Summary and PowerPoint

Make a summary and PowerPoint (A,B, and C) (Please use easy words and easy sentences)

A.prepare a 3 pagessummary of each paper (including the Libby box summary)

Focus on summarizing the most salient points of the article.

. After that make sure the summaries address the following questions (I will send them).

B.and make PowerPoint slides for each study.

C. and A separate Word file listing the key terms mentioned in the article, along with important, simplified explanations for each.”

JOURNAL OF MANAGEMENT ACCOUNTING RESEARCH
Vol. 26, No. 2
2014
pp. 123–149
American Accounting Association
DOI: 10.2308/jmar-50550
Institutional Investor Preferences for
Corporate Governance Mechanisms
Brian J. Bushee
University of Pennsylvania
Mary Ellen Carter
Boston College
Joseph Gerakos
The University of Chicago
ABSTRACT: We examine institutional investors’ preferences for corporate governance
mechanisms. We find little evidence of an association between total institutional
ownership and governance mechanisms. However, using revealed preferences, we
identify a small group of ‘‘governance-sensitive’’ institutions that exhibit persistent
associations between their ownership levels and firms’ governance mechanisms. We
also find that firms with a high level of ownership by institutions sensitive to shareholder
rights have significant future improvements in shareholder rights, consistent with
shareholder activism. Further, we find that factors describing the characteristics of
institutions’ portfolios are correlated with governance preferences. Large institutions,
those holding a large number of portfolio stocks, and those with preferences for growth
firms are more likely to be sensitive to corporate governance mechanisms, suggesting
those mechanisms may be a means for decreasing monitoring costs and may be more
essential for firms with a high level of growth opportunities. Finally, our results suggest
that common proxies for governance sensitivity by investors (e.g., legal type,
blockholding) do not cleanly measure governance preferences.
Keywords: corporate governance; institutional investors; board of directors.
JEL Classifications: G11; G20; G34.
We appreciate helpful comments and suggestions from Tracie Woidtke and workshop participants at the Drexel
University Conference on Corporate Governance Research, Harvard Business School, Michigan State University, and
The University of North Carolina/Duke Fall Accounting Conference. We are grateful for the funding of this research
provided by the Wharton, Carroll, and Booth Schools. Joseph Gerakos gratefully acknowledges financial support from
the KPMG and Nathan Foundations.
A prior version of this paper circulated under the title ‘‘Do Institutional Investors Care about Corporate Governance?’’
Published Online: July 2013
123
Bushee, Carter, and Gerakos
124
INTRODUCTION
I
nstitutional investors are commonly assumed to be a key component of corporate
governance—monitoring and disciplining managers through explicit actions or ‘‘voting with
their feet.’’ Prior research finds that a small number of institutional investors take an active role
in the governance of their portfolio firms by waging public and private campaigns, sponsoring
shareholder proposals, and voting against management attempts to entrench (Gillan and Starks
2003; Barber 2006). But such actions are costly, have uncertain outcomes, and may require
collective action. For institutional investors that are sensitive to corporate governance, an alternative
approach is to simply invest in firms with existing, preferred governance mechanisms. In general,
little is known about the extent to which firms’ corporate governance mechanisms are an explicit
determinant of institutional investors’ portfolio weighting decisions.
We examine institutional investors’ revealed preferences for firm-level corporate governance
mechanisms. In contrast to prior research that examines the association between corporate
governance and institutional ownership on a firm level (e.g., Borokhovich, Brunarski, Harman, and
Parrino 2006; Chen, Harford, and Li 2007; Li, Ortiz-Molina, and Zhao 2008), we investigate this
association on an institution level by examining the extent to which institutional investors tilt their
portfolios toward firms with preferred governance mechanisms.
Institutional investors have a number of incentives that would lead them to prefer firms with
‘‘better’’ corporate governance mechanisms.1 First, institutional investors often hold large
portfolios, for which external monitoring costs are high. Bushee and Noe (2000) find that
institutions with a large number of portfolio stocks prefer higher-quality disclosure as a way to
offset monitoring costs. Thus, institutional investors could prefer firms with strong internal
monitoring mechanisms that serve as a substitute for the institutions’ own costly monitoring
activities. Second, there could be an association (perceived or actual) between corporate
governance mechanisms and superior firm performance that is not captured by other firm
fundamentals. For example, Gompers, Ishii, and Metrick (2003), Brown and Caylor (2006), and
Larcker, Richardson, and Tuna (2007) find that ‘‘better’’ governed firms exhibit higher firm value,
better operating performance, and a reduction in potentially wasteful corporate investment. Third,
the presence of stringent fiduciary responsibilities can lead some institutions to prefer firms with
‘‘better’’ governance mechanisms, because such mechanisms can reduce the possibility of negative
outcomes due to managerial fraud or negligence (Del Guercio 1996). Fourth, institutions holding
large positions in firms or following an index strategy will find it costly to rapidly liquidate their
positions during a governance failure, which could lead to a preference for ‘‘strong’’ governance
mechanisms. Fifth, institutions following investment styles that favor small or riskier firms could
seek ‘‘better’’ governance mechanisms as a way of reducing the risk of their undiversified sector bet.
Combined, these incentives could lead institutional investors, as a whole, to exhibit preferences for
corporate governance mechanisms in their investment decisions.
We investigate three questions. First, we consider to what extent corporate governance is an
explicit determinant of institutions’ investment decisions, providing insight into whether
1
One difficulty in any governance study is determining whether a governance mechanism is ‘‘better’’ or
‘‘weaker.’’ For example, there is disagreement in the literature about whether large boards provide ‘‘better’’
governance (e.g., Dalton, Daily, Johnson, and Ellstrand 1999) or ‘‘weaker’’ governance (e.g., Yermack 1996).
However, Table 1 shows that our measure of board characteristics has improved over time. Given the pressures
for governance improvements after Sarbanes-Oxley and Enron, this improvement suggests that our measure
captures what the market perceives to be ‘‘better’’ governance mechanisms. In addition, our tests do not rely on
the assumption that certain governance mechanisms are superior. Instead, we rely only on the assumption that
governance mechanisms are observed by institutional investors, who then choose whether to include these
mechanisms in their investment decisions and monitoring activities.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
125
governance is a complement to institutional holdings. Second, we consider to what extent
institutions appear to implement their preferred governance mechanisms in their portfolio firms, as
opposed to simply investing in firms with preferred mechanisms. We examine whether the
complementary relation between institutional holdings and governance reflects preferences for
board characteristics or shareholder rights. With no theory to guide us, this analysis provides new
stylized facts to the literature. Finally, we examine which types of institutions display preferences
for corporate governance mechanisms. Prior research treats institutions as homogeneous. In
contrast, we use revealed preference to determine the types of institutions that are more sensitive to
governance. In all our analyses, we test for governance sensitivity using a broad range of
governance mechanisms within the categories of board of director characteristics and shareholder
rights.
We first investigate the association between total institutional ownership and firms’ corporate
governance mechanisms. Despite a number of potential incentives that institutional investors have
to tilt their portfolios toward firms with ‘‘better’’ governance mechanisms, we find little evidence of
an association between total institutional investor ownership and corporate governance. There is
weak evidence that firms with ‘‘better’’ board characteristics have higher levels of total institutional
ownership, but we find no association between shareholder rights and total institutional ownership.
Given the lack of an overall relation, we identify institutions that exhibit strong revealed preferences
for governance mechanisms to provide stylized facts on the proportion, influence, and
characteristics of institutional investors that are sensitive to governance in their investment
decisions and monitoring activities.
Using data from 1995 to 1997, a period in which there were no major scandals or changes in
regulation that would have induced a stronger focus on governance, we find that approximately 10
percent of institutions are sensitive to director and shareholder governance mechanisms (i.e.,
governance characteristics significantly affect their portfolio weighting decisions). Using a holdout
sample from 1998 to 2004, we confirm the validity of our revealed preference classification to
ensure it is not an artifact of our statistical cutoff for governance sensitivity. Institutions classified as
governance-sensitive continue to exhibit significant preferences for governance mechanisms.
Institutions classified as governance-insensitive do not exhibit significant preferences for board
characteristics in the later period, but do exhibit significant preferences for ‘‘weaker’’ shareholder
rights, explaining the insignificant association for total institutional ownership. Thus, revealed
preferences identify a group of institutional investors with persistent preferences for governance
mechanisms.
Next, we address the question of whether these institutions simply invest in firms with
preferred governance mechanisms or actively implement preferred mechanisms in their portfolio
firms. We find strong evidence that changes in ownership by governance-sensitive institutions are
associated with prior levels of, and contemporaneous changes in, board characteristics, implying
that governance-sensitive institutions prefer to invest in firms with existing preferred governance
mechanisms. Nevertheless, we find evidence that firms with a high level of ownership by
institutions sensitive to shareholder rights exhibit significant future improvements in shareholder
rights, implying that these institutions engage in shareholder activism. This empirical evidence
complements survey results of 118 institutional investors in the U.S. and Netherlands stating that
governance is a consideration in portfolio weighting decisions, and these institutions are willing to
engage in activities that can improve the governance of their portfolio firms (McCahery, Sautner,
and Starks 2010).
Further, we investigate the characteristics of institutions that are governance-sensitive.
Consistent with their fiduciary responsibilities, bank trusts and pensions and endowments tend to
have the highest percentage of governance-sensitive institutions. But neither type has more than 25
Journal of Management Accounting Research
Volume 26, Number 2, 2014
126
Bushee, Carter, and Gerakos
percent of its institutions classified as governance-sensitive, indicating that the legal type
classifications do not fully proxy for general governance sensitivity.
Finally, we examine the association between governance sensitivity and a set of factors that
describe the characteristics of institutions’ portfolios. We find that large institutions and institutions
holding a large number of stocks in their portfolios are more likely to be sensitive to corporate
governance mechanisms, suggesting that institutions view governance mechanisms as a means to
decrease monitoring costs. In addition, we find that institutions with preferences for growth firms
tilt their portfolios toward firms with ‘‘better’’ board characteristics, implying that institutions view
board governance as more essential for firms with a high level of growth opportunities. In contrast,
institutions with long investment horizons and small-cap investment styles are more likely to tilt
their portfolios toward firms with ‘‘better’’ shareholder rights, suggesting that shareholder
governance allows these institutions to protect their large, stable investments. Interestingly,
blockholder ownership by institutional investors is not significantly related to governance
sensitivity, suggesting that block ownership serves as a substitute for governance mechanisms,
rather than a complement. Overall, our results suggest that common proxies for governance
sensitivity by investors (e.g., legal type, blockholding) do not fully capture important aspects of the
motivation for governance sensitivity and, as a result, likely misclassify institutions with respect to
their governance sensitivity.
Prior research has examined the role of institutional investors in the governance and
decision-making of their portfolio firms. For example, research has examined the influence of
institutional investors on public campaigns, such as shareholder proposals and voting (Brickley,
Lease, and Smith 1988; Smith 1996; Wahal 1996; Gillan and Starks 2000; Aggarwal, Saffi, and
Sturgess 2012), on private negotiations with management (Strickland, Wiles, and Zenner 1996;
Carleton, Nelson, and Weisbach 1998), on anti-takeover charter amendments (Borokhovich et al.
2006), on shareholder voting rights (Li et al. 2008), on major corporate decisions such as forced
CEO turnover (Parrino, Sias, and Starks 2003), on executive compensation (Hartzell and Starks
2003; Almazan, Hartzell, and Starks 2005; Dikolli, Kulp, and Sedatole 2009), and on mergers
(Chen et al. 2007). These studies provide mixed evidence as to whether institutional investors act as
effective monitors of management and/or whether their governance actions are profitable. Possible
explanations for the mixed results are that, in general, these studies focus either on specific
corporate events or on specific classes of institutions, such as public pension funds. In addition,
these studies may not fully account for the fact that institutions can ‘‘vote with their feet’’ if they
disagree with a firm’s governance and decision-making (Bhide 1993; Admati and Pfleiderer 2009;
Edmans 2009).2
Our study contributes to the literatures on institutional investors and corporate governance in
the following ways. First, we investigate the preferences of institutional investors for corporate
governance mechanisms, rather than their role in initiating or reacting to major governance actions.3
By examining investment behavior, we are able to use a large sample of firms and institutions to
investigate the market forces that influence firm-level governance mechanisms. Second, we
2
In a similar vein, Agrawal and Knoeber (1996) find little evidence of an association between total institutional
ownership and other possible control mechanisms (e.g., insider ownership, blockholders, outside directors, CEO
human capital, and leverage). Of this list, outside directors are the only factors also considered in this study.
3
Note that the international literature finds that foreign institutional investors prefer to invest in firms with
‘‘better’’ governance practices (e.g., Leuz, Lins, and Warnock 2009; Ferreira and Matos 2008). This literature
assumes that firm-level corporate governance mechanisms substitute for weak country-level legal protections of
minority shareholders. In contrast, our study examines the preferences of U.S. institutional investors for
domestic securities. The U.S. legal system provides one of the highest levels of protection for minority
shareholders. Therefore, the motivations for governance-sensitivity in our setting are more likely associated with
investor characteristics.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
127
document the governance sensitivity of a broad set of institutional investors, taking into
consideration their heterogeneity. Not all institutional investors have the same investment
objectives or philosophy, and some are constrained by fiduciary duties or influenced by political
concerns. Understanding the heterogeneous preferences of institutional investors is increasingly
important, given recent legislative attempts to increase shareholder oversight of boards of directors
and executive compensation. For example, Section 971 of the Dodd-Frank Act would have
increased shareholder access to proxy nominations of directors had the Securities and Exchange
Commission (SEC) legislation to implement it not been overturned by the courts (U.S. House of
Representatives 2010; Holtzer 2011). Additional amendments to other SEC rules not subject to that
litigation do require companies to provide shareholder proposals regarding proxy access in
company proxy materials (Shapiro 2011). Furthermore, similar legislation (‘‘Shareholder Bill of
Rights’’) introduced by Senator Schumer had been pending in the past (Davidoff 2009). This study
provides insight into the preferences of institutional investors for such changes. Finally, we specify
and validate a parsimonious method to classify the corporate governance sensitivity of institutional
investors. In doing so, we provide evidence of the types of institutions likely to be active in
corporate governance reforms, and develop a more refined method to classify institutions in the
study of investor activism.
DATA AND SAMPLE
Sample
Our sample consists of 15,892 firm-year observations between 1995 and 2004. The sample
period is constrained by the availability of data on board of director characteristics, obtained from
the Directors database of the Investor Responsibility Research Center (IRRC). This database
contains director information for approximately 1,800 companies (S&P 500, S&P MidCap, S&P
SmallCap) from proxy statements dated 1996 to 2005. We match proxy statements to their fiscal
year (i.e., 2001 proxy data for a December fiscal year-end firm applies to the 2000 fiscal year). As a
result, the majority of the data applies to fiscal years 1995 to 2004.
As in Gompers et al. (2003), we obtain the data on shareholder rights from the IRRC
Governance database. These data are available for the years 1990, 1993, 1995, 1998, 2000, 2002,
and 2004, representing the year in which proxy statements were surveyed to gather the data. To
form a complete panel of data for our tests, we use the 1995 survey for 1995–1996 fiscal years, the
1998 survey for 1997–1998 fiscal years, the 2000 survey for 1999–2000 fiscal years, the 2002
survey for 2001–2002 fiscal years, and the 2004 survey for 2003–2004. Note that this data structure
prevents us from examining annual changes in the relation between governance factors and
institutional investors.
We obtain institutional holdings from the Thomson Financial Spectrum database. These data
compile SEC Form 13-F filings of institutional holdings. Under Rule 13(f ), all institutional
investors managing more than $100 million in equity are required to file all equity holdings greater
than 10,000 shares or $200,000 in market value with the SEC on a quarterly basis. For each firmyear observation, we calculate institutional ownership for each quarter and then use the mean of the
four quarters in empirical tests. We obtain data for our control variables from the Compustat and
CRSP databases.
Proxies for Corporate Governance
Our proxies for corporate governance mechanisms are divided into two groups. The first group
consists of board characteristics: board size, percent of independent directors, whether the CEO is
the chairperson, presence of board interlocks, and board meeting attendance. These variables
Journal of Management Accounting Research
Volume 26, Number 2, 2014
128
Bushee, Carter, and Gerakos
capture the extent to which governance serves as an internal mechanism for monitoring managers
(Hermalin and Weisbach 2003). The second group consists of the shareholder rights (or antitakeover provisions) embedded in the corporate charter, identified by Gompers et al. (2003). Lower
rights discipline managers by exposing them to the external market for corporate control. To be
consistent with that paper, all governance variables are defined so that smaller values capture
‘‘better’’ governance.
Board Characteristics
Larger boards are considered ineffective because communication, coordination, and decisionmaking problems are greater (Yermack 1996). Our proxy for board size is the log of the number of
directors (LNDIR). The combination of the CEO and chairperson positions is considered ineffective
governance because it reduces the possibility that the board will objectively monitor management.
We code an indicator variable (CEO) as 1 if the positions are combined, and 0 otherwise.
Independent directors are considered more effective monitors of management because their careers
are not dependent on the goodwill of management (Rosenstein and Wyatt 1990; Byrd and Hickman
1992). To proxy for ineffective governance, we calculate the percentage of directors that are not
independent (PNID). Interlocked directors (directors who serve on each others’ boards) are
considered indicative of ‘‘weaker’’ governance, because such directors have reciprocating
relationships that create incentives to vote in ways that benefit their counterparts and, hence,
themselves (Hallock 1997). We code an indicator variable (DLOCK) as 1 if there are any interlocks
on the board of directors, and 0 otherwise. Finally, attendance at board meetings is considered an
indication of a director’s effort in monitoring management. We include an indicator variable for bad
attendance (DBAD) coded as 1 if any director misses 75 percent or more of board meetings, and 0
otherwise.4
We combine these five variables into an index of board of director characteristics to serve as a
parsimonious measure of board quality. Bivariate correlations and factor analyses strongly suggest
that these five characteristics are independently determined.
Thus, we create a formative index similar to that of Gompers et al. (2003).5 The index
(DINDX) is computed as the sum of the three indicator variables, CEO, DLOCK, and DBAD, and
indicators for whether the firm has a high level of LNDIR and PNID. To form these indicators, we
split the distribution of LNDIR and PNID into high and low groups using k-means cluster analysis.
This approach allows for uneven clusters and is better suited to find breakpoints in the distribution
than a median split, which often divides the distribution in the center of its mass, leaving
observations in each group that are very similar.6 Thus, DINDX ranges from 0 to 5, with 0 (5)
representing boards with the best (weakest) combination of governance mechanisms.
4
There are two other board characteristics that we considered, but do not use due to data limitations: the mean
number of other boards that directors serve on and the amount of ownership by officers and directors in the
company. Both variables are missing prior to 1997, and service on other boards is reported for fewer than half of
the observations after 1997. A second problem with ownership is that large values will be mechanically related
to the percentage of institutional ownership, and small values are often missing because director ownership does
not have to be reported if it is less than 1 percent. We do use director ownership as a control variable, as
discussed later.
5
Like the GINDX of Gompers et al. (2003), our DINDX does not reflect the relative impact of each component. It
is, however, transparent, and it captures independent components in a parsimonious manner.
6
This analysis starts with a low and high observation and classifies each subsequent observation into the high or
low group based on the lower Euclidian distance between the observation and the two cluster means. Cluster
means are recomputed after each new observation is classified and the procedure iterates until all observations
are clustered. We were unable to split LNDIR for 2004 and, therefore, used the cut point for 2003.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
129
Panel A of Table 1 presents descriptive statistics on DINDX and its component variables. Over
11 percent of firm-years have a DINDX score of 0, representing the best governance mechanisms,
and over 35 percent have a score of 1. Thus, over 45 percent of the sample has boards characterized
by ‘‘better’’ governance mechanisms. Only 3.4 percent of the firm-years have ‘‘weaker’’ governance
on at least four of the five dimensions. Except for LNDIR at the highest level of DINDX, the mean
values of each of the five components of the index increase monotonically in the DINDX score.
Panel B shows the time-series change in DINDX. The percentage of firms with the best board
characteristics increased dramatically over time, from 6 percent in 1995 to 19 percent in 2004. Most
of this movement stems from firms with a DINDX score of 3. Thus, firms with relatively ‘‘weaker’’
governance mechanisms have shown dramatic changes in governance mechanisms since 2001.
Shareholder Rights
We proxy for shareholder rights using the governance index (GINDX) constructed by Gompers
et al. (2003), which is the sum of 24 individual corporate charter components and state laws relating
primarily to takeover protections, voting rules, and liability limitations. Gompers et al. (2003)
divides GINDX into five major subcomponents. DELAY is the sum of four provisions designed to
slow down hostile bidders. VOTING is the sum of six provisions related to shareholder rights in
elections or charter amendment votes. PROTECT is the sum of six provisions that protect officers
and directors from firm-related liability and provide termination-related compensation. OTHER is
the sum of six firm-level provisions relating to greenmail, directors’ duties, fair price, pension
parachutes, poison pills, and silver parachutes. These generally represent mechanisms to make
takeovers more costly to potential bidders. STATE captures whether the firm is incorporated in
states with specific anti-takeover laws and, if so, whether the firm chooses to opt out of the law.
Panel A of Table 1 presents descriptive statistics on GINDX and its component variables. For
parsimony in presentation, we divide GINDX into six groups in the table (in the analyses, we use
the continuous measure). The distribution of GINDX is more symmetric than that of DINDX, with
most firms clustered in the middle and fewer firms at the tails. The mean component scores tend to
increase monotonically in the level of GINDX, indicating that this is also a formative index with
small intra-item correlations. Panel B shows that, unlike DINDX, there has been no secular trend
toward ‘‘better’’ GINDX scores. In fact, firms with low GINDX scores have moved toward the
middle range in the latter part of our sample.
Correlations
Panel C of Table 1 presents correlations among DINDX, GINDX, and all of their components.
Note that the correlation between DINDX and GINDX is only 0.119, suggesting that these two
forms of governance mechanisms operate independently. Among the components of DINDX, the
highest bivariate correlation is between the number of directors (LNDIR) and bad attendance
(DBAD), but is only 0.154. Among the components of GINDX, there are moderately high bivariate
correlations (;0.35) among DELAY, OTHER, and PROTECT, but no other correlation greater than
0.13. DINDX exhibits only small correlations (less than 0.15) with the components of GINDX,
whereas GINDX is moderately correlated with the number of directors (positive) and the percent of
non-independent directors (negative). Thus, big boards with many independent directors exhibit a
slight tendency toward lower shareholder rights, especially in the PROTECT and OTHER category.
Descriptive Statistics
Table 2 presents descriptive statistics for the variables used in our analyses. The mean
institutional ownership in the sample firms is 60.5 percent. Two-year changes in institutional
Journal of Management Accounting Research
Volume 26, Number 2, 2014
5706
0.60
2.09
0.33
0.01
0.04
1
5618
0.83
2.29
0.38
0.06
0.15
2
2200
0.87
2.43
0.45
0.27
0.52
3
489
0.93
2.50
0.55
0.62
0.75
4
1332
Total
5
4
3
2
1
84
6%
376
28%
522
39%
266
20%
78
6%
6
0%
1995
0
DINDX
1582
132
8%
473
30%
595
38%
295
19%
79
5%
8
1%
1996
1767
137
8%
572
32%
647
37%
324
18%
78
4%
9
1%
1997
1808
175
10%
565
31%
680
38%
297
16%
81
4%
10
1%
1998
1743
166
10%
600
34%
618
35%
291
17%
60
3%
8
0%
1999
1793
203
11%
671
37%
607
34%
258
14%
50
3%
4
0%
2000
5
52
1449
186
13%
550
38%
501
35%
180
12%
31
2%
1
0%
2001
207
14%
597
41%
514
35%
137
9%
17
1%
1
0%
253
17%
628
42%
490
33%
102
7%
9
1%
3
0%
2003
1485
Total
DELAY
VOTING
PROTECT
OTHER
STATE
Variable
2002
1473
1.00
2.46
0.63
1.00
1.00
Panel B: Time-Series Movements in Governance Index Scores
1827
0.00
2.00
0.29
0.00
0.00
CEO
LNDIR
PNID
DLOCK
DBAD
Total
0
Variable
DINDX Scores
1460
284
19%
674
46%
444
30%
50
3%
6
0%
2
0%
2004
286
0.70
1.57
0.63
0.12
0.77
2–4
Panel A: Mean Values of Governance Variables for Each Level of Governance Index
Total
17–19
14–16
11–13
8–10
5–7
2–4
GINDX
1985
1.39
1.88
1.44
0.36
1.27
5–7
3344
1380
49
4%
307
22%
517
38%
420
30%
86
6%
1
0%
88
5%
481
28%
622
36%
440
26%
83
5%
2
0%
1716
3.03
2.46
2.97
1.60
2.09
11–13
1677
50
3%
384
23%
726
43%
423
25%
90
5%
4
0%
2001–
2002
425
3.45
3.07
3.80
2.24
2.58
14–16
1767
45
3%
384
22%
808
46%
445
25%
81
5%
4
0%
2003–
2004
13
4.00
3.92
3.69
1.85
5.69
17–19
(continued on next page)
1679
54
3%
429
26%
671
40%
438
26%
85
5%
2
0%
1999–
2000
2166
1997–
1998
2.39
2.12
2.12
0.91
1.70
8–10
GINDX Scores
1995–
1996
Descriptive Statistics for Corporate Governance Mechanisms
TABLE 1
130
Bushee, Carter, and Gerakos
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Journal of Management Accounting Research
Volume 26, Number 2, 2014
1.000
0.483
0.595
0.381
0.373
0.454
0.129
0.048
0.001
0.143
0.068
0.083
0.477
1.000
0.057
0.089
0.014
0.008
0.126
0.083
0.031
0.132
0.096
0.029
CEO
0.577
0.057
1.000
0.101
0.110
0.154
0.248
0.107
0.004
0.248
0.183
0.115
LNDIR
0.394
0.089
0.101
1.000
0.122
0.016
0.192
0.117
0.042
0.170
0.202
0.049
PNID
0.422
0.014
0.110
0.122
1.000
0.059
0.033
0.005
0.011
0.033
0.030
0.035
DLOCK
0.483
0.008
0.154
0.016
0.059
1.000
0.019
0.008
0.002
0.029
0.007
0.037
DBAD
0.119
0.125
0.241
0.188
0.035
0.015
1.000
0.618
0.381
0.598
0.651
0.311
GINDX
0.040
0.081
0.103
0.117
0.003
0.008
0.614
1.000
0.283
0.081
0.345
0.121
DELAY
0.012
0.030
0.010
0.050
0.006
0.004
0.404
0.278
1.000
0.053
0.128
0.019
VOTING
0.129
0.123
0.239
0.169
0.031
0.029
0.609
0.083
0.042
1.000
0.338
0.083
PROTECT
0.071
0.096
0.189
0.191
0.029
0.005
0.642
0.333
0.119
0.325
1.000
0.048
OTHER
0.057
0.022
0.095
0.045
0.030
0.024
0.342
0.122
0.023
0.084
0.061
1.000
STATE
Variable Definitions:
DINDX ¼ an index of board of director characteristics and is calculated as the sum of the three indicator variables for whether the CEO is also the chairperson (CEO), whether there
are one or more director interlocks (DLOCK), and whether one or more directors miss 75 percent or more of board meetings (DBAD), and indicators for whether the firm has a
large board size (LNDIR) and a large number of non-independent directors (PNID);
GINDX ¼ an index of shareholder rights and is calculated as the sum of 24 individual corporate charter components. The 24 individual charter components are aggregated into five
subgroups:
DELAY ¼ the sum of four provisions designed to slow down hostile bidders;
VOTING ¼ the sum of six provisions related to shareholder rights in elections or charter amendment votes;
PROTECT ¼ the sum of six provisions that protect officers and directors from firm-related liability and provide termination-related compensation;
OTHER ¼ the sum of six provisions related to greenmail, directors’ duties, fair price, pension parachutes, poison pill, and silver parachutes; and
STATE ¼ whether the firm is incorporated in states with specific anti-takeover laws and, if so, whether the firm chooses to opt out of the laws.
This table presents descriptive statistics for the governance indices and their components. Panel A presents mean values of the components for each level of the respective index.
Panel B presents time-series movements in the levels of the governance indices for the period 1995–2004. Panel C presents correlations among the governance indices and all of
their components. Correlations greater than 0.024 in absolute value are significantly different from 0 at the 0.01 level.
DINDX
CEO
LNDIR
PNID
DLOCK
DBAD
GINDX
DELAY
VOTING
PROTECT
OTHER
STATE
DINDX
Panel C: Pearson (Upper Diagonal) and Spearman (Lower) Correlations among Governance Indices and Governance Variables
TABLE 1 (continued)
Institutional Investor Preferences for Corporate Governance Mechanisms
131
Bushee, Carter, and Gerakos
132
TABLE 2
Descriptive Statistics for Variables Used in Analyses
Variable
Mean
Std. Dev.
Q1
Median
Q3
IH_TOTAL
CIH_TOTAL
DINDX
CDINDX
GINDX
CGINDX
ODOWN
CODOWN
LMV
CLMV
LEV
CLEV
EP
CEP
BP
CBP
DP
CDP
SGR
CSGR
MRET
CMRET
IRISK
CIRISK
BETA
CBETA
TURN
CTURN
SP500
CSP500
RATE
CRATE
ROE
CROE
LTIME
CSHRS
MDAGE
MNA
CEOTURN
0.605
0.049
0.353
0.028
0.486
0.015
0.100
0.010
7.443
0.119
0.249
0.011
0.049
0.009
0.542
0.042
0.012
0.001
0.240
0.064
0.194
0.015
3.682
0.016
1.041
0.048
0.158
0.012
0.281
0.031
4.369
0.230
0.170
0.030
2.660
0.631
4.067
0.623
0.181
0.198
0.104
0.203
0.177
0.139
0.049
0.142
0.062
1.465
0.603
0.188
0.101
0.071
0.086
0.350
0.275
0.017
0.010
0.383
0.383
0.561
0.753
0.444
0.383
0.546
0.450
0.144
0.076
0.450
0.211
2.490
1.274
0.276
0.247
1.077
1.344
0.076
0.839
0.392
0.472
0.003
0.200
0.200
0.368
0.000
0.015
0.013
6.384
0.210
0.081
0.038
0.032
0.026
0.289
0.089
0.000
0.002
0.053
0.179
0.125
0.425
3.997
0.267
0.665
0.228
0.066
0.010
0.000
0.000
1.000
0.000
0.100
0.094
2.092
0.009
4.027
0.000
0.000
0.628
0.039
0.400
0.000
0.474
0.000
0.041
0.001
7.251
0.122
0.244
0.000
0.059
0.002
0.485
0.014
0.004
0.000
0.143
0.029
0.095
0.040
3.697
0.015
0.941
0.041
0.106
0.010
0.000
0.000
5.000
0.000
0.163
0.012
2.779
0.152
4.078
0.000
0.000
0.757
0.089
0.400
0.000
0.579
0.053
0.120
0.005
8.364
0.461
0.374
0.052
0.082
0.021
0.721
0.149
0.019
0.001
0.294
0.090
0.386
0.412
3.373
0.290
1.307
0.334
0.193
0.035
1.000
0.000
6.000
0.000
0.239
0.045
3.437
0.963
4.116
1.000
0.000
This table presents descriptive statistics for all variables used in the empirical analyses. Variables with the prefix ‘‘C’’
represent two-year changes.
Variable Definitions:
IH_TOTAL ¼ percentage of shares outstanding held by all institutional investors;
DINDX ¼ an index of board of director characteristics and is calculated as the sum of the three indicator variables for
whether the CEO is also the chairperson (CEO), whether there are one or more director interlocks (DLOCK), and
whether one or more directors miss 75 percent or more of board meetings (DBAD), and indicators for whether the
firm has a large board size (LNDIR) and a large number of non-independent directors (PNID);
(continued on next page)
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
133
TABLE 2 (continued)
GINDX ¼ an index of shareholder rights and is calculated as the sum of 24 individual corporate charter components. Both
GINDX and DINDX are divided by their maximum values, so they range between 0 and 1;
ODOWN ¼ percentage of shares outstanding held by officers and directors;
LMV ¼ natural log of the market value of equity (CS#24 3 CS#25);
LEV ¼ ratio of debt (CS#34 þ CS#9) to total assets (CS#6);
EP ¼ ratio of income before extraordinary items (CS#18) to the market value of equity (CS#24 3 CS#25);
BP ¼ ratio of the book value of equity (CS#60) to the market value of equity (CS#24 3 CS#25);
DP ¼ ratio of dividends (CS#21) to the market value of equity (CS#24 3 CS#25);
SGR ¼ percentage change in sales (CS#12);
MRET ¼ market adjusted buy-and-hold stock return measured over a year’s time;
IRISK ¼ log of the standard deviations of the market-model residuals of daily stock returns measured over a year’s time;
BETA ¼ market model beta calculated from daily stock returns measured over a year’s time;
TURN ¼ the average monthly trading volume relative to total shares outstanding measured over a year’s time;
SP500 ¼ an indicator variable set to 1 if the firm is in the S&P 500 index, and 0 otherwise;
RATE ¼ S&P stock rating (9 ¼ Aþ,. . .,1 ¼ not rated);
ROE ¼ ratio of income before extraordinary items (CS#18) to the book value of equity (CS#60);
CSHRS ¼ change in the shares outstanding;
MDAGE ¼ mean director age;
MNA ¼ an indicator for the number of mergers and acquisitions activities that occurred (AFTNT#1 populated with
‘‘AA,’’ ‘‘AB,’’ ‘‘AR,’’ and ‘‘AS’’) during the period; and
CEOTURN ¼ an indicator variable set to 1 if there is a turnover in the CEO position during the two year window, and 0
otherwise.
ownership (4.9 percent) over the 1998–2004 period have been positive, on average, consistent with
the long-term trend in increasing institutional ownership. The mean and median values for DINDX
and GINDX differ from those reported in Table 1 because we divide each index by the maximum
value. This transformation causes the variable to range between 0 and 1, which aids in the
interpretation of the coefficients. Mean changes in the two governance indices (CDINDX and
CGINDX) are small and the median changes are 0. These small, relatively infrequent changes in
governance indices reduce the power of our changes analyses.
TOTAL INSTITUTIONAL OWNERSHIP AND CORPORATE GOVERNANCE
We first examine the relation between total institutional investor ownership and governance
mechanisms by regressing the percentage of total institutional ownership in a firm on a governance
measure—DINDX or GINDX—and a set of control variables that capture previously documented
determinants of institutional ownership (Bushee 2001; Gompers and Metrick 2001). Prior research
finds that, on average, institutions prefer large, liquid stocks that can be justified as prudent
investments. We control for size with the log of the market value of equity (LMV). We also proxy
for the prudence of the investment by including the S&P 500 stock rating (RATE) and an indicator
variable for whether a firm is listed in the S&P 500 Index (SP500). We use turnover (TURN),
calculated as the average monthly trading volume over the year divided by shares outstanding, to
control for liquidity preferences of institutions.
To control for institutions’ preferences for good recent performance, we include annual market
adjusted returns (MRET). We also include the following fundamental growth and income ratios,
upon which institutional investors base their trading decisions: the earnings-to-price ratio (EP), the
book-to-price ratio (BP), the dividend-to-price ratio (DP), sales growth (SGR), and the return on
equity (ROE). To proxy for risk, we include beta (BETA), calculated from a market model using
daily returns, idiosyncratic risk (IRISK), calculated as the standard deviation of the market model
residuals, and leverage (LEV), calculated as the debt-to-asset ratio. As an additional control
variable, we include the percentage of officer and director ownership (ODOWN). Ideally, we would
include director ownership as an indicator of the effectiveness of corporate governance. Because of
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Bushee, Carter, and Gerakos
134
the mechanical (negative) relation between institutional holdings and ownership, we include it as a
control variable instead.
We estimate the following regression for the period 1998 to 2004 using Rogers (1993) robust
standard errors:
IH TOTALit ¼ a þ b1 GOVit þ b2 LMVit þ b3 RATEit þ b4 SP500it þ b5 TURNit þ b6 MRETit
þ b7 EPit þ b8 BPit þ b9 DPit þ b10 SGRit þ b11 ROEit þ b12 BETAit
4
X
þ b13 IRISKit þ b14 LEVit þ b15 ODOWNit þ
b16þk DYEARit þ ejt ;
k¼0
where IH_TOTAL ¼ percent of institutional ownership by all institutions; GOV ¼ GINDX or
DINDX; and DYEAR ¼ year indicator.7 Recall that lower DINDX and GINDX scores represent
‘‘better’’ governance; a negative b1 coefficient indicates preferences for ‘‘better’’ governance.
Table 3 presents the results for these regressions. We find that the total level of institutional
ownership is negatively related to DINDX (with a two-tailed p-value of 0.062), but not significantly
related to GINDX.8 In terms of economic magnitude, a one-standard-deviation decrease in DINDX
is associated with a 0.6 percent increase in total institutional ownership. Thus, on average, firms
with ‘‘better’’ governance in terms of board characteristics tend to have higher levels of total
institutional ownership, but the effect is not large in either statistical or economic magnitude.9
Our analysis suggests weak governance sensitivity by institutional investors as a group. We,
therefore, identify those institutions for which ‘‘better’’ corporate governance complements their
investment decisions to provide greater insight into the characteristics of these institutions and their
governance preferences. This analysis provides insight into the importance of corporate governance
to particular institutional investors.
CLASSIFYING INSTITUTIONAL INVESTORS BASED ON REVEALED
PREFERENCES FOR GOVERNANCE
Classification
In this section, we identify individual institutional investors that are governance-sensitive. We
classify institutional investors as ‘‘governance-sensitive’’ if they significantly tilt their portfolio
weights toward firms with ‘‘better’’ corporate governance, as measured by board characteristics
(DINDX) and shareholder rights (GINDX). Governance sensitivity is likely to be a second-order
effect in choosing of portfolio weights based on the strong evidence that factors such as size,
performance, risk, and liquidity are first-order determinants of institutional investment (Bushee
2001; Gompers and Metrick 2001). Because these determinants could be correlated with
governance characteristics, we control for them to determine whether governance mechanisms have
an incremental impact on an institution’s investment decisions. We estimate the following Tobit
regression annually for each institutional investor, and base our classification on the sign and
significance of the coefficient on the governance proxy:
7
GINDX and DINDX are included separately so that we may investigate future changes in these variables.
We also estimate annual regressions. The coefficient on DINDX is negative every year and significant at the 0.10
level only in 1999 and 2000. The coefficient on GINDX is negative five out of the seven years and not significant
in any of the years.
9
We also test whether changes in total institutional ownership are associated with changes in governance indices.
Results for these tests are consistent for those of the levels tests—we find only limited evidence of
contemporaneous changes in total institutional ownership and changes in governance indices.
8
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
135
TABLE 3
Level of Total Institutional Ownership and Level of Corporate Governance
IH_TOTAL
Intercept
DINDX
GINDX
LMV
RATE
SP500
TURN
MRET
EP
BP
DP
SGR
ROE
BETA
IRISK
LEV
ODOWN
Adj. R2
n
0.433***
0.031*
0.002
0.009***
0.030***
0.219***
0.011**
0.227***
0.013
3.811***
0.019**
0.042***
0.004
0.057***
0.085***
0.438***
0.328
8992
0.436***
0.001
0.003
0.010***
0.030***
0.222***
0.011***
0.222***
0.016
3.838***
0.019**
0.041***
0.004
0.057***
0.082***
0.443***
0.327
8992
*, **, *** Signify difference from 0 at the 10 percent, 5 percent, and 1 percent levels, respectively (two-tailed test).
This table presents results of OLS regressions of total institutional ownership on the governance indices for the period
1998–2004. Regressions are estimated using Rogers (1993) robust standard errors to control for firm-specific
dependence. Included in regressions, but not tabulated, are year dummies. Definitions of control variables are provided in
Table 2.
Variable Definitions:
IH_TOTAL ¼ the percentage of shares outstanding held by all institutional investors;
DINDX ¼ an index of board of director characteristics and is calculated as the sum of the three indicator variables CEO,
DLOCK, and DBAD, and indicators for whether the firm has a high level of LNDIR and PNID;
GINDX ¼ an index of shareholder rights and is calculated as the sum of 24 individual corporate charter components and
state laws relating primarily to takeover protections, voting rules, and liability limitations.
PWGTijt ¼ ajt þ cjt GOVijt þ b1jt LMVijt þ b2jt RATEijt þ b3jt SP500ijt þ b4jt LTIMEijt
þ b5jt TURNijt þ b6jt MRETijt þ b7jt EPijt þ b8jt BPijt þ b9jt DPijt þ b10jt SGRijt
þ b11jt ROEijt þ b12jt BETAijt þ b13jt IRISKijt þ b14jt LEVijt þ ejt
where PWGT ¼ institution j’s portfolio weight in firm i at time t; and GOV ¼ DINDX or GINDX.10
We estimate each institution-specific regression on the entire panel of firms for which we have
data for governance and control variables. The portfolio weight (PWGT) is the percentage of the
institution’s total equity portfolio that is invested in a firm. If an institution has no investment in a
sample firm, the portfolio weight equals zero. Because few institutions engage in short-selling,
10
We do not include ODOWN in this regression because it is missing in 1995 and 1996. When we estimate this
regression after 1996, both including and excluding ODOWN, we find very similar proportions of firms with
significant coefficients on the governance variables.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
136
Bushee, Carter, and Gerakos
these zero weights represent a truncated distribution and a Tobit model is the appropriate
specification. In cases where an institution owns only a small number of the sample firms, the Tobit
model will not converge and we are unable to classify the institution.
We estimate the model for each year from 1995–1997, and label an institutional investor as
‘‘governance-sensitive’’ if the cjt coefficient is negative and significant at the 0.10 level (one-tailed)
in at least two of the three years. We label these institutions as GSID (GSIG) if they are sensitive to
DINDX (GINDX). If the cjt coefficient is insignificant and/or positive in each year during 1995–
1997, we classify the institution as ‘‘governance-insensitive’’ (GIND and GING for insensitivity to
DINDX and GINDX, respectively). For institutions with only one year of data during 1995–1997 or
with multiple years of data, but only one negative and significant cjt coefficient, we do not attempt
to classify the institution because of the uncertainty over its governance preferences. We perform
the classification on the 1995–1997 period to allow for a holdout sample to test out-of-sample
validity. In addition, this period arguably provides a more powerful setting to assess governance
preferences because there were no major scandals or changes in regulation that could have induced
a stronger focus on governance.
Panel A of Table 4 presents the number of institutions classified into these three groups. No
more than 11 percent of institutional investors during the 1995–1997 period explicitly tilt their
portfolio weights based on ‘‘better’’ governance mechanisms: 11 percent are sensitive to board of
director characteristics (DINDX) and 9 percent are sensitive to shareholder rights (GINDX).11,12
Thus, only a small percentage of institutions consistently incorporate firms’ governance
characteristics in their portfolio weights. Moreover, only 23 institutions are classified as
governance-sensitive to both shareholder rights and board characteristics, suggesting that even if
institutions choose to tilt their portfolio toward governance characteristics, they generally focus on
only one aspect of governance.
Because only 11 percent of institutions significantly tilt their portfolio weights toward DINDX,
the results from the total institutional ownership regressions presented in Table 3 suggest that either
these institutions tend to take significant ownership stakes or that enough institutions slightly tilt
toward governance characteristics (i.e., have a negative, but insignificant, cjt coefficient in the
classification regression) to produce an overall effect on percentage ownership. Alternatively, this
result could be driven by institutions that are classified as insensitive to governance during 1995–
1997 becoming more sensitive to governance in the holdout sample. To investigate this latter
possibility, we estimate a three-year rolling classification over the full sample period. As presented
in Panel B, the number of institutions classified as sensitive to DINDX is fairly constant in the
sample period. The relatively constant percentages of institutions sensitive to DINDX may be driven
by the secular improvement in DINDX, which may have reduced the incentives for institutions to
tilt their portfolios toward ‘‘better’’ board characteristics. In contrast, there is a secular increase in
the number of institutions sensitive to GINDX over the sample period: from 9 percent in 1997 up to
20 percent in 2002 and 2003. As the distribution of GINDX was relatively constant over the sample
period, the increased focus on governance after Enron and Sarbanes-Oxley possibly provided an
incentive for more institutions to tilt their portfolios toward firms with ‘‘better’’ shareholder rights.13
11
We find similar results when we compare the percentage of total market capitalization of all institutions in each
category: 11 percent are sensitive to board of director characteristics (DINDX) and 9 percent are sensitive to
shareholder rights (GINDX).
12
One potential concern about our classification method arises from the fact that approximately 10 percent of
institutions are sensitive to DINDX or GINDX, and our statistical cutoff for sensitivity is 10 percent. Note,
however, that we require that the coefficient on governance be negative and significant for at least two out of
three years and that we find persistence in governance sensitivity during the holdout period, thereby reducing the
possibility that governance-sensitive classification is driven by random sampling variability.
13
We also estimated all of our analyses using this alternative classification method and found similar results.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
137
TABLE 4
Revealed Preferences for Corporate Governance
Panel A: Number of Institutions Classified as Governance-Sensitive Based on 1995–1997
Coefficients
Frequency Percent
Frequency Percent
Sensitive to DINDX (GSID)
Insensitive to DINDX (GIND)
Not classified (NC)
149
809
354
11%
62%
27%
Sensitive to GINDX (GSIG)
Insensitive to GINDX (GING)
Not classified (NC)
122
853
334
9%
65%
26%
Total
1312
100%
Total
1309
100%
Panel B: Time-Series of the Percentage of Institutions Classified as Governance-Sensitive
1997
1998
1999
2000
2001
2002
2003
2004
Sensitive to DINDX (GSID)
Insensitive to DINDX (GIND)
Not classified (NC)
11%
62%
27%
13%
59%
27%
14%
55%
31%
14%
55%
31%
11%
58%
31%
12%
58%
30%
9%
61%
30%
10%
60%
30%
Sensitive to GINDX (GSIG)
Insensitive to GINDX (GING)
Not classified (NC)
9%
65%
26%
10%
66%
25%
10%
65%
25%
11%
62%
26%
13%
58%
28%
20%
53%
27%
20%
56%
24%
16%
60%
24%
Panel C: Percentage Ownership of Firm by Governance-Sensitive and -Insensitive
Institutions
Mean
Std. Dev.
Q1
Median
Q3
Ownership sensitive to DINDX (IH_GSID)
Change in ownership sensitive to DINDX (CIH_GSID)
Ownership insensitive to DINDX (IH_GIND)
0.112
0.009
0.273
0.065
0.049
0.117
0.064
0.017
0.190
0.103
0.007
0.271
0.149
0.035
0.348
Ownership sensitive to GINDX (IH_GSIG)
Change in ownership sensitive to DINDX (CIH_GSID)
Ownership insensitive to GINDX (IH_GING)
0.188
0.017
0.244
0.084
0.058
0.110
0.130
0.017
0.166
0.182
0.015
0.240
0.239
0.049
0.319
This table presents results of classifications of institutional investor sensitivity to governance mechanisms. Sensitivity is
measured by regressing portfolio weights on governance indices and control variables. Regressions are estimated using
the Tobit model to account for the truncation of portfolio weights at 0:
PWGTijt ¼ ajt þ cjt GOVijt þ b1jt LMVijt þ b2jt RATEijt þ b3jt SP500ijt þ b4jt LTIMEijt þ b5jt TURNijt þ b6jt MRETijt
þ b7jt EPijt þ b8jt BPijt þ b9jt DPijt þ b10jt SGRijt þ b11jt ROEijt þ b12jt BETAijt þ b13jt IRISKijt
þ b14jt LEVijt þ ejt :
Institutions are classified as governance-sensitive based on the sign and significance of the coefficient on GOV over the
three-year window for the period 1995–1997. GOV is either DINDX, the index of board of director characteristics, or
GINDX, the index of shareholder rights. If the coefficient on GOV is negative and significant (at the 0.10 level, onetailed) for two years during the window, with a minimum of two years of data required, then an institution is classified as
either GSID or GSIG based on the governance index (DINDX or GINDX) used in the regression. If the coefficient on
GOV is negative and significant (at the 0.10 level, one-tailed) for only one year during the window, then an institution is
classified as NC and not included in further analyses. All other institutions with a minimum of two years of data are
classified as GIND or GING based on the governance index (DINDX or GINDX) used in the regression. Panel A presents
the number of institutions classified as governance-sensitive. Panel B presents rolling classifications of institutions based
on their governance sensitivity. Panel C presents the descriptive statistics of the percent of shares outstanding held by
governance-sensitive and -insensitive institutions. Ownership percentages are calculated over the period 1998–2004.
Variables with the prefix ‘‘C’’ represent two-year changes.
(continued on next page)
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Bushee, Carter, and Gerakos
138
TABLE 4 (continued)
Variable Definitions:
IH_GSID ¼ the percentage of shares outstanding held by institutions that are sensitive to board characteristics;
IH_GIND ¼ the percentage of shares outstanding held by institutions that are insensitive to board characteristics;
IH_GSIG ¼ the percentage of shares outstanding held by institutions that are sensitive to shareholder rights; and
IH_GING ¼ the percentage of shares outstanding held by institutions that are insensitive to shareholder rights.
Panel C presents firm-level institutional ownership classified by governance-sensitivity.
Despite the fact that institutions sensitive to shareholder rights under our classification method
comprise no more than 10 percent of institutions, their average holdings in the sample firms
(IH_GSIG) are 19 percent, which is not far below the 24 percent average holdings of insensitive
institutions (IH_GING). This is not surprising because shareholder rights are likely to be an
important factor for institutions holding large stakes. In contrast, ownership levels for institutions
that are sensitive to board characteristics (IH_GSID) are, on average, similar to the percentage of
institutions classified as GSID.
Levels Analyses
Panel A of Table 5 presents results for institutions classified by their prior sensitivity to board
characteristics and shareholder rights. We use a specification that is similar to the one used in the
third section, ‘‘Total Institutional Ownership and Corporate Governance,’’ but replace the
dependent variable with the percentage ownership by institutions in the relevant governance
classification (IH_GSID, IH_GIND, IH_GSIG, and IH_GING) and control for the total level of
institutional ownership:
IH GOVSENit ¼ a þ b1 GOVit þ b2 IH OTHERit þ b3 LMVit þ b4 RATEit þ b5 SP500it
þ b6 TURNit þ b7 MRETit þ b8 EPit þ b9 BPit þ b10 DPit þ b11 SGRit
þ b12 ROEit þ b13 BETAit þ b14 IRISKit þ b15 LEVit þ b16 ODOWNit
4
X
þ
b17þk DYEARit þ ejt ;
k¼0
where IH_GOVSEN ¼ percent of institutional ownership held by the relevant class of institutions
(GSID, GIND, GSIG, and GING); IH_OTHER ¼ percent of institutional ownership by all
institutions less the level of the ownership of the relevant governance-sensitive subgroup; and GOV
¼ GINDX or DINDX.
Ownership by institutions classified as sensitive to board characteristics (IH_GSID) is
significantly negatively associated with DINDX, whereas ownership by institutions insensitive to
board characteristics (IH_GIND) is not significantly related to DINDX. Holding total institutional
ownership constant, a one-standard-deviation decrease in DINDX is associated with a 0.4 percent
increase in IH_GSID. This is two-thirds of the effect for the association between total institutional
ownership and DINDX shown in Panel A, Table 3. When we classify based on sensitivity to
shareholder rights, ownership by governance-sensitive (-insensitive) institutions is significantly
negatively (positively) related to GINDX. Holding total institutional ownership constant, a onestandard-deviation decrease in GINDX is associated with a 0.3 percent increase in IH_GSIG.
These results confirm that our classification methodology has descriptive validity out-ofsample and with firm-level percentage ownership, rather than the institution-level portfolio weights,
as the measure of institutional investment. In addition, the results show that the association between
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
139
TABLE 5
Governance-Sensitive Institutions and Corporate Governance
Panel A: Levels of Governance-Sensitive Institutional Ownership and Governance Indices
Intercept
DINDX
GINDX
IH_OTHER
LMV
RATE
SP500
TURN
MRET
EP
BP
DP
SGR
ROE
BETA
LEV
IRISK
ODOWN
Adj. R2
n
IH_GSID
IH_GIND
IH_GSIG
IH_GING
0.101***
0.018***
0.169***
0.001
0.070***
0.292***
0.063***
0.001
0.001**
0.004
0.047***
0.007***
0.023
0.015***
0.507***
0.000
0.000
0.002
0.002
0.001
0.102***
0.012
0.005*
0.004***
0.007
0.081***
0.004
0.110***
0.008
2.480***
0.009*
0.029***
0.004
0.048***
0.063***
0.198***
0.025**
0.066***
0.005***
0.003***
0.028***
0.038***
0.005***
0.037**
0.024***
0.764***
0.011***
0.012**
0.004
0.014***
0.030***
0.122***
0.054***
0.220***
0.006**
0.005***
0.008
0.132***
0.006*
0.187***
0.025***
2.765***
0.004
0.021**
0.011*
0.041***
0.041***
0.276***
0.152
8992
0.141
8992
0.195
8992
0.170
8992
Panel B: Levels and Changes in Governance-Sensitive Institutional Ownership and Levels
and Changes in Corporate Governance Indices
Intercept
CIH_GSID
IH_GSID
CDINDX
DINDX
CIH_OTHER
IH_OTHER
CLMV
CBP
CMRET
CIRISK
CSHRS
MDAGE
MNA
CEOTURN
CLEV
CEP
CDINDX
(1)
CIH_GSID
(2)
0.120
0.093**
0.054
0.034***
0.315***
0.001
0.005
0.008*
0.003
0.002
0.009
0.001
0.048*
0.003
0.024***
0.291***
0.007**
0.019***
0.073***
0.027***
0.013***
0.001
0.003***
0.009***
0.001
0.003
0.028***
Intercept
CIH_GSIG
IH_GSIG
CGINDX
GINDX
CIH_OTHER
IH_OTHER
CLMV
CBP
CMRET
CIRISK
CSHRS
MDAGE
MNA
CEOTURN
CLEV
CEP
CGINDX
(3)
CIH_GSIG
(4)
0.092***
0.024**
0.050***
0.062***
0.080***
0.012*
0.003
0.002
0.002
0.002***
0.002
0.001
0.009
0.003***
0.002
0.279***
0.036**
0.009**
0.154***
0.041***
0.010***
0.009***
0.002*
0.012***
0.002***
0.002
0.010
(continued on next page)
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Bushee, Carter, and Gerakos
140
TABLE 5 (continued)
CDINDX
(1)
0.009
0.003**
0.056***
0.004
0.000
0.004*
0.002*
0.016*
CDP
CSGR
CTURN
CSP500
CRATE
CROE
CBETA
CODOWN
Adj. R2
n
CIH_GSID
(2)
0.150
7608
0.216
7608
CGINDX
(3)
0.194***
0.000
0.061***
0.002
0.000
0.006*
0.000
0.026**
CDP
CSGR
CTURN
CSP500
CRATE
CROE
CBETA
CODOWN
Adj. R2
N
CIH_GSIG
(4)
0.113
6415
0.233
6415
*, **, *** Signify difference from 0 at the 10 percent, 5 percent, and 1 percent levels, respectively (two-tailed test).
This table presents results of regressions of governance-sensitive institutional ownership and governance indices. Panel
A presents OLS regressions of governance-sensitive institutional ownership on governance indices and control variables
for the period 1998–2004. Panel B presents OLS regressions of changes in governance indices and changes in
governance-sensitive institutional ownership for the period 1999–2004. Regressions are estimated using Rogers (1993)
robust standard errors to control for firm-specific dependence. Included in regressions, but not tabulated, are year
dummies. Variables without prefixes are levels at the beginning of the change period. Variables with the prefix of ‘‘C’’
are concurrent two-year changes. Definitions of control variables are provided in Table 2.
Variable Definitions:
IH_GSID (IH_GIND) ¼ the percentage of shares outstanding held by all institutions that are sensitive (insensitive) to
DINDX;
IH_GSIG (IH_GING) ¼ the percentage of shares outstanding held by all institutions that are sensitive (insensitive) to
GINDX;
DINDX ¼ an index of board of director characteristics and is calculated as the sum of the three indicator variables CEO,
DLOCK, and DBAD, and indicators for whether the firm has a high level of LNDIR and PNID; and
GINDX ¼ an index of shareholder rights and is calculated as the sum of 24 individual corporate charter components and
state laws relating primarily to takeover protections, voting rules, and liability limitations.
total institutional ownership and DINDX in Panel A, Table 3, is primarily driven by the governancesensitive institutions, and that conflicting preferences of institutions classified as GSIG and GING
drive the insignificant coefficient on GINDX in Panel A of Table 3.
Changes Analyses
In this section, we expand on the model in the prior section to test whether levels of, and
changes in, governance-sensitive ownership are associated with contemporaneous changes in, and
prior levels of, governance mechanisms. We include both levels and changes to test whether the
results in the prior section are driven by governance-sensitive institutions investing in firms with
preferred governance mechanisms or by governance-sensitive institutions actively implementing
preferred governance mechanisms. For example, a finding that the level of institutional ownership is
associated with future changes in governance would be suggestive of institutional activism, while
contemporaneous changes are suggestive of institutions ‘‘voting with their feet.’’ Such a model
introduces the possibility of an association in both directions; i.e., changes in institutional
ownership both drive and respond to changes in governance.
We estimate the regressions over the period 1999 to 2004 using Rogers (1993) robust standard
errors. Changes in institutional ownership, governance indices, and the control variables are all
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
141
measured as two-year changes. Levels of institutional ownership, governance indices, and the
control variables are measured at the beginning of the change period. We add the change in shares
outstanding (CSHRS) to control for any new equity issues or repurchase programs that could affect
the change in institutional ownership. We also include an indicator for the number of mergers and
acquisitions (MNA) and an indicator for CEO turnover (CEOTURN) to proxy for any major changes
in the company management or its capital structure that could lead to changes in governance. We
estimate the following regressions using OLS:
CGOVit ¼ a þ b1 CIH GOVSENit þ b2 IH GOVSENit þ b3 GOVit þ b4 CIH OTHERit
þ b5 IH OTHERit þ b6 CLMVit þ b7 CBPit þ b8 MRETit þ b9 CIRISKit
4
X
þ b10 CSHRSit þ b11 MDAGEit þ b12 MNAit þ b13 CEOTURNit þ
b14þk DYEARit
k¼0
þ eit ;
CIH GOVSENit ¼ a þ b1 CGOVit þ b2 GOVit þ b3 IH GOVSENit þ b4 CIH OTHERit
þ b5 IH OTHERit þ b6 CLMVit þ b7 CBPit þ b8 MRETit þ b9 CIRISKit
þ b10 CSHRSit þ b11 CLEVit þ b12 CEPit þ b13 CDPit þ b14 CSGRit
þ b15 CTVOLit þ b16 CSP500it þ b17 CRATEit þ b18 CROEit þ b19 CBTAit
4
X
þ b20 CODOWNit þ
b21þk DYEARit þ eit ;
k¼0
where GOV ¼ DINDX or GINDX; IH_GOVSEN ¼ percent of institutional ownership by
governance-sensitive institutions (GSID and GSIG); and IH_OTHER ¼ percent of institutional
ownership by all institutions less the level of the ownership of the relevant governance-sensitive
subgroup. Variables with a prefix of ‘‘C’’ are two-year changes. All other variables are prior
levels. We include prior levels of governance and institutional ownership variables to control for
situations in which changes are constrained (e.g., firms with the best governance score cannot
improve their governance) and to capture any changes in response to existing levels (e.g.,
institutions buying firms with existing ‘‘better’’ governance, but no concurrent changes in
governance).
Panel B of Table 5 provides results for governance-sensitive institutions using OLS. In the first
and third columns, the results show that both the level of and change in ownership by governancesensitive institutions are significantly negatively associated with changes in governance mechanisms
for both board characteristics and shareholder rights. Thus, changes in ownership by governancesensitive institutions have a significant incremental effect on governance improvements beyond the
prior ownership levels. Changes in ownership by other institutions exhibit no significant association
with improvements in governance, consistent with the results for the level of ownership. In the
second and fourth columns, the results show that changes in governance mechanisms are
significantly associated with changes in ownership by governance-sensitive institutions for both
board characteristics and shareholder rights. In addition, the change in GSID (GSIG) ownership is
associated with the prior level of DINDX (GINDX), implying that governance-sensitive institutions
accumulate holdings in firms with preferred governance mechanisms.14
We find significant evidence of a contemporaneous association between changes in ownership
by governance-sensitive institutions and changes in other governance mechanisms. In addition, we
find that governance-sensitive institutions accumulate holdings in firms with ‘‘better’’ governance.
14
In unreported tests, we attempt to estimate the models 2SLS, but our findings generally lack significance, likely
due to the inability to find suitable instruments.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Bushee, Carter, and Gerakos
142
Finally, we find evidence consistent with governance-sensitive institutions actively improving
shareholder rights in their portfolio firms. However, as these tests document associations in
contemporaneous changes, we are unable to determine causality.
CHARACTERISTICS OF GOVERNANCE-SENSITIVE INSTITUTIONS
In this section, we investigate the characteristics of governance-sensitive institutions. We first
examine whether governance sensitivity is associated with an institution’s legal type. Next, we compare
the portfolio characteristics of governance-sensitive and -insensitive institutions. The analyses are
exploratory in that they investigate which characteristics and effects are associated with governance
sensitivity, as opposed to what determines governance sensitivity. The resulting stylized facts do,
however, provide insight into the incentives that can lead an institution to be governance-sensitive.
Legal Type
We first classify institutional investors based on legal type, using the Spectrum database. The
database identifies bank trusts (BNK), insurance companies (INS), investment companies,
independent investment advisors, and other. Because investment companies and independent
investment advisers are both governed by the Investment Company Act of 1940 and have similar
low levels of fiduciary responsibility, we combine them to form a group called investment advisers
(IA). In addition, we identify the pensions and endowments (PNE) as the corporate or private
pensions, public pensions, and university and foundation endowments within the ‘‘other’’ group.15
Note that these holdings only represent internally managed investments; any externally-managed
investments will be recorded as holdings by investment advisers, which often serve as external
managers for pension and endowments in addition to managing mutual funds.
Although all fund managers are legally considered fiduciaries, the strictness of the prudent
person standard differs depending on the legal form of the institution. Because of state trust laws
and the Employee Retirement Income Security Act (ERISA), bank trusts and pensions face a higher
standard of prudence, including the requirement that each investment be analyzed individually, than
standards faced by investment advisers and insurance companies (Del Guercio 1996). Failure to
adhere to the standard of prudence can lead to investor lawsuits. Therefore, those institutional
investors subject to more stringent fiduciary standards likely have greater preferences for ‘‘better’’
corporate governance mechanisms as a defense against investor lawsuits. We, therefore, predict that
bank trusts and pensions and endowments are more likely to be sensitive to governance than
insurance companies and investment advisors.16
Table 6 presents a cross-tabulation of our governance sensitivity classification with the
classification by legal type. There is significant heterogeneity across types in terms of governance
sensitivity. Although Chi-square tests show that the distribution of legal types among governancesensitive institutions significantly differs from the distribution for all institutions, no one type is
dominated by governance-sensitive institutions. As expected, a high percentage of governancesensitive institutions within a legal type are found among bank trusts (BNK) and pensions and
endowments (PNE), consistent with their fiduciary incentives to demonstrate prudence in selecting
portfolio firms. Pensions and endowments exhibit an especially high percentage of institutions
15
Among the ‘‘other’’ category, there are also law firms, individuals acting as institutions, and other miscellaneous
institutions that are difficult to classify. We do not include these in any of our analyses.
16
Bank trusts may be sensitive to pressure from portfolio firms because of other business relations, such as banking
and lending services (Brickley et al. 1988). As a result, banks may be more likely to acquire firms with better
governance and less likely to actively implement governance changes in portfolio firms.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
143
TABLE 6
Legal Type and Governance Sensitivity
BNK
INS
IA
PNE
Total
Prob. (v2)
GSID
GIND
NC
Total
GSIG
GING
NC
Total
31
21.5%
10
6.9%
95
66.0%
8
5.6%
94
12.2%
42
5.6%
615
77.8%
34
4.4%
51
14.0%
16
4.3%
261
77.6%
14
4.1%
176
13.9%
68
5.4%
971
76.4%
56
4.4%
19
16.1%
9
7.6%
76
64.4%
14
11.9%
120
14.9%
46
5.9%
624
74.5%
36
4.7%
37
12.1%
13
4.2%
269
81.2%
6
2.5%
176
13.9%
68
5.4%
969
76.4%
56
4.4%
144
0.024
785
0.660
342
0.835
1271
118
0.000
826
0.753
325
0.037
1269
This table classifies governance-sensitive institutions by their legal type: bank trusts (BNK), insurance companies (INS),
investment advisers (IA), and corporate pension funds, private pension funds, public pension funds, and university and
foundation endowments (PNE). Chi-square tests examine whether the distribution of legal types among governancesensitive institutions is significantly different from the distribution of legal types among all institutions.
sensitive to shareholder rights, which have been a traditional target of public pensions like
CalPERS, while bank trusts show a high percentage of institutions sensitive to board characteristics.
Investment advisers (IA), which are largely exempt from fiduciary responsibility, show lower
rates of governance sensitivity. Insurance companies (INS), which can face some fiduciary
responsibility, show slightly higher governance sensitivity. For every type except pensions and
endowments, a lower percentage of institutions are sensitive to shareholder rights than to board
characteristics.
Despite the fact that pensions and endowments and bank trusts tend to have the highest
percentage of governance-sensitive institutions, neither of these types has more than 25 percent of
its institutions classified as governance-sensitive, indicating that the legal type classifications do not
proxy for general governance sensitivity. Therefore, we next investigate portfolio characteristics of
governance-sensitive and -insensitive institutions.17
Portfolio Characteristics
To measure the characteristics of the institution’s portfolio, we use factor analysis to create
seven composite measures: institution size, portfolio turnover, size of investment positions, market
capitalization of the portfolio firms, prudence of investments, value versus growth preferences, and
riskiness of portfolio firms. Bushee (2001) and Abarbanell, Bushee, and Raedy (2003) use these
factors to combine a large number of variables that have been used in prior literature into a
parsimonious set of factors that describe institutional investor portfolios. Table 7 provides
definitions of all of the variables that comprise each factor.
The first three factors capture the institutional investors’ decisions with respect to how they
manage their portfolio (Bushee 2001). We measure the institution’s size with a factor, ISIZE, which
17
In additional analyses, we regress the levels of bank and pensions and endowment ownership on governance
indices and control variables (similar to Table 5). In these tests, we find that the levels of bank and pension
ownership are positively associated with the governance indices, providing further evidence that legal type is not
a good proxy for governance sensitivity because it suggests preferences for ‘‘weaker’’ governance.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Bushee, Carter, and Gerakos
144
TABLE 7
Portfolio Characteristics and Governance Sensitivity
Panel A: Portfolio Characteristics of Institutions Classified by their Governance Sensitivity
ISIZE
NSTK
TE
PTURN
PT1
PT2
STABPN
STABPH
BLOCK
LBPH
LBPN
WAPH
FSIZE
WAMC
WASP
WATIME
WAEPRC
PRUDENCE
WADUP
WAPED
WARATE
WADE
VALUE
WAEP
WADP
WABP
RISK
WAEGR
WASGR
WABTA
WASTD
GSID
GIND
Difference
GSIG
GING
Difference
0.500
5.444
13.946
0.165
0.434
0.263
0.674
0.580
0.130
0.054
0.022
0.010
0.660
9.328
0.692
8.684
4.240
0.385
0.285
0.947
6.752
0.278
0.960
0.043
0.019
0.299
0.149
0.240
0.181
1.016
0.075
0.066
4.816
13.409
0.063
0.462
0.287
0.638
0.472
0.028
0.072
0.028
0.014
0.189
8.816
0.583
8.553
4.091
0.089
0.255
0.932
6.339
0.327
0.899
0.042
0.020
0.333
0.372
0.270
0.200
1.024
0.082
0.434***
0.628***
0.537***
0.228***
0.028
0.024
0.036
0.108***
0.158*
0.018
0.006
0.004*
0.471***
0.512***
0.109***
0.131***
0.149*
0.296***
0.030***
0.015**
0.413***
0.049***
0.061
0.001
0.001
0.034***
0.223**
0.030**
0.019*
0.008
0.007***
0.774
5.767
14.554
0.242
0.414
0.239
0.708
0.605
0.036
0.077
0.034
0.016
0.445
9.136
0.642
8.623
4.322
0.319
0.288
0.948
6.649
0.299
0.905
0.044
0.020
0.317
0.259
0.249
0.184
1.019
0.078
0.052
4.838
13.384
0.056
0.456
0.279
0.625
0.472
0.017
0.067
0.024
0.013
0.261
8.894
0.600
8.577
4.098
0.118
0.252
0.934
6.392
0.315
0.913
0.042
0.020
0.329
0.322
0.265
0.196
1.017
0.080
0.722***
0.929***
1.170***
0.298***
0.042*
0.040*
0.083***
0.133***
0.053
0.010
0.010
0.003
0.184*
0.242**
0.042*
0.046
0.224***
0.201**
0.036***
0.014*
0.257***
0.016
0.008
0.002
0.000
0.012
0.063
0.016
0.012
0.002
0.002
Panel B: Multivariate Comparisons of Portfolio Characteristics Based on Governance
Sensitivity
ISIZE
PTURN
BLOCK
FSIZE
PRUDENCE
VALUE
RISK
BNK
GSID
GSIG
0.062***
0.018
0.032
0.003
0.008
0.126***
0.042
0.009
0.076***
0.040***
0.011
0.031*
0.039*
0.026
0.008
0.031
(continued on next page)
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
145
TABLE 7 (continued)
GSID
PNE
Pseudo R2
Prob. (v2)
n
GSIG
0.010
0.083
0.069
0.001
836
0.112
0.001
864
*, **, *** Signify difference from 0 at the 10 percent, 5 percent, and 1 percent levels, respectively (two-tailed test).
This table examines the portfolio characteristics of governance-sensitive institutions. Panel A presents mean comparisons
of institutions’ portfolio characteristics based on their governance sensitivity. We create seven factors to measure
institutional portfolio characteristics: ISIZE, PTURN, BLOCK, FSIZE, PRUDENCE, VALUE, and RISK. Panel A
presents means for each of the factors and the individual measures that comprise each factor. Panel B presents marginal
effects at sample means for multivariate comparisons of portfolio characteristics estimated using logistic regression. The
dependent variables are indicators for whether the institution is classified as GSID or GSIG.
Variable Definitions:
ISIZE ¼ the institution’s size and is composed of the logarithm of the number of stocks in the portfolio (NSTK) and the
market capitalization of the portfolio (TE);
PTURN ¼ the duration that an institution holds an investment. The following items comprise PTURN: portfolio turnover
in terms of market capitalization (PT1), portfolio turnover in terms of sales transactions (PT2), percent of number of
firms held in portfolio for at least two years (STAB1), and percent of total holdings held for at least two years
(STAB2);
BLOCK ¼ the extent to which an institution is a blockholder. It consists of the percent of total holdings with at least a 5
percent stake (LBPH), the percent of portfolio firms in which it has at least a 5 percent stake (LBPN), and the
average percent ownership in portfolio firms (WAPH);
FSIZE ¼ the typical size of firms in the institution’s portfolio. It consists of the following measures: the weighted-average
market capitalization of portfolio firms (WAMC), the weighted-average of whether firms are members of the S&P
500 Index (WASP), the logarithm of the weighted-average number of months that portfolio firms have been
publicly listed (WATIME), and the weighted-average price per share of portfolio firms (WAEPRC);
PRUDENCE ¼ the extent to which the institution invests in prudent stocks as dictated by fiduciary responsibilities: the
percent of firms in the portfolio with five consecutive years of earnings growth (WADUP), the weighted-average of
a positive earnings indicator variable (WAPED), the weighted-average S&P stock rating (WARATE), and the
weighted-average debt to equity ratio of portfolio firms (WADE);
VALUE ¼ the extent to which the institution follows a value investing strategy: the weighted-average earnings to price
ratio (WAEP), the weighted-average dividend to price ratio (WADP), and the weighted-average book to price ratio
(WABP); and
RISK ¼ the riskiness of portfolio firms: the weighted-average earnings growth (WAEGR), the weighted-average sales
growth (WASGR), the weighted-average beta (WABTA), and the weighted-average standard deviation of stock
returns (WASTD).
is a combination of the number of stocks in the portfolio and the market capitalization of the
portfolio. Institutions that hold a large number of stocks in their portfolio may view investing in
firms with ‘‘better’’ corporate governance mechanisms as a means to reduce monitoring costs. In
contrast, large institutions could enjoy economies of scale in monitoring activities and better access
to management that would reduce reliance on internal governance mechanisms. We use the
portfolio turnover factor, PTURN, to measure the investment horizon of the institution. Institutions
with short investment horizons are less exposed to potential governance failures and are likely to be
less sensitive to governance mechanisms, whereas buy-and-hold institutions, especially those
following an index strategy, have a longer-term exposure to governance failures. We create a
blockholder factor, BLOCK, to measure the extent to which an institution holds large positions in
portfolio firms. Holding large positions in portfolio firms makes it more costly to liquidate the
positions rapidly and increases the potential cost of governance failures, which could lead to
governance sensitivity.
The remaining four factors measure investment styles (i.e., preferences for certain firm
characteristics) that are apparent in the institution’s portfolio holdings (Abarbanell et al. 2003). The
Journal of Management Accounting Research
Volume 26, Number 2, 2014
146
Bushee, Carter, and Gerakos
factor FSIZE measures the typical market capitalization of firms in the institution’s portfolio. Firm
size preferences could be associated with governance sensitivity for two reasons. First, smaller
firms have a higher incidence of fraud (Bushee and Leuz 2005). Second, larger firms have richer
public information environments and greater external monitoring by analysts and the media. Thus,
institutions following small-cap styles have incentives to tilt toward firms with ‘‘better’’ governance
mechanisms. We use the factor PRUDENCE to measure the extent to which the institution invests
in firms with characteristics that would support the ex ante prudence of the investment. As
mentioned earlier, firms with the most stringent fiduciary responsibilities have incentives to invest
in ‘‘better’’ governed firms to reduce the probability of holding a firm that experiences a governance
failure. We measure preferences for ‘‘growth’’ or ‘‘value’’ firms with the VALUE factor. It is unclear
whether growth or value firms would be more likely to experience governance failures; thus, we do
not have a prediction for how this factor would affect governance sensitivity. Finally, the factor
RISK captures the risk of portfolio firms. Institutions that invest in riskier firms may prefer ‘‘better’’
governance because high levels of risk may make it more costly for the institution to monitor the
firm’s activities.
Panel A of Table 7 compares the means of portfolio characteristics for governance-sensitive
and -insensitive institutions. We measure portfolio characteristics as of fiscal year-end 1997. In
general, governance-sensitive institutions are significantly larger, both in terms of market
capitalization and the number of individual stocks that they hold in their portfolio. In addition,
they tend to have longer investment horizons than governance-insensitive institutions. Institutions
that are sensitive to board characteristics are less likely to hold block positions in portfolio firms and
more likely to follow a growth investment strategy. Governance-sensitive institutions of both types
tend to invest in larger firms and to invest in firms that meet prudence standards.
Panel B presents estimated marginal effects at sample means for logistic regressions in which
the dependent variables are coded as 1 if an institution is classified as GSID (GSIG), and 0 if
classified as GIND (GING). The independent variables are the seven factors that measure portfolio
characteristics and indicator variables for whether an institution is either BNK or PNE:18
GOVSENi ¼ a þ b1 ISIZEi þ b2 PTURNi þ b3 BLOCKi þ b4 FSIZEi þ b5 PRUDENCEi
þ b6 VALUEi þ b7 RISKi þ b8 BNKi þ b9 PNEi þ ei ;
where GOVSEN ¼ 1 (0) if the institution is classified as GSID (GIND) or GSIG (GING).
The p-values for both regressions are less than 0.01, indicating that the independent variables
provide explanatory power. The multivariate results are similar to the univariate results presented in
Panel A. The marginal effects for ISIZE are positive and significant for both GSID and GSIG,
indicating that larger institutions are more likely to be sensitive to both governance mechanisms.
For GSID, the marginal effect for VALUE is negative and significant, indicating that institutions that
follow a growth investment strategy are more likely to be sensitive to board characteristics.
Institutions that invest in growth stocks may prefer ‘‘better’’ governance because they believe that
firms with higher growth opportunities require higher-quality internal oversight. For GSIG, PTURN
and FSIZE are negative and significant, implying that institutions that hold their positions for longer
periods and that invest in smaller firms are more likely to be sensitive to shareholder rights.
Consistent with the univariate test, the marginal effect for PRUDENCE is positive and significant,
implying that institutions that follow an investment strategy concordant with fiduciary duties are
more likely to be sensitive to shareholder rights. In contrast with the results presented in Table 6,
pensions and endowments are not more likely to be sensitive to shareholder rights. Finally, note that
18
Results are qualitatively similar in terms of sign and significance if we combine institutions that are not classified
with the GIND and GING institutions.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
147
blockholder ownership by institutional investors is not significantly related to governance
sensitivity for either governance mechanism, suggesting that block ownership serves as a substitute
for governance mechanisms, rather than a complement.
Although prior research commonly uses legal type to proxy for governance sensitivity, we find
that governance sensitivity is more associated with the characteristics of an institution’s portfolio
than its legal type. In general, larger institutions are more likely to tilt their portfolio toward firms
with ‘‘better’’ governance, suggesting that governance mechanisms decrease monitoring costs.
Other than large institutions, only institutions with preferences for growth firms tilt their portfolios
with ‘‘better’’ board characteristics, implying that board governance is viewed by institutions as
more essential for firms with a high level of growth opportunities. In contrast, institutions with long
investment horizons and small-cap investment styles are more likely to tilt their portfolios toward
firms with more shareholder rights, suggesting that governance allows these institutions to protect
their investments.
CONCLUSION
We examine institutional investors’ preferences for firm-level corporate governance
mechanisms within the categories of board of director characteristics and shareholder rights.
Specifically, we investigate three questions: (1) To what extent is corporate governance a
determinant of institutions’ investment and trading decisions? (2) To what extent do institutions
actively implement preferred governance mechanisms in their portfolio firms, as opposed to simply
investing in firms with preferred mechanisms? (3) Which types of institutions display preferences
for corporate governance mechanisms?
Using data from 1995 to 1997, we find that approximately 10 percent of institutions are
sensitive to each set of governance mechanisms; governance characteristics significantly affect their
portfolio weighting decisions. We find strong evidence that changes in ownership by
governance-sensitive institutions are associated with prior levels of, and contemporaneous changes
in, governance mechanisms. Despite the evidence that governance-sensitive institutions prefer to
invest in firms with existing preferred governance mechanisms, firms with a high level of
institutional ownership sensitive to shareholder rights exhibit significant future improvements in
shareholder rights, implying some activism by these institutions.
Finally, we investigate the characteristics of institutions that are governance-sensitive.
Consistent with their fiduciary responsibilities, bank trusts and pensions and endowments tend to
have the highest percentage of governance-sensitive institutions. But neither type has more than 25
percent of its institutions classified as governance-sensitive, indicating that the legal type
classifications are poor proxies for general governance sensitivity. Therefore, we examine the
association between governance sensitivity and a set of factors that describe the characteristics of
institutions’ portfolios. We find that large institutions and institutions holding a large number of
portfolio stocks are more likely to be sensitive to corporate governance mechanisms, suggesting
that they view governance mechanisms as means to decrease monitoring costs. In addition,
institutions with preferences for growth firms tilt their portfolios toward firms with ‘‘better’’ board
characteristics, implying that institutions view board governance as more essential for firms with
higher growth opportunities. In contrast, institutions with long investment horizons and small-cap
investment styles are more likely to tilt their portfolios toward firms with ‘‘better’’ shareholder
rights, suggesting that governance allows these institutions to protect their investments.
Interestingly, blockholder ownership by institutional investors is not significantly related to
governance sensitivity, suggesting that block ownership serves as a substitute, rather than
complement, for governance mechanisms.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Bushee, Carter, and Gerakos
148
Our findings have implications for investors and researchers. For investors, understanding
institutional investors’ preferences for governance mechanisms becomes increasingly important as
regulators and legislators continue to consider increased shareholder access as a means to
improving corporate governance (Davidoff 2009; Holtzer 2011). For researchers, our results
suggest that common proxies for governance sensitivity by investors (e.g., legal type, blockholding)
do not fully capture important aspects of the motivation for governance sensitivity and, as a result,
may misclassify institutions with respect to their governance sensitivity.
REFERENCES
Abarbanell, J., B. Bushee, and J. Raedy. 2003. Institutional investor preferences and price pressure: The
case of corporate spin-offs. Journal of Business 76: 233–261.
Admati, A., and P. Pfleiderer. 2009. The Wall Street walk as a form of shareholder activism. Review of
Financial Studies 22: 2645–2685.
Aggarwal, R., P. Saffi, and J. Sturgess. 2012. The Role of Institutional Investors in Voting: Evidence from
the Securities Lending Market. Working paper, Georgetown University.
Agrawal, A., and C. Knoeber. 1996. Firm performance and mechanisms to control agency problems
between managers and shareholders. Journal of Financial and Quantitative Analysis 31: 377–397.
Almazan, A., J. Hartzell, and L. Starks. 2005. Active institutional shareholders and costs of monitoring:
Evidence from executive compensation. Financial Management 34: 5–34.
Barber, B. 2006. Monitoring the Monitor: Evaluating CalPERS’ Activism. Working paper, University of
California, Davis.
Bhide, A. 1993. The hidden costs of stock market liquidity. Journal of Financial Economics 34: 31–51.
Borokhovich, K., K. Brunarski, Y. Harman, and R. Parrino. 2006. Variation in the monitoring incentives of
outside stockholders. Journal of Law and Economics 49: 651–698.
Brickley, J., R. Lease, and C. Smith. 1988. Ownership structure and voting on antitakeover amendments.
Journal of Financial Economics 20: 267–291.
Brown, L., and M. Caylor. 2006. Corporate governance and firm valuation. Journal of Accounting and
Public Policy 25: 409–434.10.1016/j.jaccpubpol.2006.05.005
Bushee, B. 2001. Do institutional investors prefer near-term earnings over long-run value? Contemporary
Accounting Research 18: 207–246.
Bushee, B., and C. Leuz. 2005. Economic consequences of SEC disclosure regulation: Evidence from the
OTC bulletin board. Journal of Accounting and Economics 39: 233–264.
Bushee, B., and C. Noe. 2000. Corporate disclosure practices, institutional investors, and stock return
volatility. Journal of Accounting Research 38: 171–202.
Byrd, J., and K. Hickman. 1992. Do outside directors monitor managers? Evidence from tender offer bids.
Journal of Financial Economics 32: 195–221.
Carleton, W., J. Nelson, and M. Weisbach. 1998. The influence of institutions on corporate governance
through private negotiations: Evidence from TIAA-CREF. Journal of Finance 53: 1335–1362.
Chen, X., J. Harford, and K. Li. 2007. Monitoring: Which institutions matter? Journal of Financial
Economics 86: 279–305.
Dalton, D., C. Daily, J. Johnson, and A. Ellstrand. 1999. Number of directors and financial performance: A
meta-analysis. Academy of Management Journal 42: 674–686.
Davidoff, S. 2009. Strangling corporate governance to save it. New York Times (May 22).
Del Guercio, D. 1996. The distorting effect of the prudent-man laws on institutional equity investments.
Journal of Financial Economics 40: 31–62.
Dikolli, S., S. Kulp, and K. Sedatole. 2009. Transient institutional ownership and CEO contracting. The
Accounting Review 84: 737–770.
Edmans, A. 2009. Blockholders, market efficiency, and managerial myopia. Journal of Finance 64: 2481–
2513.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
Institutional Investor Preferences for Corporate Governance Mechanisms
149
Ferreira, M., and P. Matos. 2008. The colors of investors’ money: Which firms attract institutional investors
from around the world? Journal of Financial Economics 88: 489–533.
Gillan, S., and L. Starks. 2000. Corporate governance proposals and shareholder activism: The role of
institutional investors. Journal of Financial Economics 57: 275–305.
Gillan, S., and L. Starks. 2003. Corporate governance, corporate ownership, and the role of institutional
investors: A global perspective. Journal of Applied Finance 13: 4–22.
Gompers, P., and A. Metrick. 2001. Institutional investors and equity prices. Quarterly Journal of
Economics 116: 229–259.
Gompers, P., J. Ishii, and A. Metrick. 2003. Corporate governance and equity prices. Quarterly Journal of
Economics 118: 107–156.
Hallock, K. 1997. Reciprocally interlocking boards of directors and executive compensation. Journal of
Financial and Quantitative Analysis 32: 331–344.
Hartzell, J., and L. Starks. 2003. Institutional investors and executive compensation. Journal of Finance 58:
2351–2374.
Hermalin, B., and M. Weisbach. 2003. Boards of directors as an endogenously determined institution: A
survey of the economic literature. Economic Policy Review Federal Reserve Bank of New York 9: 7–
26.
Holtzer, J. 2011. Corporate news: Court deals blow to SEC, activists. The Wall Street Journal (July 23).
Larcker, D., S. Richardson, and A. Tuna. 2007. Corporate governance, accounting outcomes, and
organizational performance. The Accounting Review 82: 963–1008.
Leuz, C., K. Lins, and F. Warnock. 2009. Do foreigners invest less in poorly governed firms? Review of
Financial Studies 22: 3245–3285.
Li, K., H. Ortiz-Molina, and X. Zhao. 2008. Do voting rights affect institutional investment decisions?
Evidence from dual-class firms. Financial Management 37: 713–745.
McCahery, J., Z. Sautner, and L. Starks. 2010. Behind the Scenes: The Corporate Governance Preferences
of Institutional Investors. Working paper, The University of Texas.
Parrino, R., R. Sias, and L. Starks. 2003. Voting with their feet: Institutional ownership changes around
forced CEO turnover. Journal of Financial Economics 68: 3–46.
Rogers, W. 1993. Regression standard errors in clustered samples. Stata Technical Bulletin 13: 19–23.
Rosenstein, S., and J. Wyatt. 1990. Outside directors, board independence, and shareholder wealth. Journal
of Financial Economics 26: 175–191.
Shapiro, M. 2011. Statement by SEC Chairman Mary L. Schapiro on Proxy Access Litigation. Available at:
http://www.sec.gov/news/press/2011/2011-179.htm
Smith, M. 1996. Shareholder activism by institutional investors: Evidence from CalPERS. Journal of
Finance 51: 227–252.
Strickland, D., K. Wiles, and M. Zenner. 1996. A requiem for the USA: Is small shareholder monitoring
effective? Journal of Financial Economics 40: 319–338.
U.S. House of Representatives. 2010. Dodd-Frank Wall Street Reform and Consumer Protection Act of
2010. Public Law 111-203 [H.R. 4173]. Washington, DC: Government Printing Office.
Wahal, S. 1996. Pension fund activism and firm performance. Journal of Financial and Quantitative
Analysis 31: 1–23.
Yermack, D. 1996. Higher market valuation of companies with a small board of directors. Journal of
Financial Economics 40: 185–211.
Journal of Management Accounting Research
Volume 26, Number 2, 2014
THE ACCOUNTING REVIEW
Vol. 94, No. 2
March 2019
pp. 357–361
American Accounting Association
DOI: 10.2308/accr-10651
PRESIDENTIAL SCHOLAR
Accounting Information in Corporate Governance:
Implications for Standard Setting
S. P. Kothari
Massachusetts Institute of Technology
ABSTRACT: Accounting standards are crucially relevant in the context of the use of accounting information in
corporate governance. Notwithstanding highly liquid capital markets, large and small shareholders, many activist
shareholders, sophisticated analysts, vigilant press reporters, and a vibrant litigious environment, corporate
governance challenges continue to make media headlines, and they seem to occur with a great degree of regularity.
The essay offers a high-level description of the objectives of accounting standards, a quick run through the evolution
of accounting research over the past half-century, and, finally, offers three examples of standards and disclosure
requirements that might be worthwhile to reexamine in light of the governance role of accounting information.
I. INTRODUCTION
A
s I reflect on more than three decades in the profession as an educator and a researcher, and also on my activities in
quasi-academic roles as a manager, dean, and consultant, the unmistakable conclusion I draw is th…