Does Industry-Specific Expertise Improve Board Functioning? Evidence from Forced Bank CEO Turnovers

Page created by Suzanne Cortez
 
CONTINUE READING
Does Industry-Specific Expertise Improve Board Functioning?
                      Evidence from Forced Bank CEO Turnovers

                                         Zhongdong Chen1

                                              June, 2012

Abstract:

This study investigates whether independent directors’ expertise in the industry in which the
firm operates improves board functioning. In this study, board functioning is measured
through the lens of firm performance following a CEO turnover. Using a sample of 157
bank CEO turnovers at 202 banks from 1995 to 2009, I find that market responds more
positively to forced CEO turnover decisions when they are made by a more independent
board. I document that improvements in bank performance following a forced bank CEO
turnover are significantly positively related to independent financial expertise on the board.
In contrast, board independence improves post-turnover bank performance for the overall
sample. It does not particularly benefit forced turnovers of a bank CEO. This is likely
because that a properly functioning board is particularly important when a forced CEO
turnover becomes necessary, and industry-specific expertise greatly improves boards’
ability to monitor and advise management.

JEL classification: G21, G34, G38
Keywords: board advising, board structure, banking industry,
            CEO turnover, financial expertise

1
 University of Tennessee, Department of Finance, College of Business Administration, 427 Stokely Management
Center. Email: zchen12@utk.edu

I thank Harold Black, David Cicero, Phillip Daves, Ramon DeGennaro, Mike Ehrhardt, Larry Fauver, Deborah
Harrell, Jason Howell, Ulrike Malmendier, Usha Mittoo, Andy Puckett, Alvaro Taboada, John Wachowicz, James
Wansley, Tracie Woidtke, and seminar participants at the University of Tennessee and the 2011 FMA PhD
Consortium. I am especially grateful to my advisor, Alvaro Taboada, as well as to Andy Puckett, Ulrike
Malmendier, and Tracie Woidtke, for valuable discussion and suggestions.
1. Introduction

    The recent policy debate over corporate governance reform has sparked new

interests in board structure (Adams, Hermalin, and Weisbach, 2010). When exploring

how board structure impacts board effectiveness in monitoring and advising

management, previous studies largely focus on board independence (e.g., Weisbach,

1988; Denis and Denis, 1995), board size (e.g., Yermack, 1996; Coles, Daniel, and

Naveen, 2008), and directors’ individual knowledge that is not related to the industry

in which the firm operates (e.g., Agrawal and Chadha, 2005; Güner, Malmendier, and

Tate, 2008). Less attention has been paid to independent industry-specific expertise

on boards in previous literature.     Fama (1980) and Fama and Jensen (1983)

acknowledge the importance of insider knowledge for board functioning and suggest

that it may be optimal to have insiders on boards (also see, e.g., Klein, 1998).

However, as noted in Williamson (1984), agency cost can be higher when

management has information advantages over the board of directors due to insider

knowledge (also see, e.g., Beasley, 1996). Independent industry-specific expertise on

boards, which potentially combines inside directors’ knowledge and independent

directors’ vigilance (e.g., Mace, 1986; Brickley, Coles, and Terry, 1994; Rosenstein

and Wyatt, 1997), may mitigate this issue.       In this study, I examine whether

independent directors’ expertise in the industry in which the firm operates improves

board functioning by investigating bank performance following a forced CEO

turnover.

                                          1
I focus on the banking industry for three reasons.                     Most importantly, the

complexity of banks and the opacity of bank operations (e.g., Levine, 2004) make the

banking industry a unique platform to test the significance of directors’ industry-

specific expertise. Arguably, more so than in other industries, an independent bank

director needs a solid financial background to be able to monitor and advise the

management. Some extreme examples of the primary occupations of independent

bank directors in this study are dentists, professional volunteers, chiropractors, and

judges. It is reasonable to assume that without a financial background, independent

bank directors are less able to fulfill their monitoring and advising duty. Kroll,

Walters, and Wright (2008) suggest that independent expertise may be the key for a

board to function properly. Brickley and Zimmerman (2010) also emphasize directors’

occupational backgrounds over board independence. Second, financial expertise is

more readily observable than other types of industry expertise. The definition of a

financial expert in this study follows Güner et al. (2008) and Minton, Taillard, and

Williamson (2011). A director is classified as a financial expert if the director is (was)

employed by a financial organization (e.g., venture capital firm; consumer lending

company; mutual fund; hedge fund) or a banking regulator (e.g., the FRB; the FDIC).1

Last but not least, following the recent financial crisis, government regulators and

institutional investors are requiring banks to restructure their boards with more

1
    This classification is more conservative than that used in Güner et al. (2008) and Minton et al. (2011).
Finance-related officers in non-financial firms and business professors (e.g., Finance and Accounting)
are also classified as financial experts in these two studies. I exclude these directors because they may
not necessarily have a strong understanding of bank operations. Furthermore, the two cited studies do
not consider directors with working experience with a banking regulator.
                                                      2
independent directors that have financial expertise.2 However, our knowledge on the

relationship between independent directors’ financial expertise and bank performance

is quite limited.

        Forced bank CEO turnovers make a unique set to test the impact of industry-

specific expertise on board functioning.             Because, according to Mace (1986),

“directors serve as a source of advice and counsel… and act in crisis situations” when

a forced CEO turnover is necessary (also see Adams et al., 2010).                      Previous

governance studies suggest that board of directors plays a significant role in corporate

strategies. Therefore, a properly functioning board may be particularly important at

the point where a bank needs to oust its incumbent CEO. In this study, board

functioning is measured through the lens of post-turnover bank performance.

Consistent with previous literature, this study finds that in general, a forced bank CEO

turnover is preceded by a significant decline in bank performance. In the case of poor

performance, a troubled bank may be in greater needs of better monitoring and

advices from professionals. Conventional wisdom and previous literature suggest that

independent directors that have financial expertise may be better able to satisfy these

needs (e.g., Anderson and Reeb, 2003a, 2003b; Kor and Misanyi, 2008; Schmidt,

2008).      Therefore, independent financial expertise should positively affect board

functioning and, ultimately, bank performance following a forced CEO turnover.

        Independent financial expertise may improve the functioning of a bank board

through several channels. Independent financial experts may be better able to locate a

2
    “Help Wanted: Bank Boards Seeking Competent Directors”, Stephen Gandel, TIME, May 20, 2009.
                                                 3
superior successor CEO, which, according to the Improvement Management

Hypothesis (e.g., Weisbach, 1988; Fredrickson, Hambrick, and Baumrin, 1988; Denis

and Denis, 1995), leads to improved ex-post bank performance. As noted in Walsh

and Seward (1990) and others, boards constantly face the difficult task of attributing

poor performance to exogenous negative performance shocks or CEO quality (also

see, e.g., Jenter and Kanaan, 2010). Directors with financial expertise may be better

able to select a superior successor CEO because their expertise may facilitate the

identification of exogenous performance shocks when assessing the quality of

incumbent CEO and candidates. Furthermore, as mentioned earlier, with a financial

background, independent bank directors may be better able to monitor and advise

management. This may be especially important for banks that need to replace their

incumbent CEOs.

    In this study, I investigate the relationship between independent financial

expertise and the functioning of a bank board with a sample of 157 bank CEO

turnovers, including voluntary and forced turnovers. I find a higher rate of forced

bank CEO turnovers during periods of financial turmoil, consistent with Jenter and

Kanaan (2010). In an event study examining cumulative abnormal stock returns

(CARs) around the announcements of a bank CEO turnover, I find that market

response to forced CEO turnovers is significantly positive while the response to

voluntary CEO turnovers is statistically and economically insignificant. I further find

that the positive market response is driven by forced turnover decisions made by a

more independent board. The evidence indicates that the market recognizes the
                                          4
significant role of board independence in a CEO firing decision and believes that a

more independent board is likely to help improve the status quo.          In contrast,

independent financial expertise does not have a similar impact on the market response

toward forced bank CEO turnovers.

    In an examination of post-turnover bank performance, I find that forced CEO

turnover and outsider succession significantly improve bank performance. After the

interactive effects between forced CEO turnover and board structure are controlled in

a Heckman two-step model, forced CEO turnover becomes negatively correlated,

although not significantly, with improvements in post-turnover bank performance. I

document that improvements in bank performance following a forced CEO turnover

are driven by banks with more independent financial experts on their boards. In

contrast with the market belief, board independence does not particularly benefit

forced CEO turnovers in terms of ex-post performance. Moreover, for the overall

sample, board independence is positively correlated with improvements in

performance and independent financial expertise tends to be negatively correlated

with such improvements, consistent with the findings in Huson et al. (2004) and

Minton et al. (2011), respectively. The evidence suggests that independent financial

expertise improves the functioning of a bank board in the period following a forced

CEO turnover. It is likely because that a bank board with more independent financial

expertise is better able to monitor and advise the management in that critical period.

These findings are not driven by mean reverting or risk-taking behavior. The results

                                          5
are robust to different specifications of board independence and alternative measures

of risk-taking and governance quality.

     The findings add to the growing literature suggesting the importance of directors’

expertise in board functioning. As noted in Carpenter and Westphal (2001) and others,

to advance the knowledge on the effectiveness of board functioning, governance

research may need to go beyond a focus on board independence to directors’ relevant

expertise.   The evidence of a positive correlation between independent financial

expertise and ex-post bank performance following a forced CEO turnover is new.

This study also adds to the literature on the relationship between forced CEO turnover

and ex-post firm performance.       Previous literature tends to attribute improved

performance following a forced CEO turnover to “improved management”. The new

evidence suggests that the performance consequence of a forced CEO turnover may

partly depend on boards’ effectiveness in monitoring and advising management in the

post-turnover period.    Finally, this study has implications for policy issues as

government regulators and institutional investors are requiring banks to restructure

their boards with more independent directors that have a financial background.

     The reminder of the paper is organized as follows. In section 2, I discuss related

literature and provide some background information on regulatory requirements on

bank board structure. In section 3, I describe the data. Sections 4, 5, and 6 are

dedicated to reporting the empirical results. I present robustness tests in section 7,

and concluding remarks in section 8.

                                          6
2. Related literature and background information

2.1. Directors’ expertise

     Rosenstein and Wyatt (1990) examine stock market reaction to the

announcement of the appointment of an independent director.         They find that a

director’s occupation does not impact abnormal stock returns around announcements.

They argue that independent directors of any occupation are equally valuable. The

evidence in Shivdasani and Yermack (1999) is supportive of this argument. However,

recent studies suggest that independent directors with industry-specific expertise may

be comparatively more valuable in some circumstances. Anderson and Reeb (2003a,

2003b) find better performance at US firms with founding-family ownership. This is

likely because that, as suggested in Anderson and Reeb (2004), families may place

independent directors with industry-specific expertise on the board for performance

related reasons.    This indicates that founding families recognize the value of

independent industry-specific expertise on a board. Kor and Misangyi (2008) find

evidence that independent industry-specific expertise can, to some extent, offset the

lack of industry experience by top management. However, Minton et al. (2011) find

different results. Their evidence from the banking industry shows that independent

financial expertise on a bank board is positively correlated with poor stock

performance and risk-taking during the recent financial crisis, and the correlation is

                                          7
stronger among large commercial banks. They argue that it is because financial

experts may encourage bank management to exploit the “too big to fail” effect.

    Previous studies have shown the importance of independent directors’ expertise

in corporate governance. Eng and Mak (2003) argue that independent expertise on a

board and corporate disclosure are substitutes, suggesting the important role of

independent expertise in corporate control. Gul and Leung (2004) find consistent

evidence. They also find that CEO/chairman duality limits corporate disclosure and

independent expertise on a board weakens such negative association.

2.2. Determinants of post-turnover performance

    The turnover literature suggests that, in general, forced CEO turnover leads to

improved ex-post performance because of improved management (e.g., Denis and

Denis, 1995; Weisbach, 1988; Fredrickson et al., 1988).        Huson et al. (2004)

specifically point out that the magnitude of improvement in performance is positively

correlated with board independence. Successor origin also impacts post-turnover

performance. For instance, Khurana and Nohria (2000) find that following a forced

CEO turnover, an insider succession leads to statistically and economically

insignificant performance changes while an outsider succession leads to better

performance. In line with this notion, Adams and Mansi (2009) find evidence that

successor origin is correlated with the change in firm value. Huson et al. (2004)

document a positive correlation between outsider succession and expected change in

                                          8
subsequent performance. As a result, Borokhovich et al. (1996) find that abnormal

stock returns around forced CEO turnover announcements are significantly positive

for outsider successions but significantly negative for insider successions.

     Post-turnover performance may be related to board size.             Hermalin and

Weisbach (2003) argue that board size is negatively related to firm performance. This

is likely because of inefficiency and the free-rider problem associated with large

boards (e.g., Jensen, 1993). However, Coles, Daniel, and Naveen (2008) suggest that

very small and very large boards can both be optimal, depending on firm complexity.

In line with this argument, Adams and Mehran (2008) find that Tobin’s Q is

increasing in bank board size, given that banks are comparatively large and complex.

Moreover, Pathan (2009) argues that small boards and more powerful boards tend to

restrain bank risk-taking.

     A staggered board deters takeover and thus hinders market discipline (e.g.,

Bebchuk, Coates, and Subramanian, 2002), resulting in less managerial effort.

Consistent with this, Bebchuk and Cohen (2005) find a lower Tobin’s Q at firms with

a staggered board. Faleye (2007) further suggests that a staggered board decreases the

probability of forced CEO turnover at underperforming firms.             Mahoney and

Mahoney (1993) and Faleye (2007) both find a negative market reaction to the

announcement of the adoption of a classified board. Therefore, a staggered board

may be negatively associated with post-turnover improvements in firm performance.

                                           9
An independent board is believed to be associated with better governance quality.

Beasley (1996) suggests that firms committing financial statement fraud tend to have

a significantly lower percentage of independent directors. Core, Holthausen, and

Larcker (1999) find that firms with a weak board are more likely to overcompensate

their CEOs. Although in general board independence does not significantly impact

long term firm performance (e.g., Hermalin and Weisbach, 1988; Klein, 1998; Bhagat

and Black, 2002), it is positively correlated with firm performance following a CEO

turnover (e.g., Huson et al., 2004).

     According to Yermack (2004), with all incentives accounted for, a $1,000

increase in firm value increases the wealth of a Fortune 500 firm director by 11 cents.

A one standard deviation change in firm performance implies a change of about

$285,000 in the director’s wealth. Thus a well-motivated independent director has a

good reason to monitor management. In line with this reasoning, studies have shown

that the percentage of equity-based compensation and ownership by independent

directors are positively correlated with a higher probability of forced CEO turnover at

underperforming firms (e.g., Perry, 1999; Brickley, Lease, and Smith, 1988).

However, a higher percentage of equity-based compensation as well as a greater

ownership by independent bank directors may encourage excess risk-taking,

maximizing the value of under-priced deposit insurance.         The impact of these

incentives on bank performance following a forced CEO turnover is not clear.

                                          10
Woidtke (2002) argues that institutional investors are likely to monitor

management, although the value effect of institutional monitoring depends on the

objectives of the institutions. Therefore, institutional ownership may be positively

correlated with improvements in performance following a forced bank CEO turnover.

2.3. Regulatory requirements on bank board structure

        Recent efforts by government regulators to increase bank board independence

can be traced back to the Financial Institutions Reform Recovery and Enforcement

Act of 1989 (FIRREA) and the Federal Deposit Insurance Corporation Improvement

Act of 1991 (FDICIA). The Sarbanes-Oxley Act of 2002 (SOX) also emphasizes

board independence but it applies only to public financial institutions. The Federal

Deposit Insurance Corporation (FDIC) issued the “Pocket Guide for Directors” (the

guide) to non-public banks following the passage of SOX.         The guide, which

emphasizes “the need for a strong and independent board of directors”, encourages

practices quite similar to, or even identical to those in SOX. The Federal Reserve

Board (FRB), the Office of Thrift Supervision (OTS) 3 , and the Office of the

Comptroller of the Currency (OCC) also issued similar guidance to non-public banks

which they supervised.          Banks listed on the NYSE, AMEX, and NASDAQ are

required by the exchanges to have a board that is comprised of a majority of

independent directors.

3
    On July 21, 2011, the OTS became part of the OCC.
                                                 11
At the same time, the definition of “independence” has become stricter in the last

decade. The Amendments to the Housing Act on July 30, 2008 specifies that “an

individual is not eligible to be an independent bank director if the individual serves as

an officer, employee, or director of any member of the bank, or of any recipient of

loans from the bank”. The three stock exchanges also adopted stricter definition of

“independent director” since 2002.

        Previously, regulations largely focused on board independence. Although FDIC

implemented new requirements on independent directors’ financial expertise

following the banking crisis in the 1990s, they are applied only to bank audit

committees. Until recently, little attention was paid to the occupational background

of independent directors at banks. Following the recent crisis, government regulators

and institutional investors are requiring banks to restructure their boards with more

independent directors that have financial expertise.

3. Data and summary statistics

        In this study, I explore the relation between independent financial expertise and

post-turnover bank performance using a sample of CEO turnovers at US banks in the

1995-2009 period.4 To indentify changes in CEO position, I start with all firms in

Standard & Poor’s Execucomp database with an SIC code between 6000 and 6300

from January 1st, 1995 to December 31st, 2009. Of these 295 firms, I exclude non-

4
    I start with the year 1995 because board information prior to 1994 is not available in the Edgar filing
database.
                                                     12
lending firms with SIC codes 6172 (Finance Lessors), 6099 (Functions Rel to Dep

Bkg, Nec), 6282 (Investment Advice), 6163 (Loan Brokers), 6200 (Security &

Commodity Brokers), and 6211 (Security Brokers & Dealers). I also exclude the

three federal credit agencies (Freddie Mac, Fannie Mae, and Sallie Mae) because they

have important differences compared with other financial institutions due to their

unique role as quasi-government entities.5 I manually check the remaining firms and

exclude those firms (e.g., American Express, Mellon Financial Corp, State Street

Corporation, Finova Group, and Financial Federal Corporation) whose primary

service is not making loans (e.g., wire transferring and investment advice). This

yields a final raw sample of 202 firms, including 150 commercial banks, one finance

services company, five mortgage bankers, 33 federal-chartered savings institutions

and 13 non-federal-chartered savings institutions. Following Fahlenbrach and Stulz

(2011), I define the lending firms from this sample as banks. The sample has a mean

age of 23 years (median age of 24 years), and is potentially biased toward large banks

because Execucomp tends to cover larger firms, as shown by the market capitalization

in panel A of Table 1.6 The silver lining from this basis is that those banks represent

the bulk of the market value of all banks in the US and their CEO firing decisions

have a significant impact on the industry and its “aggregate wealth” (Huson et al.,

2004).

5
    For example, Sallie Mae functioned as a government-sponsored enterprise until its privatization
process was finished in 2004 and even after that point it was still widely regarded as quasi-government
backed. However, the main results do not change after I include CEO turnovers at these three
institutions.
6
    Market capitalization is averaged across the 1995-2009 period for each bank.
                                                    13
[Insert Table 1]

    After I identify CEO changes at the 202 sample banks through the Execucomp

database, I search for banks’ announcements and related news on their top

management changes through LexisNexis and Factiva. For the purpose of this study,

I exclude CEO turnovers that are directly related to M&A activities. I also exclude

turnovers of interim/acting CEOs that serve less than one year. Of the 160 bank CEO

turnovers generated by the above selection process, I further exclude the three CEO

turnovers that are preceded by M&A activities in the seven-quarter pre-turnover

period. This yields a cleaner sample allowing me to focus on performance changes

that are caused only by turnovers.        Together with information provided by

Execucomp, I am able to obtain turnover announcement date and effective date, CEO

age and tenure at turnover for all 157 turnovers. As shown in panel A of Table 1, the

median outgoing bank CEO is 62 years old and has served as a CEO for 9 years.

Quarterly data on bank total assets, operating income, and capital assets ratio from

1993 to 2011 are either obtained from Bank Regulatory database, COMPUSTAT or

manually collected from Form 10-Q, Form 10-K, or Form 8-K.

    Information on board structure in the turnover year is obtained from the bank’s

proxy statement (Form 14A) immediately before the turnover announcement through

SEC’s Edgar filing database. Board information for ten turnovers is set to missing

either because the proxy statements are not found in the Edgar database (e.g., Hanmi

Financial Corp, 1999), or because the proxy statements do not provide enough

                                         14
information (e.g., the primary occupation of the directors is not reported by Bancwest

Corp in 2004). Data are available for 147 turnovers, as reported in panel A of Table 1.

Following the literature (e.g., Huson et al., 2004), I classify a director as an insider if

the director is, or was employed by the bank, its subsidiary, or an affiliated

organization. Other directors are classified as outside directors. I go through each

outside director’s detailed information reported in the proxy statement and classify an

outside director as “grey” if the director (1) is employed by one of the following firms:

consulting/advisory firm, investment bank, law firm, insurance company; or (2) has

significant business relationship with the bank; or (3) is related to an officer of the

bank (e.g., Huson et al., 2001). Other outside directors are classified as independent

directors. In panel A of Table 1, a median bank board has 13 directors, of which

66.67% (81.67%) are independent (outside) directors. This percentage of independent

directors (the number of independent directors/board size) is very close to those

reported in Adams and Mehran (2003) and Booth, Cornett, and Tehranian (2002). A

median independent bank director has served on the board for 11 years.

     As previously mentioned, a director is classified as a financial expert if the

director is (was) employed by a financial organization (e.g., venture capital firm;

consumer lending company; mutual fund; hedge fund) or a banking regulator (e.g.,

the FRB; the FDIC). I find that other director occupations include, but are not limited

to, attorneys, dentists, realtors, professional volunteer, professors, surgeons, farmers,

dairy farmers, chiropractors, judges, and executive officers of industrial firms. While

the percentage of independent director is high (median=66.67%), the median (mean)
                                            15
percentage of independent financial experts (the number of independent financial

experts on a board/board size) is only 9.09% (11.93%). The correlation coefficient

between the percentages of independent directors and independent financial experts is

0.2115 (p-value=0.01).

        Director compensation information in the turnover year for the 147 bank CEO

turnovers is obtained either from the Execucomp database or from bank filings. I

calculate the percentage of director equity-based compensation (EBC) using the year-

end stock price in the prior year for stock grants and I use the modified Black-Scholes

option valuation model for option grants (e.g., Guay, 1999; Core and Guay, 2002).

As reported in panel A of Table 1, the median percentage of equity-based

compensation for directors is 52.34%. CEO/chairman duality and ownership structure

are found in the proxy statement immediately before a turnover announcement.7 I

exclude ownership by institutions affiliated with the bank, such as a subsidiary bank

or employee stock ownership plan trust, because these institutions are not likely to

monitor bank management (e.g., Adams and Mehran, 2003). The median ownerships

by CEO and his or her immediate family, independent directors as a group, and

institutional investors are 0.82%, 0.81%, 5.61%, respectively.

        Following the turnover literature (e.g., Huson et al., 2001), I classify a turnover

as forced if: (1) the announcement mentions strategy or opinion differences between

the CEO and the board; or (2) related news uses words such as “ousted, fired, or

forced out”; or (3) the CEO is under the age of 60 and he (or she) is leaving for

7
    Out of the 147 outgoing CEOs, 104 (70.75%) are also board chairmen.
                                                  16
reasons other than death, health issues, or reposition; or (4) the retiring CEO is under

the age of 60 and does not announce his (or her) intention to retire at least 6 months

prior to the effective turnover date. All other turnovers are classified as voluntary.

Panel B of Table 1 reports the frequency of bank CEO turnovers over the 1995-2009

period. Of all the 157 bank CEO turnovers identified in this study, 65 are classified as

forced turnovers and 92 are classified as voluntary turnovers. The average annual

turnover rate (the annual number of turnovers/the annual number of banks) is 7.04%,

similar to that reported in Hubbard and Palia (1995) (6%, a sample of 116 banks) over

the three years before the introduction of interstate bank regulation.8 The average

annual forced turnover rate (41.40%) is higher than that reported in Taylor (2010)

(21.3%, financial firms, 1970-2007) and that in Huson et al. (2004) (16%, a sample of

firms across industries over 1971-1994). This is because the sample period in this

study covers two crises (the 1998 LTCM/Asian financial crisis and the recent 2007-

2009 crisis) and one recession (2000-2002) 9, in which the frequency of forced CEO

turnover spiked. The rate of forced bank CEO turnover (the number of forced bank

CEO turnovers/then number of bank CEO turnovers) in panel B of Table 1 is

consistent with Jenter and Kanaan (2010) where they find that CEOs are more likely

to be fired during periods of industry turmoil.

        A succession is labeled an insider succession if the successor holds, or held a

position in the bank, its subsidiaries or affiliates for more than a year (e.g., Huson et

8
    Hubbard and Palia (1995) find a cumulative turnover rate of 18.1% in the three years before interstate
banking legislation. I divide this percentage by three to find the approximate annual turnover rates.
9
    CRSP value-weighted NYSE/AMEX/NASDAQ Index dropped 37.55% during 2000-2002.
                                                    17
al., 2001). All other successions are labeled an outsider succession.10 Out of the 157

successions, 34 are classified as outsider successions and 123 are classified as insider

successions. Panel B of Table 1 reports that 33.85% (13.04%) of the outgoing CEOs

are replaced by an outside successor following a forced (voluntary) turnover, lower

(higher) than the 49.6% (9.9%) reported in Parrino (1997) (a sample of firms across

industries over 1970-1989). This indicates that outsider succession is more likely

following a forced turnover than following a voluntary turnover. It should be noted

that when an insider officer is promoted to the CEO position he (or she) usually

leaves his (or her) current position. I find that out of the 123 inside successors in the

sample, 50 are currently COOs and 12 are currently CFOs. I also find that some

banks replace their CEOs and other top executive officer(s) at the same time.11 This

means that other top management changes (e.g., CFO; COO) are likely to take place

around CEO turnovers, which may also influence a bank’s stock price and post-

turnover performance (e.g., Mian, 2001).                 However, it is nearly impossible to

disentangle these effects.

4. Market reaction to bank CEO turnover announcements

10
     I do not differentiate between outside successors from other banks and those from other industries
(see Parrino, 1997). Of the 34 outside successors, only three come immediately from another industry.
However, they all have a banking background.
11
     For example, Fremont General Corp. announced on Monday, November 12, 2007 that its CEO,
Louis J. Rampino, and COO, Wayne R. Bailey had resigned, which is classified as a forced CEO
turnover in this study. http://articles.latimes.com/2007/nov/13/business/fi-fremont13

                                                   18
There have been many empirical studies examining the impact of CEO turnover

announcements on stock price. In general, these studies find an insignificant reaction

across their overall samples but significantly positive market reaction for forced

turnovers (see Furtado and Karan (1990) for a summary of market reaction studies).

In this section, I examine the impact of turnover type (forced vs. voluntary) and board

structure (i.e., board independence; independent financial experts) on stock price

around bank CEO turnover announcements, using a sample of 157 bank CEO

turnovers as described in the prior section. I do not test the difference between insider

succession and outsider succession because many banks, especially those that oust

their CEOs, have not selected a successor at the time of announcement, which

significantly reduces the sample size.

     Panel A of Table 2 presents the results of an event study centered on the turnover

announcement day (day 0). I report cumulative abnormal returns (CARs) for various

windows including [0], [-1, 0], [0, +1] and [-1, +1]. The reported CARs are returns

adjusted by a four factor model that includes the market risk premium (the spread

between CRSP value-weighted market return and risk-free rate), SMB (the return

spread between portfolios of small and big capitalization stocks), HML (the return

spread between portfolios of high and low book-to-market stocks), and a momentum

factor. The market model parameters are estimated with daily returns over the 150-

trading-day period 30 days before a turnover announcement (see, e.g., Ellis, 2010).

To be included in this event study, a bank’s stock must trade around the

announcement day. I go through related news and bank filings around turnover
                                           19
announcements and exclude observations that may be biased by significant

contemporary events. Appendix A cites several examples of excluded observations. I

report the results for the remaining 138 CEO turnover announcements in panel A of

Table 2.

                                   [Insert Table 2]

     Panel A of Table 2 shows that, for the overall sample, average market reaction is

positive but not significant on the announcement day. Market reaction to forced

turnovers is positive over four event windows. CARs on the day of forced turnover

announcement are statistically (at the 5% level) and economically significant (1.74%).

CARs for voluntary turnovers are statistically and economically insignificant. The

spread between market reactions to forced turnovers and voluntary turnovers on the

announcement day is statistically (at the 1% level) and economically (2.07%)

significant.

     To examine whether board structure impacts CARs around CEO turnover

announcements, I estimate OLS model (1) and report the results in panel B of Table 2.

Pairwise correlations between independent variables are reported in Table 3.

                                   [Insert Table 3]

       =         +                            +       +                            (1)

     where CARs are abnormal stock returns around bank CEO turnover

announcements. Forced is a forced turnover indicator that is equal to 1 if a turnover

                                         20
is classified as forced and 0 otherwise. Governance Variables include board size, a

staggered board indicator that is equal to 1 if the directors are classified and 0

otherwise, independent director tenure, the percentage of director equity-based

compensation (% Director EBC), ownership by independent directors as a group,

ownership by institutional investors, ownership by outgoing CEO and his or her

immediate family, outgoing CEO tenure, a CEO/chairman duality indicator that is

equal to 1 if there is duality and 0 otherwise, the percentage of independent directors (%

ID), and the percentage of independent financial experts (% IFE). I control for bank

size, proxied by the natural logarithm of market capitalization (Lsize). To see whether

board independence and independent financial expertise are particularly important for

the decisions of a forced CEO turnover, I include two interaction terms (ITS): one is

between the forced turnover indicator and board independence (% ID*Forced), the

other is between the forced turnover indicator and the percentage of independent

financial experts on a board (% IFE*Forced).

     Panel B of Table 2 suggests that without controlling for the interactive effects

between board structure and the forced turnover indicator, CARs for forced turnovers

are significantly higher (at the 1% level over [0], [-1, 0], and [0, +1]) than those for

voluntary turnovers.   In contrast with previous studies, board independence (the

percentage of independent directors) is not correlated with market reaction. The

percentage of independent financial experts is also not correlated with CARs.

Institutional ownership is significantly negatively correlated with CARs over the four

event windows.
                                          21
When the two interaction terms are included in the regressions, panel B of Table

2 shows that the interaction term, % ID*Forced, is positively correlated with CARs,

significant at the 1% level in all estimates. In contrast, the interaction term, %

IFE*Forced, is not correlated with CARs.12 More importantly, the forced turnover

indicator becomes significantly (at least at the 5% level in three estimates and at the

10% level on day 0) negatively correlated with CARs once the two interaction terms

are included. This is consistent with the view that forced CEO turnovers reveal

negative insider information. The F-test reported in panel B of Table 2 strongly

rejects the null hypothesis that the coefficients of Forced, %ID*Forced, and %

IFE*Forced are jointly zero.

        The evidence indicates that forced CEO turnovers tend to receive positive market

 reaction and it is driven by turnover decisions made by a more independent board. It

 is likely because that the market understands the pivotal role of board independence

 in corporate governance. One explanation is that a more independent board is better

 able to filter out exogenous negative performance shocks from CEO quality and is

 therefore less likely to scapegoat. Another explanation is that a more independent

 board may be better able to locate a superior successor, leading to truly improved

 management.          The evidence in Borokhovich, Brunarski, Donahue, and Harman

 (2006) is consistent with this explanation as they find a positive correlation between

12
     One concern is that this may be driven by the significant correlation between % ID and % IFE. I
drop %ID*Forced in unreported tests and find similar results.
                                                  22
board independence and market reaction to exogenous CEO turnovers (i.e.,

 executive deaths), which involves the assessment of the quality of successor CEOs.

5. Performance changes around bank CEO turnovers

     It has been well documented in the turnover literature that firm performance

improves following a forced CEO turnover (e.g., Denis and Denis, 1995; Huson et al.,

2004). However, this result may not hold for banks due to the unique nature of their

business and the heavy regulations they face, along with their different governance

mechanisms (e.g., Brickley and James, 1987; Adams and Mehran, 2002; Adams and

Mehran, 2003; Becher, Campbell, and Frye, 2005). Direct evidence from the banking

industry is limited to date. Hence it is necessary to examine the changes in bank

performance around bank CEO turnovers within the framework of this study.

     Specifically, I examine bank performance over the seven-quarter pre- and post-

turnover periods.    Four measures of performance are employed: unadjusted and

industry-adjusted return on assets, abnormal stock performance, and Tobin’s Q. I do

not choose a period of three years as in previous studies (e.g., Huson et al., 2004), and

admittedly, the selection of a seven-quarter period is arbitrage. This is because 60

(38.22%) successor CEOs in the sample survive less than three years.        The sample

can be further reduced when considering the tenure of the outgoing CEOs. In contrast,

only 24 (15.29%) successor CEOs survive less than seven quarters. Arguably, a

period of seven quarters is long enough to test the impact of bank CEO turnover on

                                           23
subsequent bank performance.                  In contrast with previous studies where annual

accounting data are used, I use quarterly data, which according to Huson et al. (2001)

“would provide a more precise measure of the information available to internal

monitors when the succession is announced”.                       It should be noted that quarterly

accounting data have disadvantages. 13 First of all, quarterly data are unaudited.

However, quarterly data (e.g., call report) are also used by banking regulators to

perform off-site examination of banks (e.g., Cole and Gunther, 1998; Hirtle and

Lopez, 1999), suggesting their inherent value.                      Besides, examinations by bank

regulators add credibility to quarterly data. Second, quarterly data may be seasonal

and changes over the seven-quarter pre- and post-turnover periods may partially

capture this seasonality. I utilize industry-adjusted data in the empirical tests to

alleviate this issue.

        To obtain a clean sample, I extensively check bank announcements and news

reports and exclude banks that announce M&A activity in seven quarters following a

CEO turnover announcement. Banks with M&A activities in the pre-turnover period

are already excluded from the sample as described in previous section. From the

remaining turnovers, I exclude CEO turnovers where the outgoing CEO or the

successor CEO has held the CEO position for less than seven quarters, either because

the CEO leaves his or her position or because the bank does not survive in the seven-

quarter post-turnover period. This selection process results in a sample of 119 bank

13
     In robustness tests, I find similar results with annual accounting data.
                                                       24
CEO turnovers, including 46 forced turnovers (13 of these CEOs are succeeded by an

outsider) and 73 voluntary turnovers (11 of these CEOs are succeeded by an outsider).

5.1. Changes in return on assets around CEO turnovers

        In this section, I compare changes in return on assets (ROA = net income/the

book value of total assets, see, e.g., Anderson and Reeb, 2003a; Adams and Mehran,

2008) across turnover types (forced vs. voluntary) and successor origins (outsider vs.

insider) in the seven-fiscal-quarter pre- and post-turnover periods.

                                         [Insert Figs. 1 and 2]

        Figs. 1 and 2 plot sample banks’ average industry-adjusted ROA over the period

from seven quarters before to seven quarters after CEO turnovers.14 Industry median

ROA is calculated from a pool of banks covered by COMPUSTAT from 1993 to 2011

(around 600 banks are included in each year). The upper graph in Fig. 1 compares

industry-adjusted ROA between outsider and insider successions. It shows that banks

experiencing worse performance tend to select an outside successor and their

performance improves afterward.                Insider succession does not display a similar

pattern. The lower graph in Fig. 1 compares industry-adjusted ROA between forced

and voluntary turnovers. Banks that constantly experience worse performance tend to

fire their CEOs and their performance improves thereafter. In Fig. 2, the sample

banks are divided into two groups: voluntary turnovers and forced turnovers. The

14
     I find a similar pattern with unadjusted ROA.
                                                     25
upper graph in Fig. 2 indicates that within the voluntary turnover group, banks

experiencing worse and more volatile performance tend to select an outside successor.

The lower graph in Fig. 2 suggests that among banks that fire their CEOs, those

experiencing a sharp decrease in industry-adjusted ROA tend to choose a successor

from the outside talent pool, and their performance improves in the post-turnover

period. Insider successions following a forced turnover do not show a similar pattern.

                                   [Insert Table 4]

     In Table 4, I report average changes in unadjusted and industry-adjusted ROA in

the pre-turnover (Quarter -1 minus Quarter -7) and post-turnover (Quarter 7 minus

Quarter 1) periods. Panel A indicates that unadjusted ROA declines (at the 10% level)

prior to a forced bank CEO turnover and recovers significantly (at the 1% level) in the

post-turnover period.    Unadjusted ROA does not drop significantly prior to a

voluntary bank CEO turnover, but deteriorates in the post-turnover period (at the 1%

level). When comparing between forced and voluntary turnovers (column 1 minus

column 2), banks experiencing worse performance (-2.22%, marginally significant at

the 10% level) tend to oust their CEOs, leading to a more significant improvement

(3.82%, significant at the 1% level). Panel A of Table 4 also indicates that the

selection of an outside successor is likely motivated by poor performance in the pre-

turnover period, compared with insider successions (column 3 minus column 4,

significant at the 1% level). Outside successors tend to outperform inside successors

                                          26
in the post-turnover period, although not significantly. Industry-adjusted ROA in

panel A of Table 4 tells a similar story.

     When grouping turnovers by turnover type and successor origin, panel B of

Table 4 suggests that banks with a significant decline (at the 5% level) in unadjusted

ROA tend to fire their CEOs and employ an outside successor (column 7). In contrast,

forced turnovers without a significant decrease in unadjusted ROA tend to be

succeeded by an insider (column 5). Consistent with the literature, forced bank CEO

turnover leads to a significant (at the 5% level) improvement in unadjusted ROA

regardless of successor origin (columns 5 and 7). Furthermore, outsider successions

following a forced turnover lead to a more significant improvement than those

following a voluntary turnover (column 7 minus column 8, marginally significant at

the 1% level).    As a matter of fact, outsider successions following a voluntary

turnover are characterized by a decline in unadjusted ROA (column 8).            When

examining industry-adjusted ROA in panel B of Table 4, I find similar results.

5.2. Abnormal stock performance around CEO turnovers

     In this section I examine banks’ long-run abnormal stock performance around

CEO turnovers with a sample of 119 bank CEO turnovers as described in previous

section.   Table 5 reports average abnormal stock returns over the following six

window periods around the announcement quarter (quarter 0): [-Q7, -Q1], [-Q4, -Q1],

[-Q1], [+Q1], [+Q1, +Q4], [+Q1, +Q7]. Stock returns are adjusted by a four factor

                                            27
model that includes the market risk premium (the spread between CRSP value-

weighted market return and risk-free rate), SMB (the return spread between portfolios

of small and big capitalization stocks), HML (the return spread between portfolios of

high and low book-to-market stocks), and a momentum factor. The market model

parameters are estimated with data over the 24-month period seven quarters before the

turnover announcement quarter.

                                   [Insert Table 5]

     Panel A of Table 5 suggests that banks showing a sharp decrease in pre-turnover

abnormal stock performance, especially during the [-Q4, -Q1] and [-Q1] (significant

at the 5% level and 1% level, respectively) windows, tend to oust their CEOs. Their

abnormal stock performance improves in the post-turnover period as abnormal returns

are not significantly negative any more (column 1). In contrast, following a voluntary

CEO turnover, abnormal stock returns are significantly negative (at least at the 5%

level) over all post-turnover windows (column 2). Column 3 shows that insider

successions are followed by a significant decline in abnormal stock returns. Column

4 indicates that banks with significantly negative (at the 1% level over [-Q4, -Q1] and

[-Q1]) abnormal stock returns prior to a turnover tend to recruit an outside successor.

When comparing outsider succession with insider succession, I find that the former

significantly outperforms the latter over the [+Q1, +Q4] (at the 5% level) and [+Q1,

+Q7] (at the 10% level) windows.

                                          28
As reported in panel B of Table 5, I find that insider successions following a

voluntary CEO turnover lead to significantly negative abnormal stock returns in the

post-turnover period (column 6). Abnormal stock performance continues to plunge

following a forced turnover if the successor comes from inside of the bank, especially

over the [+Q1, +Q4] window (significant at the 5% level, column 5). Among banks

that fire their CEOs, those that select an outside successor have experienced a

significant decline (at least at the 1% level) in abnormal stock performance in the pre-

turnover windows (column 7), and they significantly outperform banks that hire from

inside the firm (column 5 minus column 7).           Ultimately, outsider successions

following a forced turnover significantly outperform those following a voluntary

turnover over the [+Q1, +Q4] and [+Q1, +Q7] windows.

5.3. Changes in Tobin’s Q around CEO turnovers

     In this section, I examine changes in firm value (measured by Tobin’s Q) in the

seven-fiscal-quarter pre- and post-turnover periods with a sample of 119 turnovers as

described in previous section. Following Adams and Mehran (2008) and others,

Tobin’s Q is defined as the ratio of a bank’s market value to the book value of total

assets. A bank’s market value is calculated as the market value of equity plus the

book value of total assets minus the book value of equity.

                                 [Insert Figs. 3 and 4]

                                          29
Figs. 3 and 4 plot sample banks’ average Tobin’s Q over the 15-quarter period

centered on the turnover quarter. When comparing insider and outsider successions,

the upper graph in Fig. 3 shows that outsider succession is preceded by a decrease in

Tobin’s Q and followed by a significant increase in Tobin’s Q. In contrast, insider

succession leads to a decline in firm value. The lower graph in Fig. 3 indicates that

voluntary turnover is followed by a decrease in Tobin’s Q. Banks that oust their

CEOs have a lower and declining firm value in the pre-turnover period. Their firm

value stops decreasing in the post-turnover period, but remains low relative to banks

with a voluntary turnover.

        Within the group of voluntary turnovers (upper graph in Fig. 4), Tobin’s Q

declines following an insider succession and increases following an outsider

succession.        Outsider succession has a higher Tobin’s Q in both pre- and post-

turnover periods. Among firms that complete forced turnovers (lower graph in Fig. 4),

those that have a lower and declining Tobin’s Q tend to choose an outside successor

and their Tobin’s Q increases in the post-turnover period.                            Insider successions

following a forced turnover do not lead to a similar improvement.

                                               [Insert Table 6]

        I report average percentage change in Tobin’s Q in the seven-quarter pre- and

post-turnover periods in Table 6.15 Panel A of Table 6 suggests that forced turnovers

and outsider successions are both preceded by a significant decrease in Tobin’s Q

15
     I find similar results in empirical tests using the level change in Tobin’s Q.
                                                       30
(columns 1 and 4). Panel B of Table 6 further indicates that with banks experiencing

a significant decline in Tobin’s Q tend to oust their CEOs and select an outside

successor.

6. Determinants of post-turnover bank performance

        Previous results have showed that in general, forced bank CEO turnovers are

preceded by a decline in performance and that turnover type and successor origin do

impact post-turnover bank performance. Forced turnovers significantly increase bank

performance in the seven-quarter post-turnover period, especially forced turnovers

followed by the succession of an outside CEO. In this section, I systematically

examine the determinants of post-turnover bank performance.                     I am particularly

interested in the relationship between board structure and bank performance following

a forced CEO turnover. Following Huson et al. (2004), I use a two-step model

developed in Heckman (1979) 16 to deal with the selectivity issue because some

successors in the sample do not survive until the end of the seven-quarter post-

turnover period. To obtain consistent regression results, in the first step I use probit

model (2) to estimate the probability of survival and the inverse Mill’s ratio (IML):

               =            +       +                                +                         (2)

        where Survivor is a binary variable that takes a value of 1 if a successor holds the

CEO position for seven quarters and 0 otherwise. Out is an outsider succession

16
     See Huson et al. (2004) for a discussion of this two-step Heckman model.
                                                    31
indicator that is 1 if a succession is labeled an outsider succession and 0 otherwise.

Other variables are defined as in model (1). I report results for probit model (2) in

column 1 of Table 7.

                                    [Insert Table 7]

     In the second step, I estimate OLS model (3) with the IML added as an

independent variable. The IML is supposed to capture the omitted variables in OLS

regressions where data are censored (e.g., Heckman, 1979). Regression results for

OLS model (3) are presented in Table 7.

    =          +       +                         +       +       +      +            (3)

     where PC is performance change over the seven-quarter post-turnover period,

including changes in unadjusted and industry-adjusted ROA, abnormal stock

performance, and the percentage change in Tobin’s Q. I control for the change in

bank risk-taking in the post-turnover period. Following Laeven and Levine (2009), I

use the natural logarithm of the z-score to proxy bank risk-taking. Since the z-score is

defined as “the inverse of the probability of insolvency” (Laeven and Levine, 2009), it

is a proper measure of bank risk-taking. A positive change in z-score (LZC) implies a

decrease in bank risk-taking. Following the literature, the z-score is calculated as a

bank’s average return on assets (ROA) plus capital assets ratio (CAR) divided by the

standard deviation of return on assets (σ(ROA)), or mathematically, z =

(ROA+CAR)/σ(ROA). Other variables are defined as in probit model (2).

                                          32
There are studies suggesting that CEO/chairman duality jeopardizes board

independence by increasing CEO power over the board (e.g., Palvia, 2011; Goyal and

Park, 2002; Adams, Almerda, and Ferreira, 2005). These studies take the view that

duality entrenches CEOs and decreases firm performance (e.g., Dalton and Rechner,

1991). I do not include a CEO/chairman duality indicator for successor CEOs in OLS

model (3) for the following reasons. First, there also are studies suggesting that

duality is merely a “natural result of succession process” and thus has no systematic

impact on firm performance (e.g., Brickley, Jeffrey, and Gregg, 1997; Adams et al.,

2005). Second, Zajac and Westphal (1996) argue that the power of a board is decided

by its tenure relative to that of the CEO. It is reasonable to assume that duality alone

will not significantly increase the power of a successor CEO over the board since the

former is new to the CEO position. Ultimately, it is difficult to test the impact of

duality on ex-post performance because of the temporary separation of CEO and

chairman following a CEO turnover. Typically, an outgoing CEO will remain on the

board as the chairman for several months (or even several quarters) before the

successor CEO takes over the chair position. 17 However, the main results do not

change after I add a CEO/chairman duality indicator for outgoing CEOs in the

Heckman two-step model.

17
     For instance, AmSouth Bancorporation announced on December 21, 1995 that current CEO John W.
Woods would be succeeded in January 1996 by C. Dowd Ritter, president and COO of the bank. Mr.
Woods would continue as the chairman of the board, an additional role to be assumed by Mr. Ritter in
August 1996. http://www.highbeam.com/doc/1G1-17931956.html
                                                33
Results in Table 7 (columns 2, 4, 6, and 8) indicate that without controlling for

the interactive effects between forced CEO turnover and board structure, forced CEO

turnover significantly (at the 1% level) increases unadjusted and industry-adjusted

ROA. The magnitude is economically significant. On average, forced turnover

increases unadjusted (industry-adjusted) ROA by 3.471% (3.18%).             Outsider

succession leads to a 3.68% (3.27%) improvement in ex-post bank performance

(significant at the 5% level) when performance is measured by unadjusted (industry-

adjusted) ROA. It also improves abnormal stock performance significantly. The

change in bank risk-taking over the seven-quarter post-turnover period is negatively

correlated with performance changes, significant at least at the 10% level in three

estimates. The stagger board indicator and independent directors’ ownership are

negatively related to improvements in bank performance.          The percentage of

independent directors is significantly related to improvements in ROA following a

bank CEO turnover, consistent with Huson et al. (2004). Bank size is negatively

related to ex-post improvements in ROA. Consistent with Minton et al. (2011),

independent financial experts are negatively correlated with abnormal stock

performance and the change in Tobin’s Q.

    So far I have confirmed that forced turnover of a bank CEO improves ex-post

bank performance.     Next I investigate whether board structure impacts board

functioning, which ultimately impacts ex-post improvements in bank performance.

As discussed previously, a properly functioning board can be particularly important

when a forced CEO turnover becomes necessary. Independent financial expertise is
                                        34
You can also read