Dou 2017 Leaving Before Bad Times Does The Labor Market Penalize

Journal of Accounting and Economics 63 (2017) 161–178
Contents lists available at ScienceDirect
Journal of Accounting and Economics

journal homepage: www.elsevier.com/locate/jae
Leaving before bad times: Does the labor market penalize

preemptive director resignations?$
Ying Dou
School of Banking and Finance, UNSW Business School, Australia
a r t i c l e i n f o abstract
Article history: When firms experience negative events such as lawsuits or earnings restatements, their
Received 14 December 2014 directors also suffer. But what about those who leave shortly before the events? I show
Received in revised form that directors who leave prior to negative events experience greater declines in the
8 February 2017
number of their directorships than directors who stay through the events, but smaller
Accepted 9 February 2017
declines than directors who leave after the events. These declines do not appear to be
Available online 29 March 2017
voluntary or driven by forced departures. Instead, they appear to be the results of labor
JEL classification: market penalties. The results suggest that resigning pre-emptively does not protect di-
G30 rectors from labor market penalties.
G34
& 2017 Elsevier B.V. All rights reserved.
Keywords:
Director departures
Reputational concerns
Board seats
Labor market settling-up
1. Introduction
How effective is the director labor market? Prior studies suggest that, when firms experience negative events (e.g., class
action lawsuits, and earnings restatements), their directors also suffer. Some directors are removed from the boards in the
post-crisis period (e.g., Brochet and Srinivasan, 2014; Srinivasan, 2015); while some others experience a decline in their
subsequent number of board seats (e.g., Fich and Shivdasani, 2007; Fos and Tsoutsoura, 2014). However, while there is
ample evidence that the labor market penalizes directors who are present when the events occur, whether directors can
avoid these penalties by resigning shortly before the events is less clear.
I examine the labor market consequences for independent directors who resign before the firm experiences a negative
event. Pre-emptive resignations are not a new topic in the director turnover literature. Brown and Maloney (1999) find that
when outside directors of acquirer firms resign prior to acquisitions, acquirers experience lower announcement returns. The
☆
I would like to thank Renée Adams, Ronald Masulis and Jason Zein for their valuable guidance on this paper. I have received helpful comments and
suggestions from Ali Akyol (FIRN discussant), Lindsay Baran (FMA discussant), David Denis (FMA Doctoral Student Consortium Leader) Kevin Davis (Sirca
discussant), Mark Humphery-Jenner, Jonathan Karpoff, Jing Ma (AFBC discussant), Gwenaël Roudaut, Sidharth Sahgal, Robert Tumarkin, Terry Walter, Jing
Xu, Emma Zhang, and participants at a UNSW brown bag seminar, 3rd Sirca Young Research Workshop, 2013 Financial Research Network (FIRN) Annual
Conference, the 26th Australasian Finance and Banking Conference (AFBC), 2014 Financial Management Association Annual Meeting and its Doctoral
Student Consortium, and a seminar at the Monash Business School. I also thank Jonathan Karpoff, Allison Koester, Scott Lee, and Gerald Martin for
providing the Federal Securities Regulation Database; and Michael Roberts and Amir Sufi for providing the data on debt covenant violations. This paper is
the first chapter of my doctoral dissertation at the University of New South Wales. All errors are my own.
E-mail address: ying.dou@unsw.edu.au
http://dx.doi.org/10.1016/j.jacceco.2017.02.002
0165-4101/& 2017 Elsevier B.V. All rights reserved.
162 Y. Dou / Journal of Accounting and Economics 63 (2017) 161–178
authors' interpretation is that outside directors have an incentive to leave when they expect the firm to perform poorly.
Asthana and Balsam (2007) find complementary evidence that anticipation of subsequent poor performance is a significant
predictor of independent director departures. Fahlenbrach et al. (2017) show that anticipation of negative events such as
lawsuits or earnings restatements leads independent directors to resign from a firm to protect their reputation and avoid an
increase in their workload.
There are many reasons why independent directors might leave prior to negative events. First, directors have to invest
additional time and effort when the performance of the firm declines. Vafeas (1999) finds an increased number of board
meetings following share price declines. Since independent directors tend to hold multiple board seats and face time
constraints (Fich and Shivdasani, 2006), this workload increase can be particularly problematic. Also, staying with the firm
during bad events exposes directors to litigation risks (Brochet and Srinivasan, 2014). The evidence in Agrawal and Chen
(2011) suggests that directors who remain at the firm's board are more likely to be named as defendants in subsequent class
action lawsuits compared with those who have already left. Lastly, for some directors, staying also exposes them to the risk
of being targeted in proxy contests, which severely damages their reputation (Fos and Tsoutsoura, 2014). Since being as-
sociated with a negative event brings several expected costs as well as damaging a director's reputation, it is reasonable that
some directors find it tempting to resign pre-emptively upon anticipating a negative event.
Using class action lawsuits, earnings restatements, severe dividend reductions, and debt covenant violations as proxies
for negative events, I show that the labor market does penalize pre-emptive resignations. Specifically, directors who resign
prior to negative events experience significantly greater directorship declines than directors who have stayed through these
events. Compared to these staying directors who have already been shown by the literature to suffer a reputation damage,
resigning directors receive an additional 10.8% reduction in their number of board seats, which is economically significant.
Even though resigning pre-emptively brings stronger penalties than staying through the events, such penalties are still
smaller compared to leaving a troubled firm shortly after the negative event occurs. Directors who leave in the post-event
period suffer an 18.5% further board seat reduction relative to directors who stay, which is 71% greater than the penalties
that directors who leave in the pre-event period receive. The stronger penalties for ex-post departures are not surprising,
because these directors may have exposed themselves to lawsuits and proxy contests mentioned above, and are likely held
most accountable for the negative events.
Directors may leave a board for other reasons. For example, some director departures may be forced rather than vo-
luntary; some directors may leave a firm to voluntarily reduce their board commitments due to poor health or other time
constraints; and some directors may not have anticipated the subsequent negative events when they resign, and thus their
resignations should not be considered as “pre-emptive.” To address these alternative explanations, I infer a director's reason
for resigning by looking at whether multiple directors leave the same firm during the same year, whether the director
acquires a new board seat shortly after resigning, and whether the director sold any company stock shortly before resigning.
These alternative explanations do not fully explain the decline in board seats. Lastly, as a placebo test, I examine whether
directors who resign from matched benchmark firms that do not experience a negative event suffer similar directorship
declines. The results suggest that leaving a non-troubled firm during the same period does not have any impact on a
director's subsequent number of directorships, confirming that the penalties for resigning only exist among negative-event
firms.
After documenting the labor market consequences for pre-emptive resignations, I examine the channels through which
these consequences occur. Cross-sectional analysis based on directors' other existing directorships shows that directors who
leave a troubled firm shortly before and shortly after the negative events are more likely to also lose their other boards
relative to directors who have stayed during the events. Consistent with the baseline results, such likelihood is higher for
those who leave in the post-event period than for those who leave in the pre-event period. In addition, I find evidence that
directors who stay subsequently acquire more directorships at firms that experience similar negative events. This can be
another explanation why directors who leave a negative-event firm suffer greater directorship declines than the staying
directors.
Labor market penalties extend beyond the loss of board seats. For directors that leave, there is a higher likelihood of
losing committee chairmanships at their other boards compared to directors that stay through the events. Such effect is,
once again, stronger for those who leave after the events occur than for those who leave before the events occur. Directors
who leave (both before and after the events) also tend to subsequently hold less prestigious directorships. Upon the initial
revelation of these negative events, the stock prices of the other firms where the ex-ante resigning directors are still board
members react negatively, especially when the resigning directors play important roles on these boards. Relatedly, share-
holders in these connected firms tend to react positively when these directors subsequently leave these boards.
This paper aims to fill a gap in the literature on the director labor market. Studies that examine firms experiencing
negative events such as bankruptcy (Gilson, 1990), dividends cuts (Kaplan and Reishus, 1990), earnings restatements
(Srinivasan, 2005), financial fraud (Fich and Shivdasani, 2007), option backdating (Bereskin and Smith, 2014), proxy contests
(Fos and Tsoutsoura, 2014), or rejections of takeover offers (Harford, 2003) document negative consequences for the di-
rectors in these firms. Yet, it is unclear whether such penalties also extend to those who leave before the problem is
revealed. Overall, the results in this study suggest that directors cannot avoid the penalties by resigning pre-emptively. Even
though the penalties for resigning pre-emptively are smaller than for leaving the firm after the event occurs, they are still
much stronger than staying at the firm through the event. Therefore, the labor market appears effective in providing ex-post
accountability even if a director tries to abandon a firm ahead of difficult times.
Y. Dou / Journal of Accounting and Economics 63 (2017) 161–178 163
2. Data and sample
To compare the labor market consequences among directors that resign before, after, and stay during a negative event, I
examine the change in the number of board seats for each director within the three years after the negative event. A greater
board seat reduction for resigning directors would be evidence that these directors receive penalties from the labor market
for abandoning the firm. The first step in constructing the sample is to define negative events. In this paper, negative events
include class action lawsuits, earnings restatements, dividend reductions of 10% or more, and debt covenant violations.
These events are generally viewed as detrimental to director reputation.1
The class action lawsuit data are from the Securities Class Action Clearinghouse (SCAC); the earnings restatement data
are from the US Government Accountability Office (GAO); the dividend reductions are calculated based on data from
COMPUSTAT; and the debt covenant violation data are provided by Roberts and Sufi (2009).2 After obtaining these four
event samples, I pool them together to construct a “combined” sample which is likely to be more representative. If a firm
appears in the sample more than once in any two consecutive years, I retain only the first appearance. Finally, I exclude
financial and utility firms from the sample because these firms are highly regulated.
Thereafter, I merge the event sample with my director data, taken from RiskMetrics, and classify the independent di-
rectors in these firms into those who resign before the event, those who depart after the event, and those who stay during
the event.3 The main focus of this paper is on directors who resign during the period when they are most needed. Therefore,
the resignation should occur (1) before the public learn about the negative event (motivated by the reputational concern),
and (2) before/during the period when the workload gets heavy (motivated by time constraints). As such, I define Resigning
Ex-Ante directors as those who no longer work for a negative-event firm based on the last annual proxy statement before the
negative event. I exclude directors who cannot be found in the database after the departure because these departures are
likely driven by deaths or complete retirements. Including these directors can favor finding a significant decline of board
seats for Resigning Ex-Ante directors.
Further, I define Departing Ex-Post directors as those who leave a negative-event firm within the three years after the
event. The main distinction between a Resigning Ex-Ante and a Departing Ex-Post director is whether the director leaves the
firm before or after the negative event. Lastly, I define staying directors as those who joined the firm before the event and
are still present three years after the event. I recognize that the Departing Ex-Post directors who leave in the second or third
year after the event are also likely to leave for other reasons. In unreported results, I classify directors who leave a negative-
event firm in the second or third year after the event as staying directors, rather than Departing Ex-Post directors, and find
results that are quantitatively similar to the baseline results in Table 3.
The coverage of RiskMetrics starts in 1996 but is limited to S&P 1500 firms. To maximize the sample, I merge negative-
event firms that cannot be merged with RiskMetrics to BoardEx and Corporate Library, which cover a broader range of firms
since 2003 and 2001, respectively. Identifying Resigning Ex-Ante directors requires the availability of director data at least
two years before the negative event, while identifying directors' subsequent board seats requires the availability of director
data three years after the negative event. The sample period for the director data (1996–2011) consequently limits the
negative-event sample period to 1998–2008. The only exception is the earnings restatements sample, whose coverage by
the GAO ends in 2006. In addition, I exclude firms that have missing values in Compustat or CRSP.
The last step in constructing the analysis is to collect the number of board seats each director has in the event year (Year 0)
and three years after the event (Year 3). A detailed description of the steps taken to collect this information is in Appendix A.
On average, 30% of the independent directors are excluded from the sample due to having missing information in board seats.
Table 1 reports the number of events, the number of Resigning Ex-Ante directors, the number of Departing Ex-Post directors, and
the number of staying directors each year. Based on the combined sample, there are 1,202 negative events (from 1,014 unique
firms), 387 Resigning Ex-Ante directors, 2,227 Departing Ex-Post directors, and 4,071 staying directors.
Table 2 presents the sample summary statistics and univariate results. For brevity, I only report the results based on the
combined sample of all negative events. Results based on the individual samples are quantitatively similar. Panel A of Table 2
reports firm characteristics. Financial variables such as total assets and return on assets (ROA) are from Compustat, while
stock returns are from CRSP. Corporate governance variables such as board size, board independence, CEO duality indicator,
presence of classified boards, and busy board indicator are from the director databases, while information on institutional
ownership is from the Institutional Holdings data by Thomson Reuters.4
1
For a few examples, see Fich and Shivdasani (2007), Masulis et al. (2012), Kaplan and Reishus (1990), Nini et al. (2012), Desai et al. (2006), and Masulis
and Mobbs (2016).
2
Hennes et al. (2008) show that the earnings restatements sample provided by the GAO contains both intentional irregularities and unintentional
errors. The authors also show that the negative consequences of earnings restatements are mainly concentrated among intentional irregularities. Therefore,
I only include restatements that are due to irregularities as classified by Hennes, Leone and Miller.
3
Independent directors are the outside directors with no material affiliations with the firm. I only analyze independent directors because their
incentives are more dependent on reputational concerns than those of other directors. Insider and grey directors, on the other hand, may base their
decision to stay at the firm on their employment status and pre-existing familial and business relationships.
4
Data on classified boards are available in RiskMetrics and Corporate Library, but not in BoardEx. Therefore, for BoardEx firms in my sample, I first try
to merge them to the other two databases to obtain this information. For the remaining firms, I merge them to Voting Analytics, which provides director
election data for Russell 3000 firms. Observing whether the entire board is up for election each year enables me to infer whether the board is classified.
Lastly, for firms that still have missing information, I read their proxy filings in the relevant year to collect this data.
Table 1
Number of events, resigning directors, departing directors, and staying directors. This table shows the number of negative events, number of Resigning Ex-
Ante directors, number of Departing Ex-Post directors, and number of staying directors. All directors in this table are independent directors. A Resigning Ex-
Ante director is a director whose name can no longer be found in the last annual proxy statement before the negative event; a Departing Ex-Post director is a
director who leaves a negative-event firm within three years after the event; and a staying director is a director who has joined the firm before the negative
event and is still present at the firm three years after the event. The sample period is 1998–2008, except for the earning restatements sample, whose
coverage by GAO ends in 2006.
Combined Sample Lawsuits Earning Restatements
Year Events Resigning Departing Staying Events Resigning Departing Staying Events Resigning Departing Staying
1998 4 0 6 8 3 0 1 4 1 0 1 0
1999 65 17 111 175 28 1 53 76 6 1 9 19
2000 73 21 136 239 28 6 58 112 5 0 7 12
2001 78 13 134 301 24 1 42 82 9 1 13 38
2002 115 11 199 395 48 4 107 176 14 1 25 44
2003 122 50 267 427 60 18 115 183 19 5 42 60
2004 135 58 291 457 62 14 147 197 22 10 41 76
2005 247 72 452 857 72 11 118 241 53 16 105 194
2006 163 74 297 485 60 12 134 162 23 15 117 186
2007 106 38 189 396 73 18 152 288 – – – –
2008 94 33 145 331 43 10 85 127 – – – –
Total 1202 387 2227 4071 500 96 1011 1649 152 50 361 631
Dividend Reductions Covenant Violations
Year Events Resigning Departing Staying Events Resigning Departing Staying
1998 0 0 0 0 3 0 1 5
1999 36 18 55 79 3 0 5 4
2000 44 15 86 134 10 5 7 22
2001 39 13 61 144 10 0 24 28
2002 33 6 41 115 25 0 25 72
2003 28 25 59 126 10 3 26 34
2004 28 16 73 109 13 8 18 22
2005 33 20 53 100 31 5 42 83
2006 48 48 74 124 26 5 46 79
2007 40 25 55 148 30 7 46 92
2008 68 30 99 259 14 3 7 42
Total 394 215 655 1337 177 35 247 483
Panel B reports director characteristics as well as the differences across the three groups of directors. The cross-group
comparison reveals several notable differences. First, compared to Departing Ex-Post directors, Resigning Ex-Ante directors
have longer tenure and more committee memberships. One interpretation of these differences is that the Resigning Ex-Ante
directors tend to be more capable than the Departing Ex-Post directors. Consequently, the better capability makes these
directors better able to relinquish their poorly performing directorships. Alternatively, the longer tenure and greater
committee roles may also mean that these directors would be held more accountable if they were still present when the
event occurs. Ex ante, this concern might give them greater incentives to resign pre-emptively.
On the other hand, the Resigning Ex-Ante directors hold significantly fewer committee memberships than directors who
stay through the events. In fact, among all three types of directors, the staying directors have the longest tenure and greatest
committee commitment. Meanwhile, they hold the lowest number of directorships. These characteristics may suggest that
the board seat at the negative-event firm is more important to the staying directors than it is to the Resigning Ex-Ante and
Departing Ex-Post directors, which is consistent with their decision to stay at the firm and try to fix the problem.
Lastly, Panel B summarizes the number of directorships for each director in Year 0 and Year 3. For Resigning Ex-Ante
directors, the number of directorships at Year 0 is collected after their resignations from the negative-event firms. Overall,
Resigning Ex-Ante directors have slightly more directorships than Departing Ex-Post and staying directors when the event
occurs, although neither difference is statistically significant. For all three types of directors, the average number of board
seats decreases from Year 0 to Year 3, suggesting that the negative events, in general, hurt director reputation. By definition,
Departing Ex-Post directors lose their board seats at the negative-event firm within the three years after the event. Pre-
sumably, these directors will try to add new directorships to their portfolios to compensate for this loss. Nevertheless, the
fact that these directors experience the largest board seat declines indicates that most of them cannot successfully recover
from the directorship loss.
Table 2
Summary Statistics. This table reports sample summary statistics. Panel A reports firm characteristics, while Panel B reports director characteristics and
tests the differences among Resigning Ex-Ante, Departing Ex-Post, and staying directors. *, **, and *** indicate significance at the 10%, 5%, and 1% levels
respectively. Resigning Directors are directors who resign from a negative-event firm in the year before the event occurs; Departing Directors are directors
who leave a negative-event firm within three years after the event; and Staying Directors are directors who joined the firm before the negative event and
are still present at the firm three years after the event. Return on Assets is the operating income before depreciation and amortization divided by total
assets. CEO Duality is an indicator variable that takes the value of one if the CEO and the chairman in the firm are the same person and zero otherwise.
Classified Board is an indicator variable indicating whether the directors of the board are divided into more than one class. Busy Board is an indicator
variable that takes the value of one if more than half of the independent directors of the firm hold three or more directorships. Institutional Ownership is the
total percentage ownership by all institutional shareholders. Audit, Compensation, and Nomination are three indicator variables indicating whether the
director is a member of that committee. Committee Memberships is the total number of memberships the director has among these three Committees.
Panel A: Firm Characteristics
Mean Std Dev P(25) P(75)
Total Assets (in millions) 13.236 41.723 0.986 7.851

Return on Assets 0.132 0.448 0.1 0.21
Prior Year Stock Return 0.135 0.507 0.14 0.388
Board Size 10.276 2.466 9 12
Board Independence 0.739 0.11 0.688 0.813
CEO Duality (0,1) 0.681 0.395 0 1
Classified Board (0,1) 0.556 0.497 0 1
Busy Board (0,1) 0.258 0.355 0 1
Institutional Ownership 0.463 0.301 0.201 0.735
Panel B: Director Characteristics
Mean Std Dev P(25) P(75) Resign Depart Stay R-D R-S D-S
Age 59.625 8.499 54 65 60.044 60.377 59.174 0.333 0.87 1.203***

Tenure 7.364 5.967 3 10 7.612 6.327 7.908 1.285*** 0.295 1.580***
Female (0,1) 0.121 0.327 0 0 0.116 0.098 0.135 0.018 0.018 0.036***
Audit Committee (0,1) 0.472 0.499 0 1 0.393 0.344 0.549 0.048* 0.156*** 0.204***
Compensation Committee (0,1) 0.464 0.499 0 1 0.455 0.362 0.521 0.092*** 0.066** 0.159***
Nomination Committee (0,1) 0.348 0.476 0 1 0.279 0.279 0.392 0 0.113*** 0.113***
Committee Memberships 1.262 0.82 1 2 1.167 1.009 1.409 0.158*** 0.242*** 0.400***
Directorships at Year 0 2.489 1.666 1 3 2.633 2.567 2.433 0.066 0.201 0.135***
Directorships at Year 3 2.245 1.487 1 3 2.22 2.085 2.258 0.135 0.038 0.173*
3. Baseline results
To examine whether the labor market penalties vary across different types of directors, I estimate the following baseline
regression:
Directorship Change ¼ α þ β1 Resigning Ex-Anteþ β2 Departing Ex-Postþf(controls) þ ε
The two key independent variables, Resigning Ex-Ante and Departing Ex-Post, are indicator variables that take the value of
one for Resigning Ex-Ante and Departing Ex-Post directors respectively. The omitted group is the staying directors. Existing
findings in the literature, which focus on the comparison between Departing Ex-Post and staying directors, suggest that the
coefficient of Departing Ex-Post should be significantly negative. Based on the arguments in this paper, I expect the coef-
ficient of Resigning Ex-Ante to also be negative and significant, indicating that the Resigning Ex-Ante directors are penalized
for abandoning the firm. In addition, I expect the coefficient of Resigning Ex-Ante to be of a smaller magnitude than that of
Departing Ex-Post. Even though both types of directors leave the firm, the pre-emptive resignations may help these directors
to avoid certain embarrassments such as receiving a low vote at elections, or being targeted by shareholder activists.
I control for director characteristics (age, gender, number of board seats, tenure, and number of committee member-
ships), firm characteristics (firm size, ROA, and stock returns), and corporate governance characteristics. Corporate gov-
ernance dimensions include board size, the fraction of independent directors on the board, whether the CEO is also the
board chairman, whether the firm has a classified board, whether the board is likely to be busy, and the level of institutional
ownership. Studies have shown that these dimensions can impact firm performance, which plays an important role in
determining a director's reputation in the labor market.5 All regressions include both industry (2-digit SIC) fixed effects and
year fixed effects. Following Petersen (2008), I cluster standard errors both at the firm level and the year level.
Table 3 reports the regression results. In Column 1, the coefficients of Resigning Ex-Ante and Departing Ex-Post are both
5
For examples, see Coles et al. (2008) for board size; Adams and Ferreira (2007) and Duchin et al. (2010) for board independence; Brickley et al. (1999)
and Goyal and Park (2002) for CEO duality; Bebchuk and Cohen (2005) and Faleye (2007) for classified boards; Fich and Shivdasani (2006) and Field et al.
(2013) for busy directors, and Gompers and Metrick (2001) and Chen et al. (2007) for institutional ownership.
Table 3
Changes in directorships. This table shows regression results of directorship changes for Resigning Ex-Ante, Departing Ex-Post, and staying directors. The
dependent variable is the change in number of directorships from Year 0 to Year 3. The two key independent variables, Resigning Ex-Ante and Departing Ex-
Post, are both indicator variables. Resigning Ex-Ante takes the value of one for directors who resign from a negative-event firm in the year before the event
occurs; and Departing Ex-Post takes the value of one for directors who leave a negative-event firm within three years after the event. The omitted group is
the staying directors who joined the firm before the negative event and are still present at the firm three years after the event. Committee Memberships is
the total number of committee memberships a director has. CEO Duality is an indicator variable that takes the value of one if the CEO and the chairman in
the firm are the same person and zero otherwise. Classified Board is an indicator variable indicating whether the directors of the board are divided into
more than one class. Busy Board is an indicator variable that takes the value of one if more than half of the independent directors of the firm hold three or
more directorships. Institutional Ownership is the total percentage ownership by all institutional shareholders. All regressions include both industry (2-digit
SIC) fixed effects and year fixed effects. Standard errors are clustered at both the industry level and year level. Beneath each coefficient estimate is its t-
value. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively.
(1) (2)
Resigning Ex-Ante 0.279*** 1.822*

( 4.751) ( 1.768)
Departing Ex-Post 0.476*** 0.641*
( 10.484) ( 1.650)
Age 0.012*** 0.012***
( 10.397) ( 9.313)
Female 0.004 0.001
( 0.137) (0.028)
Directorships at Year 0 0.345*** 0.296***
( 14.663) ( 10.478)
Tenure 0.011*** 0.008***
( 5.672) ( 3.838)
Committee Memberships 0.031** 0.013
( 2.140) (0.843)
Ln(Total Assets) 0.033*** 0.038***
(4.144) (3.086)
Return on Assets 0.053 0.057
(1.313) (1.469)
Event-Year Stock Return 0.014* 0.026*
(1.704) (1.776)
Board Size 0.011** 0.007
( 2.301) ( 1.380)
Board Independence 0.046 0.119
(0.463) (1.073)
CEO Duality 0.079*** 0.070***
( 3.477) ( 3.125)
Classified Board 0.035 0.042*
( 1.265) ( 1.701)
Busy Board 0.084 0.083
(1.260) (1.352)
Institutional Ownership 0.089** 0.102**
( 2.118) ( 2.440)
Directorships*Resigning 0.054
(0.831)
Directorships*Departing 0.146***
( 5.167)
Tenure*Resigning 0.017**
( 2.020)
Tenure*Departing 0.004
( 1.068)
Committees*Resigning 0.068
( 0.810)
Committees*Departing 0.058
( 1.629)
Assets*Resigning 0.092**
( 2.070)
Assets*Departing 0.030*
( 1.728)
Returns*Resigning 0.092
( 0.520)
Returns*Departing 0.005
(0.146)
Resigning ¼Departing t-test 14.84 2.52
Resigning ¼Departing p-value 0.000 0.113
Observations 6685 6685
Adjusted R-squared 0.328 0.344
Fixed Effects Industry, Year
significantly negative. For Resigning Ex-Ante directors, a coefficient of 0.28 means that they suffer an incremental loss of
0.28 directorships compared to the directors who have stayed through the events. This result supports the argument that
the labor market penalizes directors for abandoning a troubled firm. As shown in Table 2, Resigning Ex-Ante directors hold
2.6 board seats in Year 0. Therefore, this additional penalty represents a 10.8% further reduction in their number of board
seats, which is economically significant. This number can only under-estimate the magnitude of the penalties, as there are
also other benefits from holding a board seat that are being lost such as social connections with other directors.
Further, the coefficient of Departing Ex-Post is 0.48, which indicates that the penalties for directors who leave a ne-
gative-event firm after the event are strongest among all three groups of directors. For these directors, a coefficient of 0.48
represents an 18.5% further board seat reduction relative to staying directors. This coefficient has a much greater magnitude
(71% higher) than Resigning Ex-Ante; and the difference is statistically significant. Therefore, even though the labor market
penalizes directors for resigning pre-emptively from a troubled firm, the penalties are still smaller than if the director leaves
the firm after the event occurs.
There are a number of reasons why Resigning Ex-Ante directors face smaller penalties relative to Departing Ex-Post di-
rectors. As mentioned earlier, this result may reflect the additional reputational damage that Departing Ex-Post directors
suffer from events such as receiving a low election vote, being named in a lawsuit, or being targeted in a proxy contest. Also,
some Departing Ex-Post directors may initially try to stay and fix the problem, but later on decide to leave. In this case, these
“under-crisis” resignations could put the firm in an even more difficult situation than the pre-emptive resignations. This can
further explain the greater penalties that Departing Ex-Post directors receive.6
Among the control variables, director age and tenure appear to negatively impact directorship change. In addition,
consistent with Yermack (2004), the number of directorships a director already has at the time of the event appears to have
a significantly negative impact on the director's subsequent directorships. This negative relation is intuitive, as a director
with many board seats should have a lower probability of taking another directorship than a non-busy director.
I also control for the firm's pre-event ROA and event-year stock return in the regression. ROA aims to capture firms' pre-event
performance, while the stock return during the event-year reflects the magnitude of the damage brought by the event. According
to the results, the stock return in the event-year tends to positively affect a director's subsequent board seat change. Lastly,
among the corporate governance variables, the results suggest that smaller boards and the absence of CEO-Chairman duality,
which are generally viewed as signals of strong governance, can positively affect directors' board seats. On the other hand, the
relation between the number of board seats and the institutional ownership appears to be significantly negative.
To better understand how certain director characteristics can affect the changes in board seats for Resigning Ex-Ante,
Departing Ex-Post, and staying directors, in Column 2, I interact the Resigning Ex-Ante indicator and the Departing Ex-Post
indicator with the number of directorships, director tenure, number of committee memberships, firm size, and the event-
year stock return respectively. According to the results, the labor market penalties for both Resigning Ex-Ante and Departing
Ex-Post directors tend to be heavier if they leave a large firm. Since larger firms tend to be more “visible”, this result is
consistent with the intuition that directors who leave these firms draw more attention and therefore suffer stronger pe-
nalties. Additionally, a director's tenure and number of board seats tend to magnify the penalties for Resigning Ex-Ante and
Departing Ex-Post directors, respectively.
4. The sources of penalties
Having documented the labor market consequences for different types of directors, in this section I attempt to identify
how these consequences occur. There are at least two effects that can lead to a greater overall directorship decline for
Resigning Ex-Ante (and Departing Ex-Post) directors. First, while the negative events reveal the monitoring failure of all the
directors associated with the troubled firms, the departures that take place shortly before (i.e., Resigning Ex-Ante directors)
and after (i.e., Departing Ex-Post directors) the events further reveal that these directors are either abandoning a troubled
firm, or being forced to leave due to being accountable for the events. In this case, firms where these directors are still board
members may have stronger incentives to remove these directors from their boards, compared with firms where the staying
directors are board members.
Second, while Resigning Ex-Ante, Departing Ex-Post, and staying directors all suffer a reputational damage due to the
monitoring failure, it is possible that staying directors are better able to obtain new board seats relative to those who leave
before or during the events. For example, a firm that is currently facing a class action lawsuit may find it useful to appoint to
its board a director who has dealt with a lawsuit in the past. Such preference towards staying directors can be because
staying directors accumulate certain experience in managing the crisis which these firms value. Alternatively, these firms
might believe that the staying directors are inherently more competent than the Resigning Ex-Ante and Departing Ex-Post
directors. In either case, this effect can further contribute to the greater directorship declines for Resigning Ex-Ante and
Departing Ex-Post directors. I examine these two possibilities separately.
6
There can be several alternative explanations to the baseline results, which are discussed in Section 6.
4.1. Losing existing board seats
To examine whether the first conjecture holds, I gather the remaining board seats that Resigning Ex-Ante, Departing Ex-
Post, and staying directors hold in the year the events occur, and construct an indicator variable that equals one if the
director is no longer a member of that board three years after the event, and zero otherwise. This indicator variable is the
outcome variable in this sub-section. I expect the coefficients of Resigning Ex-Ante and Departing Ex-Post to be significantly
positive, indicating that, compared to staying directors, Resigning Ex-Ante and Departing Ex-Post directors are more likely to
lose their other existing board seats.
The results are reported in Column 1 of Table 4. I estimate a logit regression with industry and year fixed effects.
Consistent with the above prediction, the coefficients of Resigning Ex-Ante and Departing Ex-Post are both significantly
positive, suggesting that these directors face a higher likelihood of removal from their remaining board seats. Even though
Resigning Ex-Ante once again has a smaller magnitude than Departing Ex-Post, this difference is not statistically significant,
suggesting that the timing of the departure does not affect the director's likelihood of subsequently being removed from
other boards.
In Column 2, I repeat the methodology in Table 3 and interact the two key independent variables with several director
and firm characteristics. Overall, the results suggest that having a long tenure and having a large number of committee
memberships increase the likelihood that a Resigning Ex-Ante or Departing Ex-Post director leaves a board. Since both tenure
and committee membership reflect a director's relative importance to the board, one interpretation of the result is that firms
are more likely to remove these directors when they play important roles on the boards. Lastly, the likelihood of removal for
Departing Ex-Post directors is also higher at large firms.
Apart from comparing Resigning Ex-Ante directors' likelihood of losing other board seats to Departing Ex-Post and staying
directors, I also compare them to the board members from the Resigning Ex-Ante directors' other board seats in Year 0. One
concern with the comparison made in Columns 1–2 is that, because the analysis examines directors' other board seats, the
Resigning Ex-Ante, Departing Ex-Post, and staying directors are not necessarily sitting on the same boards in this sample.
Therefore, the results from this cross-board comparison could be driven by certain unobservable factors at the board level. In
light of this, in Columns 3 and 4, I compare Resigning Ex-Ante directors to the “local” directors that are currently sitting on
the same boards with the Resigning Ex-Ante directors. Such a within-board comparison enables me to replace industry fixed
effects with firm fixed effects which control for more unobservable effects. To accommodate the use of firm fixed effects, I
also replace the logit regressions with OLS regressions.
According to the results in Columns 3, the coefficient of Resigning Ex-Ante remains significantly positive, suggesting that
Resigning Ex-Ante directors are more likely to leave their other boards even when compared to the rest of the directors on
the same boards. In Column 4, I once again interact Resigning Ex-Ante with several director characteristics. The results
suggest that having a larger number of directorships or committee memberships tends to increase the likelihood that a
Resigning Ex-Ante director leaves the board.
In Columns 5 and 6, to gain further insights into the relative penalties between Resigning Ex-Ante and Departing Ex-Post
directors, I apply the same approach to Departing Ex-Post directors. Specifically, within all the board seats that Departing Ex-
Post directors hold in Year 0 (excluding the negative-event firm), I compare the likelihood that these directors leave this
board in the next three years to the rest of the directors on the board. As the results in Column 5 show, Departing Ex-Post
directors have a significantly higher likelihood to also lose these board seats compared to the remaining board members.
Also, the coefficient of Departing Ex-Post from Column 5 has a much greater magnitude than the coefficient of Resigning Ex-
Ante in Column 3, suggesting once again that Departing Ex-Post directors receive stronger penalties relative to Resigning Ex-
Ante directors.
4.2. Acquiring new board seats
The second channel predicts that firms that try to deal with a negative event prefer to appoint directors who have stayed
through similar events in the past. To test this conjecture, for each Resigning Ex-Ante, Departing Ex-Post, and staying director
in my negative-event firm sample, I examine whether he/she acquires any subsequent similar-event directorships. To qualify
as a similar-event directorship, the director appointment has to meet two conditions. First, the firm appointing this director
must experience the same type of negative event that the director was initially associated with in my sample.7 Second, the
appointment must occur in the year immediately before, the year of, or the year immediately after the event.
After collecting the number of similar-event directorships that each director acquires after the initial negative event, I
regress this value on the two key independent variables, Resigning Ex-Ante and Departing Ex-Post, along with other control
variables. The results are reported in Table 5 Column 1. Both the coefficients of Resigning Ex-Ante and Departing Ex-Post are
negative and significant, which suggests that directors who have stayed through a negative event appear to be more de-
sirable to firms facing similar events compared to those who leave ex-ante or ex-post. Further, the coefficients of Resigning
Ex-Ante and Departing Ex-Post are similar in magnitude and are not significantly different, suggesting that whether a director
7
For example, if a director previously associated with an earnings restatement acquires a new directorship at a firm that is experiencing debt
covenants violations, I do not define this as a similar-event directorship.
Table 4
Other existing board seats. This table shows regression results of a director's likelihood of being removed from the non-negative firm directorships he/she
holds at the time of the negative event. The dependent variable for all columns is an indicator variable which takes the value of one if the director leaves
this board within three years after the negative event, and zero otherwise. In Columns 1 and 2, the sample focuses on all the non-negative firm board seats
held by all the Resigning Ex-Ante, Departing Ex-Post, and staying directors identified from the negative-event firms. The regressions in Columns 1 and 2 are
logit regressions with industry-fixed effects and year-fixed effects. In Columns 3 and 4, the sample does not include Departing Ex-Post and staying directors,
but instead includes the directors who are members of the non-negative firm board seats held by the Resigning Ex-Ante directors. The Resigning Ex-Ante
indicator variable in Columns 3 and 4 is equal to one for Resigning Ex-Ante directors, and zero for these “local” directors. Similarly, in Columns 5 and 6, the
sample includes the directors who are members of the non-negative firm board seats held by the Departing Ex-Post directors. The Departing Ex-Post
indicator variable in Columns 5 and 6 is equal to one for Departing Ex-Post directors, and zero for the “local” directors. To better address un-observable
factors, the regressions in Columns 3–6 include firm-fixed effects and year-fixed effects and are OLS regressions. Independent is an indicator variable that
equals one for independent directors, and zero otherwise. CEO Duality is an indicator variable that takes the value of one if the CEO and the chairman in the
firm are the same person and zero otherwise. Classified Board is an indicator variable indicating whether the directors of the board are divided into more
than one class. Busy Board is an indicator variable that takes the value of one if more than half of the independent directors of the firm hold three or more
directorships. Institutional Ownership is the total percentage ownership by all institutional shareholders. Standard errors are clustered at the industry and
year level for Columns 1 and 2, and are clustered at the firm and year level for Columns 3–6. Beneath each coefficient estimate is its t-value. *, **, and ***
indicate statistical significance at the 10%, 5%, and 1% levels respectively.
Resigning, Departing, and Staying Resigning and Local Departing and Local
(1) (2) (3) (4) (5) (6)
Resigning Ex-Ante 1.025*** 0.854* 0.069*** 0.162*

(9.974) (1.820) (3.685) (1.848)
Departing Ex-Post 1.118*** 1.892** 0.324*** 0.371***
(16.776) (2.528) (29.877) (2.871)
Age 0.026*** 0.024*** 0.006*** 0.006*** 0.005*** 0.005***
(5.979) (5.409) (6.735) (6.818) (6.190) (6.194)
Female 0.069 0.060 0.037** 0.037** 0.021 0.021
(0.852) (0.743) ( 2.180) ( 2.181) ( 1.056) ( 1.049)
Directorships at Year 0 0.005 0.040* 0.011*** 0.013*** 0.010** 0.010**
(0.317) (1.885) ( 2.905) ( 3.414) ( 2.416) ( 2.377)
Tenure 0.013*** 0.007 0.005*** 0.005*** 0.005*** 0.005***
(2.748) (1.194) (4.680) (4.377) (4.626) (4.690)
Independent 0.137* 0.142* 0.003 0.005 0.024 0.024
(1.657) (1.732) (0.213) (0.331) ( 1.354) ( 1.356)
Committee Memberships 0.224*** 0.334*** 0.053*** 0.057*** 0.067*** 0.069***
( 6.247) ( 7.590) ( 6.668) ( 6.885) ( 7.448) ( 7.523)
Ln(Total Assets) 0.159*** 0.111*** 0.036 0.035 0.036 0.037
(6.479) (3.951) (1.351) (1.317) (0.623) (0.641)
Return on Assets 0.855*** 0.827*** 0.079 0.076 0.319*** 0.319***
( 3.873) ( 3.877) (0.247) (0.235) (2.607) (2.612)
Event-Year Stock Return 0.210*** 0.326*** 0.003 0.005 0.032 0.032
( 2.811) ( 3.626) (0.080) ( 0.109) ( 1.052) ( 1.036)
Board Size 0.120*** 0.121*** 0.048 0.048 0.018* 0.018*
( 9.747) ( 9.751) (1.503) (1.360) (1.736) (1.749)
Board Independence 1.477*** 1.443*** 0.209 0.204 0.662*** 0.661***
(5.420) (5.259) ( 1.252) ( 1.219) ( 3.016) ( 3.009)
CEO Duality 0.043 0.036 0.008 0.006 0.040 0.040
( 0.588) ( 0.502) (0.187) (0.122) ( 0.857) ( 0.861)
Classified Board 0.017 0.012 0.159*** 0.157*** 0.265*** 0.265***
( 0.258) ( 0.173) (3.336) (3.255) ( 3.458) ( 3.464)
Busy Board 0.326*** 0.330*** 0.037 0.038 0.076** 0.076**
( 4.669) ( 4.717) (0.923) (0.947) (1.995) (2.004)
Institutional Ownership 0.045 0.036 0.113 0.122 0.145** 0.144**
(0.395) (0.314) ( 0.663) ( 0.720) ( 2.047) ( 2.042)
Directorships*Resigning 0.040 0.015*
( 0.899) (1.737)
Directorships*Departing 0.060 0.001
( 1.108) ( 0.261)
Tenure*Resigning 0.030** 0.002
(1.979) (0.672)
Tenure*Departing 0.012 0.002
(1.368) ( 1.229)
Committees*Resigning 0.326*** 0.045**
(3.006) (2.228)
Committees*Departing 0.264*** 0.018**
(3.791) (2.220)
Assets*Resigning 0.063
(1.335)
Assets*Departing 0.126***
(3.818)
Table 4 (continued )
Resigning, Departing, and Staying Resigning and Local Departing and Local
(1) (2) (3) (4) (5) (6)
Returns*Resigning 0.595
(0.940)
Returns*Departing 0.231
(1.581)
Resigning ¼Departing t-test 0.82 0.89
Resigning ¼Departing p-value 0.366 0.346
Observations 11,205 11,205 7853 7853 40,970 40,970
Pseudo R-squared 0.149 0.154
Adjusted R-squared 0.161 0.162 0.239 0.239
Fixed Effects Industry, Year Firm, Year
leaves before or after a negative event does not affect a subsequent troubled firm's propensity to appoint him/her as a board
member.
The sample in Column 1 considers director appointments in any of the three years around the event in the appointing
firm. In Columns 2–4, I also divide the sample based on whether the director appointment occurs in Year t 1, t, or tþ1. As
the results indicate, the coefficient of Resigning Ex-Ante is significantly negative in the Pre-Event Year sample and the Post-
Event Year sample, while the coefficient of Departing Ex-Post is significantly negative in the Pre-Event Year and the Event-
Year sample. Overall, the evidence in this section supports the argument that staying directors appear relatively more
valuable to firms facing a negative event, which is likely due to the experience they accumulate or the greater ability they
possess.
5. Other dimensions of penalties
5.1. Committee memberships/chairmanships
This section examines whether the consequences for resigning pre-emptively are also reflected in certain other di-
mensions. Apart from board seats, the consequences for Resigning Ex-Ante directors might also be reflected in the number of
committee memberships/chairmanships. When the firm from which a director has just resigned experiences a negative
event, all other boards on which this director is still a board member learn about this director's quality. While some of these
boards may find it optimal to remove this director (as documented in Section 4.1), some other boards may find it optimal to
only remove the director from certain committees. This might be the case when firing a director is costly (e.g., when finding
a replacement director is difficult).
The same argument applies to directors who leave a troubled firm shortly after the event occurs. In addition, because
finding a replacement board member can be more difficult after the event occurs, boards where the Departing Ex-Post
directors are still members may have even stronger incentives to remove these directors from important committees
compared to boards where the Resigning Ex-Ante directors are members. Therefore in this sub-section, I test whether,
compared to staying (Departing Ex-Post) directors, Resigning Ex-Ante directors are more (less) likely to lose committee
memberships/chairmanships on their remaining boards in the three years after the negative events.
I start with all other board seats that Resigning Ex-Ante, Departing Ex-Post and staying directors have in Year 0. I then
exclude the observations where the director is no longer with the board, or the firm is no longer covered in the data three
years after the event. Within each of the remaining board seats, I count the number of memberships and chairmanships
each director has in Year 0 and Year 3, and take the differences as two separate dependent variables. Due to data availability,
I only focus on the three major monitoring committees including the audit, compensation, and nomination committee.
The results are reported in Table 6. In Column 1 where the dependent variable is the change in committee memberships,
the coefficient of Resigning Ex-Ante is insignificant. However, in Colum 2 where the dependent variable is the change in
committee chairmanships, this coefficient is negative and significant at the 1% level. Meanwhile, the coefficient of Departing
Ex-Post is significantly negative in both columns, consistent once again with the earlier findings that these directors receive
the strongest penalties. Therefore, the results in this sub-section indicate that, compared to staying directors, Resigning Ex-
Ante directors do not experience a significantly greater decline in the number of committee memberships after the event.
However, they tend to chair significantly fewer committees. On the other hand, Resigning Ex-Ante directors are better off
than Departing Ex-Post directors as they tend to experience smaller committee membership reductions. In terms of the
committee chairmanship reductions, the difference between Resigning Ex-Ante and Departing Ex-Post directors is not sig-
nificantly different.
Table 5
Acquiring new directorships. This table reports the results of estimating the effects of being associated with a negative event on the director's frequency of
acquiring similar-event directorships. Similar-event directorships refer to the new director appointments that are made in the year immediately before, the
year of, or the year immediately after the appointing firm experiences a negative event, and where the joining director was a Resigning Ex-Ante, Departing
Ex-Post, or staying director from another firm which experienced the same type of negative event in the past. In Column 1, the dependent variable is the
total number of similar-event directorships that each director acquires. In Columns 2–4, the dependent variables are the numbers of similar-event di-
rectorships that each director acquires in the pre-event year, event year, and post-event year, respectively. All dependent variables are multiplied by 100.
Independent is an indicator variable that equals one if the director joins as an independent director, and zero otherwise. CEO Duality is an indicator variable
that takes the value of one if the CEO and the chairman in the firm are the same person and zero otherwise. Classified Board is an indicator variable
indicating whether the directors of the board are divided into more than one class. Busy Board is an indicator variable that takes the value of one if more
than half of the independent directors of the firm hold three or more directorships. Institutional Ownership is the total percentage ownership by all
institutional shareholders. Standard errors are clustered at both the industry and year level. Beneath each coefficient estimate is its t-value. *, **, and ***
indicate statistical significance at the 10%, 5%, and 1% levels respectively.
Joins in: Three-Year Period Pre-Event Year Event Year Post-Event Year
(1) (2) (3) (4)
Resigning Ex-Ante 2.641*** 1.826*** 0.403 0.638***

( 3.107) ( 2.996) ( 1.017) ( 3.284)
Departing Ex-Post 2.704* 3.155** 0.950** 0.233
( 1.853) ( 2.518) ( 2.253) ( 0.717)
Age 0.051* 0.029* 0.000 0.012
( 1.884) ( 1.796) (0.008) ( 0.966)
Female 0.012 0.390 0.012 0.309
( 0.017) ( 0.936) (0.041) (0.626)
Directorships at Year 0 0.752** 0.349 0.442*** 0.064
(2.071) (1.339) (3.084) (1.091)
Independent 0.226** 0.117 0.061** 0.079***
( 2.423) ( 1.475) ( 2.028) ( 3.964)
Committee Memberships 0.010 0.362** 0.115 0.376***
(0.028) ( 1.996) ( 0.720) (2.675)
Ln(Total Assets) 0.071 0.268 0.105 0.012
(0.240) (0.948) ( 0.958) (0.113)
Return on Assets 3.572** 2.723*** 1.439*** 0.055
(2.523) (2.837) (2.735) (0.118)
Event-Year Stock Return 0.116 0.248 0.121 0.229
(0.142) (0.283) (0.479) ( 0.999)
Board Size 0.061 0.188* 0.042 0.097**
( 0.384) ( 1.857) ( 0.766) (2.034)
Board Independence 5.706*** 3.362** 1.650 1.178
(2.716) (2.141) (1.163) (1.017)
CEO Duality 0.087 0.395 0.190 0.361
(0.150) ( 1.164) ( 0.812) (1.097)
Classified Board 0.738 0.376 0.110 0.149
( 0.895) ( 0.674) ( 0.324) ( 0.661)
Busy Board 2.020 1.669 0.492 0.343
(0.949) (1.043) (1.023) (0.646)
Institutional Ownership 1.060 0.862 0.345 0.098
( 0.948) ( 0.800) ( 0.788) ( 0.218)
Resigning¼ Departing t-test 0.01 3.63 1.24 2.69
Resigning¼ Departing p-value 0.932 0.057 0.266 0.101
Observations 9107 9107 9107 9107
5.2. Directorship prestige
In this sub-section I test if the reputational penalties exist not only in the quantity of the directorships, but also in the
quality of these directorships. If a director is willing to forgo two board seats in small firms to obtain one board seat in a
large firm, then the observed directorship decline does not necessarily indicate a penalty. Following recent studies such as
Bereskin and Smith (2014) and Masulis and Mobbs (2014), I use firm size as a measure of directorship prestige. Specifically,
for each director, I measure the change in prestige as the percentage change in the total assets of the largest firm this
director works for from Year 0 to Year 3, excluding the initial negative-event firm. I only focus on the largest firm the
director works for rather than taking the average or median value across all directorships to avoid a reduction in the prestige
measure due to obtaining additional smaller directorships.
Column 3 of Table 6 reports the results. The Resigning Ex-Ante and Departing Ex-Post indicators are negative and sig-
nificant at the 1% level and 10% level respectively. The two coefficients are not significantly different. Therefore, the evidence
presented here indicates that both leaving a firm shortly before and after a negative event decrease the prestige of a
director's directorship portfolio.
Table 6
Changes in committee memberships, chairmanships, and directorship prestige. This table shows regression results of the changes in committee mem-
berships, chairmanships, and directorship prestige for Resigning Ex-Ante, Departing Ex-Post, and staying directors. The samples for Columns 1 and 2 are at
the directorship level, excluding the negative-event firm; while the sample for Column 3 is at the director level. In Column 1, the dependent variable is the
change in the number of committee memberships from Year 0 to Year 3. In Column 2, the dependent variable is the change in the number of committee
chairmanships from Year 0 to Year 3. In Column 3, the dependent variable is the percentage change in the total assets of the largest firm this director works
for from Year 0 to Year 3, excluding the initial negative event firm. CEO Duality is an indicator variable that takes the value of one if the CEO and the
chairman in the firm are the same person and zero otherwise. Classified Board is an indicator variable indicating whether the directors of the board are
divided into more than one class. Busy Board is an indicator variable that takes the value of one if more than half of the independent directors of the firm
hold three or more directorships. Institutional Ownership is the total percentage ownership by all institutional shareholders. All regressions include both
industry (2-digit SIC) fixed effects and year fixed effects. Standard errors are clustered at both the industry level and year level. Beneath each coefficient
estimate is its t-value. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively.
Membership Chairmanship Size

(1) (2) (3)
Resigning Ex-Ante 0.015 0.094*** 6.531***

( 0.271) ( 4.106) ( 2.766)
Departing Ex-Post 0.145*** 0.115*** 5.677*
( 2.590) ( 4.960) ( 1.673)
Age 0.001 0.000 0.013
(1.236) (0.469) ( 0.139)
Female 0.029 0.019 10.194
(1.009) ( 1.029) (0.888)
Tenure 0.011*** 0.004*** 0.262
( 8.833) ( 7.326) ( 0.492)
Memberships at Year 0 0.417***
( 12.309)
Chairmanships at Year 0 0.483***
( 27.576)
Directorships at Year 0 2.615***
( 3.243)
Ln(Total Assets) 0.003 0.005 4.092*
( 0.278) ( 1.542) (1.801)
Return on Assets 0.066 0.052 0.701
(1.234) (1.588) ( 0.144)
Event-Year Stock Return 0.084*** 0.012 1.691
(4.099) ( 0.596) (0.440)
Board Size 0.026*** 0.007*** 1.979*
( 4.593) ( 6.145) ( 1.779)
Board Independence 0.227* 0.045 25.742
(1.709) ( 1.572) (1.044)
CEO Duality 0.006 0.009 0.053
( 0.277) (1.067) (0.011)
Classified Board 0.037** 0.002 2.078
(1.981) (0.195) ( 0.481)
Busy Board 0.010 0.011 6.355
( 0.389) (1.373) (1.088)
Institutional Ownership 0.039 0.006 2.908
( 1.019) (0.317) ( 1.017)
Resigning ¼Departing 15.15 0.90 0.14
t-test
Resigning ¼Departing 0.000 0.343 0.709
p-value
Observations 8429 8429 4078
Adjusted R-squared 0.255 0.262 0.010
5.3. Market reaction to negative events announcements
The third dimension I examine is the market reaction when the negative event is reported to the market. Such an-
nouncements communicate two pieces of information to the public. First, the public learns that the directors associated
with these firms, both those who are presently on the board and those who left a short period ago, are responsible for the
events. Second, for firms where the Resigning Ex-Ante directors of the negative-event firms are still board members, the
event announcement also provides a signal that the director resignation which preceded the event was likely to be driven by
the director's intention to abandon a troubled firm. Since the results in Section 4.1 suggest that these firms tend to sub-
sequently remove Resigning Ex-Ante directors from their boards, it is important to show that these firms have reacted
negatively to the events in the first place.
To qualify as a valid market reaction test, the announcement of an event needs to satisfy two conditions. First, the
announcement needs to “surprise” the market in the sense that the market was not already expecting such an an-
nouncement. Second, the announcement must not come with other confounding information which also has the potential
to surprise the market. In my sample, information on dividend reductions and covenants violations is from firms' annual or
quarterly reports which contain a large amount of additional information, making these events inappropriate for event
studies. The lawsuits sample and earnings restatements sample, which are from the SCAC and GAO respectively, do not have
this weakness. However, as discussed by Karpoff et al. (2017), these databases are associated with a number of issues, one of
which is that the dates of the events recorded by these databases are either the dates that the lawsuits are filed (as in the
SCAC database), or the dates that the restatements are announced (as in the GAO database). In many cases, these dates occur
after the dates that the public initially becomes aware of the ongoing negative events. Therefore, directly using the event
dates from these databases is problematic as the market may have already become aware of these events through other
channels by this time, such as media coverage.
To overcome this issue, I conduct the event study based on the data from the Federal Securities Regulation (FSR) database
collected and shared by Karpoff et al. (2017). These data, which are used in Karpoff et al. (2008a, 2008b), contain manually
collected information on 1099 firms subject to SEC or Department of Justice (DOJ) enforcement action for financial mis-
representation from 1978 to 2011. One of the unique features of this database is that it contains the date of the first public
announcement (in the form of company announcements or media articles) that indicates a potential future enforcement
action. After merging this database to my director data, I identify 66 Resigning Ex-Ante directors who, in the event year, hold
a total of 89 other directorships. These 89 other directorship firms (hereafter “connected firms”) are my primary focus in this
sub-section, and my hypothesis is that, when the negative event is initially revealed to the public, these connected firms
exhibit a negative shareholder reaction because of the presence of a director on their boards who has previously resigned
from the negative-event firm.
To test whether a negative market reaction exists for these connected firms, I first compare the shareholder reaction
among these firms to that of their benchmark firms which do not have director connections to negative-event firms. To
identify benchmark firms, I implement the following criteria. First, for each of these 89 connected firms, I identify all
potential benchmark firms from the same 2-digit SIC industry where none of the board members is associated with a
negative event within their directorship portfolios during that year. Then, I exclude benchmark firms whose total assets are
greater or less than the connected firms' total assets by more than 10%. Thereafter, I calculate the Day-0 abnormal return for
each firm using a market model approach with an estimation window of [ 240, 11]. Day 0 is the date that the market, for
the first time, learns about the negative events. After matching and excluding observations with missing values, the final
sample contains 53 connected firms and 391 benchmark firms.
Column 1 of Table 7 reports the results of regressing the market reaction on an indicator variable Resigning Director
Connection, which equals one for the 53 connected firms and zero for the 391 benchmark firms. The regression includes
industry and year fixed effects. The coefficient of Resigning Director Connection is negative and significant at the 5% level,
indicating that shareholders react negatively to another firm's adverse news announcement if one of their current directors
was a Resigning Ex-Ante director in that firm.
Apart from comparing these connected firms to their non-connected benchmark firms, I also compare them to those
firms that are connected by Departing Ex-Post and staying directors. Even though the results in Column 1 support my
hypothesis, the fact that the benchmark firms do not have negative-event directors on their boards means that the results
could merely be driven by the connection to the negative events, rather than the pre-emptive resignations. To better capture
the effects of the resignation on shareholder reaction, it is important to show that these firms suffer an additional negative
shareholder reaction even when compared to firms that are connected by staying and Departing Ex-Post directors.
To this end, within the FSR sample, I identify the other directorships that Departing Ex-Post and staying directors were
holding at the time of the announcements, and calculate the abnormal returns for these firms. Column 2 of Table 7 com-
pares the abnormal returns of these firms to those of firms connected by Resigning Ex-Ante directors. According to the
results, the coefficients of Resigning Director Connection and Departing Director Connection are both insignificant, suggesting
that the event revelation triggers similar price reactions for firms connected by Resigning Ex-Ante, Departing Ex-Post, and
staying directors. However, the analysis so far treats all directors as equally important to the connected firms. Since the
negative event reveals the Resigning Ex-Ante director's propensity to leave the board prior to difficult times, the extent to
which this revelation hurts firm value can depend on the board's sensitivity to this director's sudden resignation.
To examine whether this conjecture holds, I divide the sample based on the importance of each director to the firm. I
define importance based on each director's number of memberships within the three major monitoring committees. Among
all directors in my sample, the median and mean values of committee memberships are 1 and 1.25 respectively. I therefore
define a director as important if he/she holds two or three memberships, and I expect firms to react more negatively if these
important directors are viewed by shareholders as having the propensity to suddenly resign. According to the results in
Columns 3 and 4 of Table 7, the Resigning Director Connection indicator remains insignificant within directors who hold zero
or one committee membership, but becomes significantly negative for those who sit on two or three committees. Therefore,
the results suggest that firms where Resigning Ex-Ante directors play important roles experience a significant shareholder
wealth reduction following a signal that these directors have a greater propensity to suddenly resign.
Unlike the results presented in all the previous tables, the Departing Director Connection indicator is insignificant in all
columns of this table. This insignificance may reflect the fact that the Departing Ex-Post directors are still present on the
negative-event boards when the initial revelation of the event occurs. Therefore, it is difficult for firms connected by these
Table 7
Market reaction on event revelations. This table examines the market reaction among firms that share common directors with the negative-event firms
when the negative event is revealed to the public for the first time. The dependent variable for all columns is the Day-0 abnormal returns estimated using a
market model approach with an estimation window of [ 240, 11]. In Column 1, the sample includes firms that have a Resigning Ex-Ante director on their
boards and the matched benchmark firms that do not share directors with the negative-event firms. The Resigning Director Connection indicator variable
equals one for the connected firms, and zero for the benchmark firms. In Columns 2–4, the sample includes three types of firms: firms connected by
Resigning Ex-Ante directors (Resigning Director Connection), firms connected by Departing Ex-Post directors (Departing Director Connection), and firms
connected by staying directors (the omitted group). CEO Duality is an indicator variable that takes the value of one if the CEO and the chairman in the firm
are the same person and zero otherwise. Classified Board is an indicator variable indicating whether the directors of the board are divided into more than
one class. Busy Board is an indicator variable that takes the value of one if more than half of the independent directors of the firm hold three or more
directorships. Institutional Ownership is the total percentage ownership by all institutional shareholders. All regressions include both industry (2-digit SIC)
fixed effects and year fixed effects. Standard errors are clustered at both the industry level and year level. Beneath each coefficient estimate is its t-value. *,
**, and *** indicate statistical significance at the 10%, 5%, and 1% levels respectively.
Matched Other Connected Firms
Firms Full Membership o ¼1 Membership 41

(1) (2) (3) (4)
Resigning Director Connection 0.014** 0.001 0.003 0.011***

( 2.382) ( 0.389) (0.762) ( 3.616)
Departing Director Connection 0.001 0.000 0.002
(0.369) (0.159) (0.837)
Age 0.000 0.000 0.000 0.000
( 0.923) ( 0.083) (1.502) ( 1.190)
Female 0.005*** 0.000 0.002 0.002
( 2.923) (0.064) (0.464) ( 0.461)
Directorships 0.002 0.000 0.000 0.001
(1.628) (0.473) ( 0.327) (0.855)
Tenure 0.000 0.000 0.000 0.000
( 0.226) ( 1.229) ( 0.375) ( 1.055)
Committee Memberships 0.001 0.001 0.000 0.004
(0.777) (1.535) ( 0.187) (0.860)
Ln(Total Assets) 0.000*** 0.000 0.000 0.000
(3.722) ( 0.851) ( 0.834) ( 0.801)
Return on Assets 0.012 0.006 0.012 0.012*
(0.384) (0.951) ( 0.817) (1.934)
Event-Year Stock Return 0.008** 0.002 0.002 0.004
( 2.613) ( 0.624) ( 0.448) ( 0.768)
Board Size 0.001 0.000 0.000 0.000
( 0.998) (0.765) (0.566) (0.678)
Board Independence 0.002 0.001 0.007 0.018
( 0.213) (1.567) (0.783) ( 1.465)
CEO Duality 0.001 0.000 0.000 0.002
( 0.432) ( 0.231) (0.219) ( 0.501)
Classified Board 0.001 0.002 0.004** 0.004*
(0.858) (1.170) (2.101) ( 1.973)
Busy Board 0.001 0.001 0.002 0.003
(0.269) ( 0.618) ( 0.747) ( 0.661)
Institutional Ownership 0.002 0.000** 0.015** 0.000
(0.376) (2.211) (2.058) (0.719)
Resigning ¼Departing t-test 0.30 0.54 8.20
Resigning ¼Departing p-value 0.584 0.464 0.005
Observations 444 981 577 404
Departing Ex-Post directors to differentiate them from the staying directors. This can explain the insignificant differences
between the shareholder reaction in firms connected by Departing Ex-Post and staying directors.
5.4. Market reaction to future appointments and departures
Having shown that firms connected by Resigning Ex-Ante directors generally react negatively to the announcements of
the events, I subsequently examine whether Resigning Ex-Ante directors also face a penalty in the form of triggering a
negative (positive) shareholder reaction when they subsequently join (leave) a board. If the public in general views these
directors as undesirable, then it is possible that shareholders view the appointments of these directors as bad news, and vice
versa.
To conduct such a test, I make use of the Key Developments data from Capital IQ, which collects news on material
corporate events that can affect stock prices such as executive/director changes, M&A rumors, SEC inquiries, and many more.
To ensure that the announcement is not confounded by other events, I exclude observations that are accompanied by
another news article covering the same firm within three days before or after the announcement. The types of events I focus
on include “Executive/Board Changes - Other” (Type ID¼16), “Executive Changes - CEO” (Type ID¼101), and “Executive
Changes - CFO” (Type ID ¼102).
Some news articles mention multiple board changes. Excluding all of these observations will yield a small sample.
Therefore, for these articles, I keep them if they mention no more than two outside director changes, or no more than one
insider change. I keep track of Resigning Ex-Ante, Departing Ex-Post, and staying directors and test if the subsequent ap-
pointments (departures) of Resigning Ex-Ante and Departing Ex-Post directors trigger a more negative (positive) market
reaction. After excluding observations that do not have sufficient CRSP data to calculate the Day-0 abnormal return (based
on an estimation window of [ 240, 11]), my sample contains 765 director appointments and 779 director departures.
Table 8 reports the results. Columns 1 and 2 report the results based on the director appointment and departure an-
nouncements respectively. The Resigning Ex-Ante indicator is insignificant in Column 1, indicating that whether a director
was a Resigning Ex-Ante director from a negative-event firm in the past does not affect the shareholder reaction when the
director joins a board. On the other hand, this indicator is significantly positive in Column 2, suggesting that the shareholder
reaction tends to be more positive when a Resigning Ex-Ante director, rather than a staying director, leaves the board. Lastly, I
combine both director appointments and departures while replacing the abnormal returns in the appointments sample
with their opposite values. Consistent with the results in Column 2, the Resigning Ex-Ante indicator in Column 3 is sig-
nificantly positive. Therefore, the results in this section suggest that shareholders generally do not appreciate the presence
of directors who have previously resigned from other firms prior to negative events.
The Departing Ex-Post indicator is also significantly positive in Columns 2–3, suggesting that Departing Ex-Post directors
also trigger a positive shareholder reaction when they leave other boards relative to staying directors. However, this effect
appears to be weaker than the effect observed for Resigning Ex-Ante directors above. This weaker effect could be due to the
same reason explained in the previous sub-section. Some Departing Ex-Post directors leave a negative-event firm in the
second or the third year after the event. Because the samples in Columns 1–3 contain all director appointment and de-
parture announcements after the event, some of these announcements at other firms may occur before the Departing Ex-
Post directors actually leave the negative-event firms. In these cases, the market may not be able to differentiate between
the Departing Ex-Post and staying directors.
In light of this, I repeat the regressions in Columns 1–3 within the samples that exclude the announcements related to
Departing Ex-Post directors' appointments or departures that predate these directors' departures from the negative-event
firms. As shown in Column 4, the Departing Ex-Post directors now trigger a significantly negative shareholder reaction when
they join other boards. In Column 5, the significance level of Departing Ex-Post has also increased from the 10% level (as in
Column 2) to the 1% level. Lastly, in the sample that contains both appointment and departure announcements, the va-
luation impact triggered by Departing Ex-Post directors is once again more significant compared to Resigning Ex-Ante di-
rectors, which is consistent with the finding in this paper that Departing Ex-Post directors suffer the strongest penalties
among all three types of directors.
6. Alternative explanations
The baseline analysis finds that both directors who leave before and after a negative event occurs suffer stronger di-
rectorship declines than directors who stay through the event. There can be several alternative explanations to this finding,
which I consider in this section. First, the Resigning Ex-Ante director sample could include directors who are fired from the
board rather than leaving voluntarily. Including these directors can favor finding a significant board seat reduction for
Resigning Ex-Ante directors. In a subsample test, I exclude directors from firms where there are also inside directors leaving
in the pre-event year, or where three or more independent directors leave the firm in the pre-event year. The coefficient of
Resigning Ex-Ante remains significantly negative in this subsample test.
Directors may also leave due to other reasons such as poor health or busyness. To exclude these directors from the Resigning
Ex-Ante sample, I only include directors who acquired additional board seats within the three years after the event, and find
similar results also in this subsample test. In addition, I conduct a firm level matching analysis (based on firm size and industry)
and examine whether directors who resign from firms that do not have negative performance shocks in the same period also
experience a decline in board seats relative to those why stay at these firms. The results suggest that leaving a non-troubled firm
during the same period does not have any impact on a director's subsequent number of directorships, confirming that the
penalties for resigning only exist among negative-event firms. For brevity, these results are not reported in the paper.
7. Conclusion
The effectiveness of the director labor market has been a widely examined topic. Many studies document that directors
associated with negative events suffer reputational damage and receive penalties from the labor market. While there is
ample evidence that the labor market penalizes directors who are present when the event is revealed, it is unclear whether
directors can avoid these penalties by pre-emptively resigning from the firm.
Table 8
Market reaction to director appointments/departures. This table examines the market reaction when Resigning Ex-Ante, Departing Ex-Post, and staying
directors subsequently join or leave a board after the negative event occurs. The dependent variable for all columns is the Day-0 abnormal returns
estimated using a market model approach with an estimation window of [ 240, 11]. Appointments and Departures refer to the samples of director
appointment announcements and director departure announcements respectively. Pooled refers to the sample containing both appointment and departure
announcements, while multiplying the values of the dependent variable in the Appointments sample by (-1). The samples in Columns 1–3 include all
available announcements for Resigning Ex-Ante, Departing Ex-Post and staying directors, while the samples in Columns 4–6 exclude announcements of a
Departing Ex-Post director's subsequent appointment or departure where the director has not yet left the negative-event firm. CEO Duality is an indicator
variable that takes the value of one if the CEO and the chairman in the firm are the same person and zero otherwise. Classified Board is an indicator variable
indicating whether the directors of the board are divided into more than one class. Busy Board is an indicator variable that takes the value of one if more
than half of the independent directors of the firm hold three or more directorships. Institutional Ownership is the total percentage ownership by all
institutional shareholders. All regressions include both industry (2-digit SIC) fixed effects and year fixed effects. Standard errors are clustered at both the
industry level and year level. Beneath each coefficient estimate is its t-value. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels
respectively.
Full Sample Excluding Pre-Departure Announcements
Appointments Departures Pooled Appointments Departures Pooled

(1) (2) (3) (4) (5) (6)
Resigning Ex-Ante 0.002 0.011*** 0.006** 0.002 0.011*** 0.006**

( 0.474) (2.631) (1.975) ( 0.521) (2.942) (2.086)
Departing Ex-Post 0.003 0.002* 0.004*** 0.006** 0.007*** 0.007***
( 1.209) (1.749) (2.620) ( 2.254) (2.762) (4.732)
Age 0.000 0.000 0.000 0.000 0.000 0.000
( 0.310) ( 1.177) ( 0.811) (0.171) ( 1.193) ( 0.602)
Female 0.007** 0.000 0.004 0.008** 0.002 0.005
( 2.188) (0.052) (1.132) ( 2.322) (0.451) (1.554)
Directorships 0.001* 0.000 0.000 0.001 0.000 0.000
( 1.796) (0.349) (0.862) ( 1.354) (0.665) (0.889)
Independent 0.008 0.016 0.008 0.004 0.010 0.005
(0.742) ( 0.804) ( 0.994) (0.345) ( 0.532) ( 0.719)
Ln(Total Assets) 0.000 0.000 0.000 0.000** 0.000 0.000
(0.628) (0.560) (0.071) (2.529) (1.007) ( 0.209)
Return on Assets 0.008 0.010** 0.005* 0.006 0.007 0.003
( 1.260) (2.178) (1.707) ( 0.879) (1.470) (0.854)
Event-Year Stock Return 0.002 0.000 0.001 0.002 0.000 0.001
(1.011) (0.017) ( 0.704) (1.185) (0.104) ( 0.635)
Board Size 0.000 0.000 0.000 0.001 0.000 0.000
(0.237) ( 0.264) ( 0.522) (0.747) (0.170) ( 0.576)
Board Independence 0.017 0.021** 0.001 0.020 0.022** 0.000
( 1.338) ( 2.595) ( 0.174) ( 1.482) ( 2.229) ( 0.043)
CEO Duality 0.001 0.003 0.003 0.002 0.003 0.003
( 0.537) (1.031) (1.339) ( 0.657) (1.215) (1.244)
Classified Board 0.004** 0.005 0.003* 0.004* 0.005 0.003
(2.028) ( 1.371) ( 1.943) (1.886) ( 1.267) ( 1.607)
Busy Board 0.001 0.001 0.001 0.001 0.001 0.001
( 0.208) (0.510) (0.620) ( 0.149) ( 0.471) (0.374)
Institutional Ownership 0.003 0.001 0.001 0.003 0.001 0.001
(0.747) (0.238) ( 0.431) (0.834) (0.185) ( 0.395)
Resigning ¼Departing t-test 0.36 1.68 0.34 1.08 0.53 0.01
Resigning ¼Departing p-value 0.552 0.196 0.561 0.300 0.469 0.912
Observations 765 779 1544 744 756 1500
Adjusted R-squared 0.053 0.040 0.025 0.065 0.045 0.032
I show that the labor market does penalize the pre-emptive resignations. Compared to directors who have stayed at the firm
during the events, those who resign ex-ante suffer an additional 10.8% board seat reduction. This result is both significantly and
economically significant. Further, the penalties for the directors who leave in the post-event period are stronger than those who
resign ex-ante, consistent with these directors presumably being held most accountable for the events. These results are robust to
several alternative explanations.
The penalties also exist in a number of other dimensions. Directors that leave face a higher likelihood of losing committee
chairmanships at their other boards compared to directors that stay through the events. They also tend to subsequently hold less
prestigious directorships. In addition, when the public initially learns about the negative event, other firms where the Resigning Ex-
Ante director is still a board member exhibit a significantly negative shareholder reaction to these news announcements, especially
when the Resigning Ex-Ante director plays an important role on these other boards. Relatedly, shareholders in these connected firms
tend to react positively when these directors subsequently leave these boards. Overall, the results in this paper suggest that the
director labor market appears effective in providing ex-post accountability even if a director tries to abandon a firm ahead of
difficult times.
Appendix A. Collecting the number of board seats
The main barrier to collecting the number of directorships in Year 3 is that all of the Resigning Ex-Ante and Departing Ex-
Post directors are no longer present in the negative-event firms. It is therefore problematic to collect the information on
board seats only from these firms. To deal with this issue, I also consider the information from the other board seats these
directors hold in the sample. Doing so, however, raises another concern. Because the paper focuses on the change in the
number of directorships from Year 0 to Year 3, it is vital that there are no systematic differences in the ways directorships are
counted in Year 0 and Year 3. Therefore, I take the following three steps to collect the board seats information.
For each director, I first attempt to identify a firm where this director remains a board member from Year 0 to Year 3.
Information should be most reliable when it is collected for the same firm and covered by the same database over time.
For these directors, I take the number of board seats as reported by these firms both in Year 0 and in Year 3.
For the remaining directors, I then gather all of their directorship observations (still within the same director database
that covers the initial negative-event firms) in Year 0 and Year 3, and use the number of board seats reported in these
directorship observations. Occasionally, the board seat information is not consistent across firms for the same director. In
this case, I take the average value of the number of board seats reported by these firms.
After the above two steps, a large fraction of directors still have missing board seats data. These are directors who simply cannot
be found in either Year 0 or Year 3. To avoid losing all of these observations just because they have missing values for board seats,
I attempt to identify these directors within the other two director databases used in the paper.8 Directors whose numbers of
board seats in Year 0 and Year 3 are both available within one of the other two databases are also included in the analysis.
After these three steps, the average decrease in sample size due to missing values in the number of board seats is
approximately 30%.
References
Adams, R.B., Ferreira, D., 2007. A theory of friendly boards. J. Finance 62, 217–250.
Agrawal, A., Chen, M.A., 2011. Boardroom brawls: an empirical analysis of disputes involving directors. Available at SSRN 1362143.
Asthana, S., Balsam, S., 2007. Determinants of outside director turnover. Working paper.
Bebchuk, L.A., Cohen, A., 2005. The costs of entrenched boards. J. Financ. Econ. 78, 409–433.
Bereskin, F.L., Smith, C.W., 2014. Mechanisms of board turnover: evidence from backdating. J. Appl. Corp. Finance 26, 65–78.
Brickley, J.A., Linck, J.S., Coles, J.L., 1999. What happens to CEOs after they retire? New evidence on career concerns, horizon problems, and CEO incentives.
J. Financ. Econ. 52, 341–377.
Brochet, F., Srinivasan, S., 2014. Accountability of independent directors: evidence from firms subject to securities litigation. J. Financ. Econ. 111, 430–449.
Brown, W., Maloney, M., 1999. Exit, voice, and the role of corporate directors: evidence from acquisition performance. Available at SSRN 160308.
Chen, X., Harford, J., Li, K., 2007. Monitoring: which institutions matter? J. Financ. Econ. 86, 279–305.
Coles, J.L., Daniel, N.D., Naveen, L., 2008. Boards: does one size fit all? J. Financ. Econ. 87, 329–356.
Desai, H., Hogan, C.E., Wilkins, M.S., 2006. The reputational penalty for aggressive accounting: earnings restatements and management turnover. Account.
Rev. 81, 83–112.
Duchin, R., Matsusaka, J.G., Ozbas, O., 2010. When are outside directors effective? J. Financ. Econ. 96, 195–214.
Fahlenbrach, R., Low, A., Stulz, R.M., 2017. Do independent director departures predict future bad events? Rev. Financ. Stud.. (Forthcoming).
Faleye, O., 2007. Classified boards, firm value, and managerial entrenchment. J. Financ. Econ. 83, 501–529.
Fich, E.M., Shivdasani, A., 2006. Are busy boards effective monitors? J. Financ. 61, 689–724.
Fich, E.M., Shivdasani, A., 2007. Financial fraud, director reputation, and shareholder wealth. J. Financ. Econ. 86, 306–336.
Field, L., Lowry, M., Mkrtchyan, A., 2013. Are busy boards detrimental? J. Financ. Econ. 109, 63–82.
Fos, V., Tsoutsoura, M., 2014. Shareholder democracy in play: career consequences of proxy contests. J. Financ. Econ. 114, 316–340.
Gilson, S.C., 1990. Bankruptcy, boards, banks, and blockholders: evidence on changes in corporate ownership and control when firms default. J. Financ.
Econ. 27, 355–387.
Gompers, P.A., Metrick, A., 2001. Institutional investors and equity prices. Q. J. Econ. 116, 229–259.
Goyal, V.K., Park, C.W., 2002. Board leadership structure and CEO turnover. J. Corp. Financ. 8, 49–66.
Harford, J., 2003. Takeover bids and target directors' incentives: the impact of a bid on directors' wealth and board seats. J. Financ. Econ. 69, 51–83.
Hennes, K.M., Leone, A.J., Miller, B.P., 2008. The importance of distinguishing errors from irregularities in restatement research: the case of restatements
and CEO/CFO turnover. Account. Rev. 83, 1487–1519.
Kaplan, S.N., Reishus, D., 1990. Outside directorships and corporate performance. J. Financ. Econ. 27, 389–410.
Karpoff, J., Koester, A., Lee, D.S., Martin, G., Proxies and databases in financial misconduct research, Account. Rev. 2017, (Forthcoming).
Karpoff, J.M., Lee, D.S., Martin, G.S., 2008a. The cost to firms of cooking the books. J. Financ. Quant. Anal., 43; 581–611.
Karpoff, J.M., Lee, D.S., Martin, G.S., 2008b. The consequences to managers for financial misrepresentation. J. Financ. Econ. 88, 193–215.
Masulis, R.W., Mobbs, S., 2014. Independent director incentives: where do talented directors spend their limited time and energy? J. Financ. Econ. 111,
406–429.
Masulis, R.W., Mobbs, S., 2016. Supply and demand for independent director services: major board decisions and corporate outcomes. Available at SSRN
2222783.
8
The director merging across databases is based on director name and year of birth. I manually adjust director names in all of the three databases to
clean all the titles, prefixes, and postfixes, and ensure that the given names, surnames, and preferred names (if any) are all formatted consistently.
Masulis, R.W., Wang, C., Xie, F., 2012. Globalizing the boardroom—The effects of foreign directors on corporate governance and firm performance. J. Account.
Econ. 53, 527–554.
Nini, G., Smith, D.C., Sufi, A., 2012. Creditor control rights, corporate governance, and firm value. Rev. Financ. Stud. 25, 1713–1761.
Petersen, M.A., 2008. Estimating standard errors in finance panel data sets: comparing approaches. Rev. Financ. Stud. 22, 435–480.
Roberts, M.R., Sufi, A., 2009. Control rights and capital structure: an empirical investigation. J. Finance 64, 1657–1695.
Srinivasan, S., 2005. Consequences of financial reporting failure for outside directors: evidence from accounting restatements and audit committee
members. J. Account. Res. 43, 291–334.
Vafeas, N., 1999. Board meeting frequency and firm performance. J. Financ. Econ. 53, 113–142.
Yermack, D., 2004. Remuneration, retention, and reputation incentives for outside directors. J. Finance 59, 2281–2308.

Dou 2017 Leaving Before Bad Times Does The Labor Market Penalize

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Dou 2017 Leaving Before Bad Times Does The Labor Market Penalize

Hochgeladen von

Copyright:

Verfügbare Formate

Journal of Accounting and Economics 63 (2017) 161–178

Contents lists available at ScienceDirect

Journal of Accounting and Economics

Leaving before bad times: Does the labor market penalize

2. Data and sample

Combined Sample Lawsuits Earning Restatements

Dividend Reductions Covenant Violations

Year Events Resigning Departing Staying Events Resigning Departing Staying

Panel A: Firm Characteristics

Mean Std Dev P(25) P(75)

Total Assets (in millions) 13.236 41.723 0.986 7.851

Panel B: Director Characteristics

Age 59.625 8.499 54 65 60.044 60.377 59.174 0.333 0.87 1.203***

Directorship Change ¼ α þ β1 Resigning Ex-Anteþ β2 Departing Ex-Postþf(controls) þ ε

Resigning Ex-Ante 0.279*** 1.822*

4. The sources of penalties

4.1. Losing existing board seats

4.2. Acquiring new board seats

(1) (2) (3) (4) (5) (6)

Resigning Ex-Ante 1.025*** 0.854* 0.069*** 0.162*

(1) (2) (3) (4) (5) (6)

5. Other dimensions of penalties

5.1. Committee memberships/chairmanships

Resigning Ex-Ante 2.641*** 1.826*** 0.403 0.638***

5.2. Directorship prestige

Membership Chairmanship Size

Resigning Ex-Ante 0.015 0.094*** 6.531***

5.3. Market reaction to negative events announcements

Matched Other Connected Firms

Firms Full Membership o ¼1 Membership 41

Resigning Director Connection 0.014** 0.001 0.003 0.011***

5.4. Market reaction to future appointments and departures

Full Sample Excluding Pre-Departure Announcements

Appointments Departures Pooled Appointments Departures Pooled

Resigning Ex-Ante 0.002 0.011*** 0.006** 0.002 0.011*** 0.006**

Appendix A. Collecting the number of board seats

Das könnte Ihnen auch gefallen

Resigning Ex-Ante 2.641* 1.826* 0.403 0.638***

Resigning Ex-Ante 0.015 0.094* 6.531*

Resigning Director Connection 0.014 0.001 0.003 0.011*

Resigning Ex-Ante 0.002 0.011* 0.006 0.002 0.011* 0.006