494)
0.431
(0.297)
2.137
(0.200)
0.180
(0.074)
10.009
(0.245)
5.094
(0.084)
0.077
(0.011)
0.894
(0.668)
0.185
(0.825)
2.143
(0.003)
0.209
(0.755)
0.679
(0.491)
0.094
(0.897)
0.021
(0.562)
0.113
(0.801)
0.200
(0.519)
0.786
(0.371)
0.018
(0.971)
3.282
(0.175)
0.257
(0.030)
16.310
(0.135)
6.880
(0.084)
0.096
(0.403)
0.612
(0.826)
0.081
(0.939)
1.815
(0.004)
0.585
(0.319)
0.533
(0.548)
0.624
(0.369)
0.004
(0.898)
0.263
(0.501)
0.112
(0.663)
0.287
(0.475)
0.516
(0.240)
2.103
(0.209)
0.161
(0.116)
13.083
(0.132)
5.128
(0.084)
0.074
(0.016)
0.957
(0.646)
0.245
(0.774)
1.748
(0.007)
0.402
(0.485)
0.668
(0.443)
0.643
(0.355)
0.010
(0.728)
0.221
(0.567)
0.246
(0.331)
0.276
(0.490)
0.380
(0.293)
2.197
(0.192)
0.182
(0.070)
10.375
(0.229)
5.054
(0.087)
0.078
(0.009)
0.909
(0.662)
0.173
(0.837)
1.871
(0.012)
0.195
(0.771)
0.754
(0.444)
0.106
(0.884)
0.017
(0.641)
0.183
(0.690)
0.206
(0.508)
0.798
(0.364)
0.006
(0.990)
3.341
(0.172)
0.262
(0.027)
17.438
(0.109)
6.903
(0.084)
0.096
(0.406)
0.613
(0.826)
0.073
(0.944)
1.648
(0.010)
0.605
(0.302)
0.590
(0.505)
0.635
(0.362)
0.008
(0.777)
0.330
(0.406)
0.122
(0.636)
0.287
(0.474)
0.475
(0.220)
2.166
(0.200)
0.164
(0.108)
13.649
(0.117)
5.088
(0.087)
0.075
(0.014)
0.970
(0.642)
0.236
(0.782)
(Continued)
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
611
Relative Size
Diff Industries
Private
% Cash
(1)
(2)
(3)
(4)
(5)
(6)
0.616
(0.180)
0.410
(0.079)
2.021
(0.000)
0.009
(0.001)
0.515
(0.229)
0.223
(0.448)
1.917
(0.000)
0.009
(0.004)
0.036
(0.461)
0.650
(0.164)
0.387
(0.099)
1.989
(0.000)
0.009
(0.000)
0.622
(0.176)
0.406
(0.082)
2.013
(0.000)
0.009
(0.001)
0.523
(0.223)
0.228
(0.439)
1.909
(0.000)
0.009
(0.004)
0.038
(0.445)
0.656
(0.161)
0.383
(0.102)
1.982
(0.000)
0.009
(0.000)
GINDEX
Staggered Board
Observations
Adjusted R2
Fixed-effects
3,599
0.044
2,464
0.050
0.205
(0.385)
3,510
3,599
0.043
0.044
Industry, Year
2,464
0.049
0.201
(0.395)
3,510
0.043
612
48-industry group and 2) have propensity scores no greater than or less than the treatment firms
score by 5%. Some of our firm-level outcomes (e.g., CEO compensation and financial restatements) concern only a specific board committee rather than the entire board. In this case, the
treatment firms refer to those that have at least one experienced director in the committee of interest, while the control firms are firms that have no experienced independent directors at the board.
Table XI reports the results. Our firm-level analysis contains five different dimensions including
1) CEO compensation, 2) CEO turnover, 3) financial irregularities, 4) acquisition decisions,
and 5) acquisition CARs. Correspondingly, each of the five columns in Table XI represents
one dimension. The key explanatory variable, PROP-G15, refers to the ratio of experienced
independent directors to all independent directors for CEO turnover, acquisition decisions, and
acquisition CARs tests. Since CEO compensation (financial irregularities) concerns only the
compensation (audit) committee, PROP-G15, in this test, refers to the ratio of experienced
independent directors on the compensation (audit) committee only. To save space, some control
variables that are specific to certain dimensions only (e.g., stock return in CEO turnover and M&A
characteristics in acquisition CARs) are included in the regressions but not reported in the table.
The matching, on average, reduces the sample size by 25%. Nevertheless, the variable
PROP-G15 is still significantly negative (positive) in the financial irregularities (acquisition
CARs) test. For the CEO turnover test, the interaction term of PROP-G15 and stock returns
remains significantly negative. Thus, even though the results in this table are relatively less
strong, they still indicate that our findings cannot be solely explained by experienced directors
staying longer in better firms.
B. Firms Selecting Directors
The second concern we consider is the possibility that firms only retain better directors on
their boards over time. If this was the case, then a higher proportion of experienced directors
might be equivalent to a higher proportion of high-quality directors, which makes it unsurprising
that these directors facilitate more effective monitoring and advising. To deal with this issue,
we make further use of the director fixed-effects. Specifically, we first focus on experienced
directors who have multiple directorships in the same year and compare their performance (both
at the director level and the firm level) across these directorships. If our results are capturing
the effect of director quality, rather than the effect of director tenure, then cross-sectionally, a
director should exhibit no differences between boards on which they have been a director for
over 10 years and boards on which they have been a director for three years.
To determine whether this is true, we restrict our sample to directors who: 1) have more than one
directorship in a given year and 2) have a tenure of more than 12 years in at least one directorship
and a tenure of less than five years in at least one directorship.8 The results are presented in Panel
A of Table XII. In Columns (1) and (2), we examine these directors committee memberships
and meeting attendance. In doing so, we rerun the regressions in Tables III and IV, but include
director year fixed-effects. As such, we drop variables that have no variation within the same
director in a given year (e.g., age and gender). Our key independent variable is director tenure.
The coefficient for director tenure is positive and significant at the 10% level in the Committee
Membership test and is insignificant in the Poor Attendance test, suggesting that these directors
are more likely to be committee members at firms where they are relatively more experienced.
Next, we examine whether the monitoring and advising are more effective in firms where these
experienced directors have stayed for a longer time period as compared to firms that they have
8
The decision to reduce the threshold from 15 to 12 years in this case is to obtain sufficient observations.
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
613
PROP-G15
Compensation
Turnover
Irregularity
Acquisition
CAR
(1)
(2)
(3)
(4)
(5)
0.019
(0.827)
0.238
(0.044)
0.537
(0.000)
0.014
(0.844)
0.008
(0.082)
0.069
(0.314)
0.192
(0.000)
2.054
(0.344)
2.023
(0.071)
1.602
(0.017)
0.355
(0.000)
0.003
(0.021)
1.250
(0.006)
0.082
(0.000)
12,601
0.363
0.014
(0.430)
0.076
(0.043)
0.011
(0.483)
0.011
(0.660)
0.043
(0.051)
0.000
(0.772)
0.138
(0.000)
0.137
(0.000)
0.130
(0.353)
0.359
(0.036)
0.148
(0.004)
0.007
(0.012)
0.001
(0.001)
0.041
(0.616)
0.008
(0.003)
10,860
0.063
0.019
(0.071)
0.002
(0.864)
0.016
(0.233)
0.005
(0.619)
0.000
(0.826)
0.003
(0.638)
0.003
(0.523)
0.591
(0.147)
0.198
(0.494)
0.067
(0.000)
0.004
(0.073)
0.000
(0.714)
0.001
(0.981)
0.002
(0.169)
6,427
0.011
Industry, Year
0.000
(0.991)
0.068
(0.002)
0.022
(0.526)
0.020
(0.405)
0.001
(0.413)
0.035
(0.006)
0.004
(0.666)
0.055
(0.209)
0.034
(0.839)
0.133
(0.055)
0.006
(0.062)
0.001
(0.012)
0.038
(0.718)
0.008
(0.019)
12,546
0.044
1.550
(0.080)
1.203
(0.149)
0.184
(0.866)
0.250
(0.721)
0.034
(0.395)
1.088
(0.132)
0.209
(0.494)
0.213
(0.702)
14.176
(0.653)
3.020
(0.227)
0.090
(0.438)
8.329
(0.441)
4.621
(0.110)
0.090
(0.308)
2,183
0.053
Observations
Adjusted R2
Control Variables
Fixed-effects
Tenure
Observations
Adjusted R2
Control Variables
Fixed-effects
Tenure
13,335
0.270
Yes
15,196
0.066
Yes
0.011
(0.000)
0.008
(0.013)
2,432
0.502
Yes
0.022
(0.267)
(3)
CEO
Compensation
11,898
0.265
Yes
0.023
(0.629)
2,465
0.037
Yes
2,809
0.169
Yes
0.072
(0.074)
0.051
(0.516)
(2)
(1)
Poor
Attendance
Committee
Member
0.000
(0.908)
0.031
(0.017)
10,295
0.057
Yes
0.005
(0.738)
0.165
(0.001)
2,155
0.143
Yes
(4)
CEO
Turnover
7,039
0.005
Yes
0.020
(0.078)
2,251
0.108
Yes
0.001
(0.047)
(5)
Financial
Irregularity
11,790
0.018
Yes
0.006
(0.010)
2,469
0.049
Yes
0.009
(0.876)
(6)
Acquisition
Decisions
This table reports results of repeating some of the previous regressions with director-fixed-effects. Panel A implements director year fixed-effects and contains firms in which at least
one independent director has more than one directorship in a given year and has a tenure of more than 12 years in at least one directorship and a tenure of less than five years in at least one
directorship. Panel B implements director firm fixed-effects and contains firms which have experienced directors and whose data are available for at least one year when the tenure of
the experienced director is below five years and when the tenure of the experienced director is over 12 years. The coefficients in Column (2) of both panels are multiplied by 10. Control
variables in Columns (1) to (6) are initially from Column (1) in Table III, Column (1) in Table IV, Column (4) in Table VI, Column (4) in Table VII, Column (2) in Table VIII, and Column
(4) in Table IX, respectively. Thereafter Panel A drops Age L65, Female Director, and Busy Director; while Panel B drops Age_L65, Female Director, Firm Age, Director Meeting Fee,
and Number of Meetings. Standard errors are robust and clustered by firm. The p-values are in parentheses.
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
615
joined recently. The sample size drops significantly as we only include firm-year observations
where at least one independent director has more than one directorship and is experienced in some
directorships and inexperienced in some others. Due to these restrictions, we exclude Acquisition
CARs from our analysis. We also replace the logit regression in Financial Irregularity with an OLS
regression. In addition, for the CEO Compensation and Financial Irregularity tests, we have very
few observations if we require the director to be on the compensation/audit committee in all of
the directorships that they have. Therefore, in this section, we do not implement such a restriction.
As the results indicate, even within the much smaller samples, firms in which these directors
are experienced make CEO turnover more sensitive to performance, and are less likely to restate
earnings due to irregularities, than firms where these directors are relatively less experienced.
To further investigate whether the positive impact experienced directors bring to the firm is
really driven by the experience they accumulate over time, in Panel B of Table XII, we repeat the
above tests, but replace director year fixed-effects with director firm fixed-effects. To this
end, we restrict our sample to firms that have experienced directors and we require that the data
for these firms are available for at least one year, both when the tenure of the experienced director
is below five years and when the tenure of the experienced director is over 12 years. Similar to
the tests in Panel A, we first examine directors committee memberships and meeting attendance.
We drop control variables that have no variation within each director-firm combination (e.g.,
gender), and variables that have extremely high correlation with tenure (e.g., director age,
firm age). For the Meeting Attendance test, we also have to drop Director Meeting Fee and
Number of Meetings. These two variables from Execucomp are only available until 2006 and, by
construction, we have very few observations in this section if our sample period ends in 2006. As
Columns (1) and (2) in Panel B indicate, as a directors tenure increases, they are significantly
more likely to be a committee member and significantly less likely to miss board meetings.
At the firm level, our methods are similar to those in Panel A with two exceptions. The first
is that the regressions in Panel B include director firm fixed-effects rather than director
year fixed-effects. The second is that we exclude Firm Age from the control variables. Overall,
the results in Panel B suggest that as a directors tenure at a firm increases, the firms CEO
turnover-performance sensitivity increases, the probability to restate earnings due to irregularities
decreases, and the propensity to attempt M&A transactions decreases. These results indicate
that directors do exhibit a significant improvement in the effectiveness of their monitoring and
advising roles as their tenures increase over time.
C. General Experience versus Firm-Specific Experience
The third issue we consider is whether the experience that makes directors more effective at
monitoring and advising can be any directorship experience. Our argument is that experienced
directors become better due to their long service at the firm. However, if directors are able
to apply the experience they accumulate across boards, then they do not have to accumulate
experience by always staying at the same firm, and firms can easily replace long-tenured
directors with directors who have served on many boards in the past. In this subsection, we
examine how our results change when we replace our variables related to director tenure with
some variables that directly measure directors general experience.
To measure general experience, we use the number of S&P 1500 directorship-year observations the director has had over the past five years. Since directors with tenure longer than 15
years account for 14% of our independent directors sample, we choose the threshold of nine
directorship-years as it also gives us approximately 14% experienced directors in the sample.
That is, a director who has had five directorships each year for two years and a director who
616
CEO
Turnover
Financial
Irregularity
(1)
(2)
(3)
0.149
(0.035)
0.017
(0.407)
0.047
(0.289)
0.026
(0.056)
0.016
(0.465)
0.047
(0.014)
0.000
(0.505)
0.146
(0.000)
0.139
(0.000)
0.090
(0.116)
0.022
(0.217)
0.114
(0.012)
0.005
(0.034)
0.001
(0.000)
0.013
(0.864)
0.510
(0.466)
0.180
(0.084)
0.471
(0.000)
0.043
(0.504)
0.001
(0.751)
0.021
(0.595)
0.182
(0.000)
0.276
(0.576)
1.062
(0.039)
0.812
(0.015)
0.406
(0.000)
0.000
(0.792)
1.414
(0.001)
0.267
(0.684)
0.849
(0.321)
0.337
(0.494)
0.004
(0.881)
0.053
(0.907)
0.229
(0.490)
46.951
(0.038)
41.430
(0.088)
11.388
(0.049)
0.374
(0.004)
0.003
(0.776)
0.498
(0.790)
Acquisition
Decisions
(4)
0.047
(0.068)
0.040
(0.028)
0.000
(0.995)
0.025
(0.245)
0.000
(0.623)
0.024
(0.036)
0.006
(0.401)
0.061
(0.002)
0.002
(0.948)
0.156
(0.018)
0.007
(0.021)
0.001
(0.053)
0.021
(0.835)
CAR
(5)
0.566
(0.451)
0.481
(0.404)
0.515
(0.547)
1.086
(0.135)
0.033
(0.264)
0.246
(0.525)
0.161
(0.514)
0.207
(0.603)
0.655
(0.079)
2.241
(0.183)
0.190
(0.072)
12.639
(0.128)
6.197
(0.037)
(Continued)
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
617
0.093
(0.000)
13,620
0.418
CEO
Turnover
(2)
0.006
(0.015)
14,105
0.064
Financial
Irregularity
(3)
0.368
(0.160)
8,134
0.121
Industry, Year
Acquisition
Decisions
(4)
0.007
(0.015)
13,429
0.044
CAR
(5)
0.069
(0.028)
3,454
0.044
has had two directorships each year for five years are both considered as experienced directors
in this section. Thereafter, we calculate the proportion of these directors on boards and on
audit/compensation committees and use these values as our key independent variables. As the
results in Table XIII suggest, having a large proportion of these directors does not facilitate more
effective monitoring or advising. The only positive impact of these directors is that they tend to
make less acquisition attempts. However, at the same time, a larger proportion of these directors
on the compensation committee appears to increase CEO compensation. Overall, the results
in this section are largely different from those in previous sections, suggesting that general
directorship experience is not a substitute for firm-specific experience.
IV. Conclusion
Advocates for improving corporate governance and regulators recommend term limits for
independent directors. The premise is that new directors will infuse innovative ideas and energy
into the boardroom and will also be less likely to align with the CEO when the time comes
to set compensation and consider leadership changes. A contrasting hypothesis in support of
experienced directors stems from the fact that these directors have significant equity stakes
in firms, have experience dealing with multiple CEOs, and have had an opportunity to learn
about the business and the industry. In this paper, we study whether calls for term limits
are justified by examining how the presence of experienced directors on boards affects firm
policies.
We use data from S&P 1500 companies over 16 years to examine monitoring and advising
outcomes using the proportion of directors with a tenure greater than 15 years (an experienced
director) as the key explanatory variable. We find that CEOs in firms with a larger number of
experienced directors tend to have lower compensation and are more likely to leave when the
firm performs poorly. These firms are also less likely to make earnings restatements. Firms with
a higher proportion of experienced directors are also less likely to make acquisitions and those
that are made are more likely to be profitable. The results on both monitoring and advising
outcomes suggest that experienced directors provide a balance of power in the boardroom.
This paper also provides evidence regarding the trade-off between the advising and monitoring
roles of board members. Previous research indicates that excessive monitoring of the CEO leads
to the possibility that managers withhold information from the board, compromising advising
quality. However, experienced directors may have developed firm-level expertise over their
618
tenure and not be totally dependent on management for insight into firm operations. The results
indicate that the presence of these directors on boards leads to better strategic advice and better
monitoring concerning decision making. Overall, the results confirm that experienced directors
make a valuable contribution to corporate governance within firms.
Appendix
The Financial Accounting Standards Board (FASB) issued FAS 123 in October 1995 that
encouraged firms to report their equity based compensation using a fair-value based method of
accounting. The intention was to provide a more realistic valuation of the payment to executives
compared to previous methods (APB 25) whereby stock option grants were often recognized
without any compensation cost. FASB introduced FAS 123R at the end of 2004, which made
the optional implementation of fair-value based accounting mandatory for all public firms.
Subsequently, Execucomp changed its methodology for equity-based compensation from the
fiscal year 2006 onward to incorporate FAS 123.9
Prior to fiscal year 2006, Execucomp calculated stock option grants using its own standardized
Black-Scholes methodology. After fiscal year 2006, Execucomp reported the value of the option
grants as calculated by individual firms. Since firms may use their own internal models to value
stock option grants, annual option compensation cannot be compared across firms after 2006.
There will also be a difference in option valuation within firms before and after fiscal year 2006.
We follow Kini and Williams (2012) and Coles et al. (2014) and use the pre-2006 Execucomp
methodology to calculate option awards for all years and all firms to ensure that compensation
is comparable.
The pre-2006 Execucomp methodology used the following inputs:
A grant date of July 1 is used for all options in a given fiscal year. To account for the fact
that executives often exercise their options early, the time to maturity of the option was
set to 70% of the actual time to maturity. This number was then rounded to the nearest
whole number.
The seven-year risk free rate of a US Treasury bond.
The 60-month stock volatility is used. If there are fewer than 12 months are available,
then the volatility of the S&P 1500 is used. Otherwise, if there are fewer than 60 months
available, then those number of months are used. The data are winsorized at the 5th
percentile and the 95th percentile.
The average dividend yield over the previous three years is used. The data are winsorized
at the 5th percentile and the 95th percentile.
The company specified strike price and market price of the option were used.
In order to ensure the correctly replicated Execucomp methodology, we calculate the option
valuation of all firms in all years before 2006 and obtain a correlation of 0.9973. We then
calculate total compensation using the formula below:
Total Compensation = SALARY + BONUS + OTHANN + ALLOTHTOT + RSTKGRNT
+ option awards calculated value + LTIP.
9
Please refer to http://www.fasb.org/summary/stsum123r.shtml for details on FAS 123R and http://www.
fasb.org/summary/stsum123.shtml for details on FAS 123.
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
619
The only change in our calculation from the Execucomp calculated total compensation, TDC1,
is the calculation of the option awards. We find the correlation between total compensation
measures is 0.996.
After the year 2006, we again calculate option values and leave all of the other inputs the
same. Prior to 2006, all inputs were available in the Execucomp table Stock Option Grants 1992 Format. After 2006, the number of options granted and the exercise price used are from the
Execucomp table Plan Based Awards, while the exercise price is inferred from the Execucomp
table Outstanding Equity Awards using exercise price and number of options in the grant.
Total Compensation = SALARY + BONUS + NONEQ INCENT + OTHCOMP
+ STOCK AWARDS FV + option awards calculated value
+ DEFER RPT AS COMP TO.
The correlation between our calculation of total compensation and TDC1 from Execucomp is
unsurprisingly a bit lower at 96.6%.
References
Adams, R. and D. Ferreira, 2007, A Theory of Friendly Boards, Journal of Finance 62, 217250.
Adams, R. and D. Ferreira, 2009, Women in the Boardroom and Their Impact on Governance and Performance, Journal of Financial Economics 94, 291309.
Adams, R., B. Hermalin, and M. Weisbach, 2010, The Role of Boards of Directors in Corporate Governance:
A Conceptual Framework and Survey, Journal of Economic Literature 48, 58107.
Adams, R.B. and D. Ferreira, 2008, Do Directors Perform for Pay? Journal of Accounting and Economics
46, 154171.
Agrawal, A. and S. Chadha, 2005, Corporate Governance and Accounting Scandals, Journal of Law and
Economics 48, 371406.
Baker, M. and P. Gompers, 2003, The Determinants of Board Structure at the Initial Public Offering,
Journal of Law and Economics 46, 569598.
Bebchuk, L. and J. Fried, 2003, Executive Compensation as an Agency Problem, Journal of Economic
Perspectives 17, 7192.
Burns, N. and S. Kedia, 2006, The Impact of Performance-Based Compensation on Misreporting, Journal
of Financial Economics 79, 3567.
Chang, S., 1998, Takeovers of Privately Held Targets, Methods of Payment, and Bidder Returns, Journal
of Finance 53, 773784.
Chen, L. and F. Moers, 2014, The Market for Independent Directors, Available at SSRN 2117058.
Coles, J., N. Daniel, and L. Naveen, 2008, Boards: Does One Size Fit All? Journal of Financial Economics
87, 329356.
Coles, J., N. Daniel, and L. Naveen, 2014, Co-Opted Boards, Review of Financial Studies 27, 17511796.
Core, J., R. Holthausen, and D. Larcker, 1999, Corporate Governance, Chief Executive Officer Compensation, and Firm Performance, Journal of Financial Economics 51, 371406.
620
Do, Q., B.D. Nguyen, and P.R. Rau, 2014, Sugar and Spice and Everything Nice: What Are Independent
Career Directors Made Of? Working paper.
Ertugrul, M. and K. Krishnan, 2010, Can CEO Dismissals Be Proactive? Journal of Corporate Finance
17, 134151.
Fahlenbrach, R., A. Low, and R. Stulz, 2010, Why Do Firms Appoint CEOs as Outside Directors? Journal
of Financial Economics 97, 1232.
Fahlenbrach, R., A. Low, and R. Stulz, 2013, The Dark Side of Outside Directors: Do They Quit Ahead of
Trouble? Available at SSRN 1585192.
Faleye, O., R. Hoitash, and U. Hoitash, 2011, The Costs of Intense Board Monitoring, Journal of Financial
Economics 101, 160181.
Frydman, C. and R. Saks, 2010, Executive Compensation: A New View from a Long-Term Perspective,
19362005, Review of Financial Studies 23, 20992138.
Gabaix, X. and A. Landier, 2008, Why Has CEO Pay Increased So Much? Quarterly Journal of Economics
123, 49100.
Goldman, E., J. Rocholl, and J. So, 2009, Do Politically Connected Boards Affect Firm Value? Review of
Financial Studies 22, 23312360.
Gompers, P., J. Ishii, and A. Metrick, 2003, Corporate Governance and Equity Prices, Quarterly Journal
of Economics 118, 107156.
Goyal, V. and C. Park, 2002, Board Leadership Structure and CEO Turnover, Journal of Corporate Finance
8, 4966.
Guner, A., U. Malmendier, and G. Tate, 2008, Financial Expertise of Directors, Journal of Financial
Economics 88, 323354.
Guthrie, K., J. Sokolowsky, and K.-M. Wan, 2012, CEO Compensation and Board Structure Revisited,
Journal of Finance 67, 11491168.
Harris, M. and A. Raviv, 2008, A Theory of Board Control and Size, Review of Financial Studies 21,
17971832.
Heidrick & Struggles, 2007, 10th Annual Corporate Board Effectiveness Study, Los Angeles, CA, Heidrick
& Struggles and USC/Center for Effective Organizations.
Hennes, K.M., A.J. Leone, and B.P. Miller, 2008, The Importance of Distinguishing Errors from Irregularities in Restatement Research: The Case of Restatements and CEO/CFO Turnover, Accounting Review
83, 14871519.
Hermalin, B. and M. Weisbach, 1998, Endogenously Chosen Boards of Directors and Their Monitoring of
the CEO, American Economic Review, 96118.
Jensen, M. and W. Meckling, 1976, Theory of the Firm: Managerial Behavior, Agency Costs and Ownership
Structure, Journal of Financial Economics 3, 305360.
Jenter, D. and K. Lewellen, 2010, Performance-Induced CEO Turnover, Stanford University Working
Paper.
Kaplan, S. and B. Minton, 2012, How Has CEO Turnover Changed? International Review of Finance 12,
5787.
Kini, O. and R. Williams, 2012, Tournament Incentives, Firm Risk, and Corporate Policies, Journal of
Financial Economics 103, 350376.
Dou, Sahgal, & Zhang r Should Independent Directors Have Term Limits?
621
Klein, A., 2002, Audit Committee, Board of Director Characteristics, and Earnings Management, Journal
of Accounting and Economics 33, 375400.
Landier, A., J. Sauvagnat, D. Sraer, and D. Thesmar, 2012, Bottom-Up Corporate Governance, Review of
Finance 17, 161201.
Masulis, R. and S. Mobbs, 2011, Are All Inside Directors the Same? Evidence from the External Directorship Market, Journal of Finance 66, 823872.
Masulis, R., C. Wang, and F. Xie, 2007, Corporate Governance and Acquirer Returns, Journal of Finance
62, 18511889.
Masulis, R.W., C. Wang, and F. Xie, 2012, Globalizing the Boardroom: The Effects of Foreign Directors
on Corporate Governance and Firm Performance, Journal of Accounting and Economics 53, 527554.
Moeller, S., F. Schlingemann, and R. Stulz, 2004, Firm Size and the Gains from Acquisitions, Journal of
Financial Economics 73, 201228.
Morck, R., A. Shleifer, and R. Vishny, 1990, Do Managerial Objectives Drive Bad Acquisitions? Journal
of Finance 45, 3148.
National Association of Corporate Directors (NACD), 2005, Report of the NACD Blue Ribbon Commission
on Director Professionals (November, 1996, reissued 2001, 2005).
Oyer, P., 2008, The Making of an Investment Banker: Macroeconomic Shocks, Career Choice, and Lifetime
Income, Journal of Finance 63, 26012628.
Roll, R., 1986, The Hubris Hypothesis of Corporate Takeovers, Journal of Business, 197216.
Tervio, M., 2008, The Difference That CEOs Make: An Assignment Model Approach, American Economic
Review 98, 642668.
Travlos, N., 1987, Corporate Takeover Bids, Methods of Payment, and Bidding Firms Stock Returns,
Journal of Finance 42, 943963.
Vafeas, N., 2003, Length of Board Tenure and Outside Director Independence, Journal of Business Finance
& Accounting 30, 10431064.
Weisbach, M., 1988, Outside Directors and CEO Turnover, Journal of Financial Economics 20, 431460.
Yermack, D., 1996, Higher Market Valuation of Companies with a Small Board of Directors, Journal of
Financial Economics 40, 185211.
Young, C., 2011, Boards: When Best Practice Isnt Enough, McKinsey Quarterly, June.