Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Compensation Peer Choice and Managerial Capital1
Jeffrey Coles
University Of Utah
Fangfang Du
Arizona State University
Daruo Xie
Australian National University
This Version: August 2017
(Do Not Distribute)
Abstract
Conventional wisdom regards a CEO’s compensation peer group as a group of peer firms. We
argue that a CEO’s compensation peer is another CEO, not a firm, and a compensation peer group
is a group of other CEOs, not a group of firms. A firm need to select CEOs with similar
marketable managerial capital, rather than similar firms, to construct a peer group. We then
examine the role of managerial capital in actual peer selections. We show that in practice firms
select compensation peers largely based on the similarity in CEO managerial capital. However,
we also find evidence that actual compensation peers are biasedly selected to exaggerate labor
market competition for CEO managerial capital.
1 We thank Carr Bettis and Incentive Lab, a compensation and governance company in Scottsdale, Arizona, for the
use of the pertinent elements of the Incentive Lab Compensation and Metrics Data.
1. Introduction
Becker (1962), and Murphy and Zabojnik (2004, 2007), distinguish between general
managerial capital (managerial skills transferable across firms, not specific to any organization,
hence marketable) from firm-specific managerial capital (managerial skills valuable only within
the organization, not transferable). This is important because a competitive labor market for CEO
talents would not exist if managerial skills were not transferable across firms. In the CEO labor
market, potential employers compete for a CEO’s marketable managerial capital, not for a CEO’s
firm-specific managerial capital. It is a CEO’s marketable managerial capital that determines the
demand and supply condition for his/her talent. Each time a firm hires a new CEO, a competitive
pay needs to reflect the market price of the CEO’s managerial capital in the labor market.
In practice many firms use a peer group as a benchmark to determine the CEO’s
compensation.2 Prior studies on peer selection think of a compensation peer group as a group
of “peer firms”, constructed based on the similarity in firm characteristics.3 We propose a simple
and new criterion on peer selection. We argue that a CEO’s peer is another CEO, not a firm, and
a compensation peer group needs to be a group of other CEOs, not a group of other firms. A
peer group need to be constructed to reflect the labor market competition for a CEO’s talent.
Hence, we propose that a firm needs to select CEOs with similar marketable managerial capital,
rather than similar firms, to construct a compensation peer group. This is because it is a CEO’s
marketable managerial capital that her/his potential employers compete for, and a competitive
labor market should generate similar prices for similar goods.
Our new criterion on peer selection matters at least in three aspects. First, our new
criterion is more consistent with the economic rationale. In theory a peer group can function as
an efficient way to determine the competitive pay to attract and retain valuable managerial capital.
Our view explicitly highlights the central role of CEO managerial capital in peer selection.
Secondly, our view provides a new practical guidance on peer selection. A group of peers selected
based on CEO managerial capital can be considerably different from a group of “peer firms” that
are selected solely based on firm characteristics. Thirdly, our new criterion implies a new
2 From our data 2006-2011, around 68% of the S&P1500 firms explicitly report that they use compensation peers
to set their CEOs’ compensations. 3 For instance, Bizjak, Lemmon, and Nguyen (2008, 2011), Faulkender and Yang (2010, 2013), Cadman and Carter
(2013).
definition of peer selection bias. Namely, if a peer group is constructed in any way to
misrepresent the labor market competition for CEO managerial capital, then the peer selection is
biased. In contrast, prior studies think of biased peer selection as mismatches in firm
characteristics.
In order to facilitate empirical tests, we first consider a broad set of observable proxy
variables that may reflect CEOs’ marketable managerial capital. Specifically, our proxy variables
belong to two main categories: the characteristics of a CEO’s current employer firm, and the
characteristics of a CEO’s lifetime work and education experiences. Such choice of two-category
variables are based on two underlying assumptions. The first assumption is that a CEO’s
managerial capital tends to be more marketable to similar firms, e.g. firms of similar size, in same
industry, etc. Hence, the characteristics of a CEO’s current employer firm can well reflect a
CEO’s marketable managerial capital. The second assumption is that a CEO’s lifetime education
and work experiences also considerably inform how the CEO has developed his/her managerial
skills. And this information is largely orthogonal to the characteristics of a CEO’s current firm.
Hence, the characteristics of a CEO’s lifetime experiences are indispensable indicators of the
CEO’s marketable managerial capital. For example, a CEO can be labeled as a generalist, if she
had worked in multiple industries, for multiple firms, and in multiple positions, etc. 4 A
generalist’s managerial capital is featured as being highly marketable. We hence follow Custodio,
Ferreira, and Matos (2013) to create the one-dimensional index of general managerial ability.
The General Ability Index will enable us to label each CEO as either a specialist or a generalist.
Following our argument that marketable managerial capital needs to play a central role
in peer selection, we lay out our testable hypothesis that the proxy variables reflecting marketable
managerial capital would be determinants of actual peer choice. We test this hypothesis by doing
multivariate analysis. The analysis shows that vast majority of the proxy variables, including
both the characteristics of a CEO’s current firm and the characteristics of a CEO’s lifetime
experiences, are statistically significant in determining actual peer choice. For instance, a firm
tends to select peer CEOs from firms with similar size, similar performance, similar market-to-
4 Note that the marketability of managerial skills is a dimension different from managerial productivity. According
to Murphy and Zabojnik (2007), a higher degree of marketability does not necessarily mean a higher productivity.
Custodio, Ferreira, and Matos (2013) empirically show a statistically insignificant relation between firm performance
and the degree of marketability.
book ratios, and in same industry, etc. A firm also tends to select peer CEOs at similar ages, with
similar multi-firm experiences, similar multi-position experiences, and similar educational
backgrounds. A CEO who worked in a conglomerate firm is more likely to have CEOs with
conglomerate experience chosen as peers. A CEO who holds the chairman position is more likely
to have chairman-CEOs chosen as peers. The results show that compensation peer selection is
not only about matches in firm characteristics, but also about matches in any variables that can
reflect CEO’s marketable managerial capital.
We then go further to address the question whether actual compensation peers are
opportunistically selected? This question captures attention because of the concern of possible
management-captured boards and inflated CEO compensations. We provide empirical evidences
that actual peer groups are biasedly constructed to misrepresent labor market competition. There
are two types of biases. First, our analysis shows that firms systematically choose peer CEOs
whose managerial capital is more marketable. For example, we find that firms are more likely to
choose generalist CEOs as compensation peers. All else equal, a specialist are 8.7% less likely to
be chosen as a compensation peer for a generalist, but a generalist are 3.5% more likely to be
chosen as a compensation peer for a specialist. The asymmetry is statistically and economically
significant. By favoring generalist CEOs, a board exaggerates the labor market competition for
its CEO’s talent and inflate CEO compensation upward.5 Second, our analysis also shows that
firms are more likely to choose CEOs from larger and better performing firms as compensation
peers. CEOs from larger and better performing firms are likely to be more productive. By
favoring more productive CEOs, a board again exaggerates the labor market competition for its
CEO’s managerial capital.
Bizjak, Lemmon, and Nguyen (2011), Cadman and Carter (2013), and Faulkender and
Yang (2010, 2013), find that firms tend to favor larger and better performing firms when selecting
compensation peers. Albuquerque, De Franco, and Verdi (2013) argue that the choice of highly
paid peers represents a reward for unobserved CEO talent. Empirically our tests partially overlap
prior studies as regards to the relation between firm characteristics and peer selection. However,
our paper suggests a new framework for examining compensation peer groups. First, we provide
5 Custodio, Ferreira, and Matos (2012) empirically show that generalist CEOs earned significant higher
compensations than specialist CEOs.
new answers to two basic questions: what is the criterion for selecting compensation peers? and
how to identify possible selection biases? Furthermore, we empirically examines actual peer
selection and find new evidences on peer selection biases. Our paper provides a deeper and
broader framework that nests prior studies.
Frydman (2005), and Murphy and Zabojnik (2004, 2007) suggest that an increasing
importance of marketable managerial capital leads to dramatic increase in equilibrium CEO
wages.6 Our paper suggests that an increasing importance of marketable managerial capital,
however, could induce or exacerbate the problem of biased peer selection. As marketable
managerial capital becomes increasingly important over time, it facilitate a firm to misrepresent
the labor market competition for its CEO’s managerial capital by preferring generalists as peers.
Over time, the systematic selection biases could lead to a ratcheting of CEO compensations, and
considerably inflate CEO compensation in the long run.
Our paper also contributes to the literature on the connections between CEOs’ lifetime
experiences and corporate polices. Several studies examines how the life experience may shape
CEOs’ risk preferences (Bernile, Bhagwat, and Rau (2014), Custodio and Metzger (2014),
Dittmar and Duchin (2014)). In contrast, our paper focuses on CEOs’ human capital accumulation
through their lifetime experiences, and its implications to CEO compensation design.
The remainder of this article is divided into several parts. The section 2 explicates our
new criterion for selecting compensation peers. Section 3 describes the data and variables. In
section 4, we provide evidences that general managerial capital affects CEO compensation, and
we empirically examine determinates of peer selection. Section 5 concludes.
2. Peer group benchmarking
2.1 Peer selection and managerial human capital
In a competitive CEO labor market, a CEO’s compensation is the market price of his/her
managerial human capital. Theories such as Becker (1962), and Murphy and Zabojnik (2004,
6 Some other theories also suggest that the dramatic CEO pay increase can be explained by economic fundamentals.
For example, Gabaix and Landier (2008) develop a simple equilibrium model of CEO pay and they suggest that the
recent rise in CEO pay can be attributed to substantial growth in firm size. Marin and Verdier (2004) argue that an
increase in international trade has led to the increase in demand for managerial talent, which in turn push up the
CEO pay.
2007), suggest that managerial human capital has two components, general managerial capital
(managerial skills transferable across firms, not specific to any organization, hence marketable)
and firm-specific managerial capital (managerial skills valuable only within the organization,
hence not marketable). This is important because the existence of marketable managerial capital
is a precondition for the presence of a competitive CEO labor market. In the CEO labor market,
potential employers compete for a CEO’s marketable managerial capital, not for his/her firm-
specific managerial capital. Hence it is a CEO’s marketable managerial capital that should
determine the demand and supply condition for his/her talent. In contrast, firm-specific
managerial capital is not marketable, so it should have little effect on the market price of a CEO’s
managerial talent.
In practice many firms use compensation peer groups as benchmarks to determine their
CEOs’ compensation. For example, from 2006 to 2011, around 68% of the S&P1500 firms, 92%
of the S&P500 firms, explicitly report that they use compensation peers to set their CEOs’
compensations. The existence of a competitive CEO labor market provides the rationale for using
peers group to set CEO pay. The rationale is that peer group benchmarking can be an efficient
way to determine the amount of pay necessary to attract and retain valuable managerial capital.
By selecting a group of other CEOs as compensation peers, a board is able to set a competitive
pay level for its CEO. In contrast, if a board is blind to the pays of other CEOs, it would be
difficult to imagine how the board can set a competitive pay for its CEO.
In the light of the economic rationale for using compensation peers, we propose a simple
and novel view on the selection of compensation peers. Prior studies on peer selection, such as
Bizjak, Lemmon, and Nguyen (2011) (BLN here after), Faulkender and Yang (2010, 2013) (FY
here after), and Cadman and Carter (2013) (CC here after), regard compensation peer group as a
group of “peer firms”, focusing on whether a board selects the “right” firms. We argue that a
CEO’s peer is another CEO, not a firm. A compensation peer group should be a group of other
CEOs, not a group of firms. A compensation peer group need to be constructed in a way to reflect
the labor market competition for the CEO’s managerial capital. Hence, we propose that a board
should select other CEOs with similar marketable managerial capital, rather than similar firms,
to construct a compensation peer group. This new view on peer selection bases on a simple
economic principle that a competitive market should generate similar prices for similar goods. In
the competitive managerial labor market, CEOs’ marketable managerial capital are the “goods”.
Our view on peer selection provides a new guidance on the selection of compensation
peers. A group of peers selected based on managerial capital can be considerably different from
a group of “peer firms”, which are selected solely based on firm characteristics. For instance, the
potential employers of a generalist CEO would be considerably different from the potential
employers of a specialist CEO. Therefore, a generalist CEO should have a different peer group
from a specialist CEO’s peer group, regardless of how similar their current firms are. Because
prior studies regard compensation peers as a group of “peer firms”, they only provide incomplete
answers to the question whether actual peers are selected in a way to reflect labor market
competition.
Practically, we need a set of observable proxy variables as indicators of CEOs’ marketable
managerial capital. Accordingly, we make two additional assumptions. Our first assumption is
that a CEO’s managerial capital tends to be more marketable to other firms which are similar to
her/his current employer firm. For example, a CEO who works for a large firm in the
pharmaceutical industry would likely to have other large pharmaceutical firms as potential
employers. Hence, this assumption implies that the characteristics of a CEO’s current firm, e.g.
firm size, firm industry, firm performance, are all important indicators of her/his marketable
managerial capital. Our second assumption is that a CEO’s lifetime education and work
experiences considerably inform her/his marketable managerial capital. For example, if a CEO
worked as a top manager in several other firms and in several different industries, he/she would
likely to be classified as a generalist CEO, and the generalist CEO’s managerial capital would
likely to be highly marketable. So this assumption implies that the characteristics of a CEO’s
lifetime education and work experiences are indispensable indicators of her/his marketable
managerial capital. As the result, our proxy variables belong to two main categories: the
characteristics of a CEO’s current employer firm, and the characteristics of a CEO’s lifetime
education and work experiences. Our hypothesis is that these variables play important roles in
determining the choice of compensation peers. We will empirically test the hypothesis and report
the results in Section 4.1.
2.2 Opportunistic peer selection
A main concern in peer studies is whether actual compensation peers are opportunistically
selected. Arguments based on captured boards suggest that top managers may influence the peer
selection process to inflate their pays. To address this concern, we first need to answer a notional
question, how to define a possible bias in peer selection? Prior studies generally define a bias in
peer selection as a mismatch in firm characteristics, e.g. a firm selects larger and better
performing firms as compensation peers. We propose a new definition of a possible bias in
compensation peer selection. If a peer group is constructed in any way to misrepresent the labor
market competition for a CEO’s managerial capital, then the peer selection is biased. Obviously,
this new definition is a direct implication from our view that a board should select CEOs with
similar marketable managerial capital to construct a peer group.
Accordingly, in theory there are two types of possible biases in constructing a
compensation peer group. A type-one bias is that a board systematically chooses peer CEOs
whose managerial capital is more marketable. For example, a specialist CEO may not have much
opportunities of outside hiring, but the board can exaggerate the marketability of her/his
managerial capital by selecting generalist CEOs as peers. A type-two bias is that a board
systematically chooses peer CEOs whose managerial capital is more productive.7 For example, a
small firm can exaggerate the productivity of its CEO’s managerial capital by selecting some
S&P500 firm CEOs as peers. In either way, the board misrepresents the labor market competition
for its CEO’s managerial capital, and inflates the CEO’s pay.
Empirically we identify peer selection biases by examining possible mismatches in the
variables that indicate CEOs’ marketable managerial capital. BLN (2011), CC (2013), and FY
(2010, 2013), find that firms tend to favor larger and better performing firms when selecting
compensation peers. Our paper doesn’t really overthrow their empirical results. Instead, our work
provides a deeper and broader framework that nests prior studies. The selection biases the prior
studies observed actually belong to the type-two biases, i.e. misrepresenting the productivity of
CEO. However, no study has ever examined the type-one biases, i.e. misrepresenting the
7 Note that the transferability of managerial skills is a dimension different from managerial productivity. According
to Murphy and Zabojnik (2007), a higher degree of transferability does not necessarily mean a higher productivity.
CFM (2012) empirically shows a statistically insignificant relation between firm performance and the degree of
transferability.
marketability of CEO. We will do a broader examination on opportunistic peer selection, which
covers both types of selection biases.
3. Data and variables
Our data on CEO compensation peer groups comes from the Incentive Lab Compensation
and Metrics Data. Incentive Lab collects compensation peer data from corporate proxy
statements. The sample data covers the Standard & Poor’s 1500 firms that reported their
compensation peers. US public firms have been required to disclose their compensation peers
every year since 2006. The sample consists of a panel of 6,989 firm-years and 156,903 peer-years
observations in the fiscal years 2006 through 2012.8 This data sample is much larger than the
data samples used in prior peer studies. FY (2010), BLN (2011), and Albuquerque, De Franco,
and Verdi (2013), have 657 firm-years, 707 firm-years, and 2,158 firm-years respectively. Table 1
presents descriptive statistics of our sample data. We find little evidences on any trends in the
usage of CEO compensation peers. The number of S&P 1500 firms that use compensation peers
increased from 878 firms in 2006 to 1061 firms in 2007, and then stayed largely at the same level
in the following years. In total, 68% of S&P1500 firms used compensation peers in setting CEO
pay. We don’t see any trends in the size of peer groups either. Each CEO had a median of 17
compensation peers. The size distribution of peer groups was largely unchanged over the years.
The sample data from the Incentive Lab contains only firm identifier information. We try
to find the corresponding CEOs for each firm and each reported peer. We either merge the raw
sample data with BoardEx database, or manually look into corporate proxy statements to identify
the corresponding CEOs. As the result, our test sample consists of a panel of 6,297 firm-years
which includes 1,674 unique CEOs, and 122,133 peer-years which include 3,987 unique peer
CEOs.
Table A1 provides a complete list of our proxy variables for CEOs’ marketable managerial
capital. We will use these variables in estimating peer choice model. The variables are either the
characteristics of a CEO’s employer firm, or the characteristics of a CEO’s lifetime experiences.
Firm characteristic variables are collected from CRSP and COMPUSTAT databases. CEO
8 Our sample does not contain all the firms reporting compensation peers in the fiscal year 2012. Our data is obtained
on summer 2013, at a time when some firms had not reported their peers for the fiscal year 2012.
experience variables are collected from BoardEx database. We largely follow BLN (2011) in
choosing our firm characteristic variables, such as size, industry, credit rating and market-to-
book ratio. In what follows, we briefly discuss a number of CEO experience variables.
Number of positions a CEO worked: A CEO who worked in many different positions is likely
to have acquired knowledge in multiple organizational areas. The CEO’s managerial skills are
likely to be more transferable.
Number of firms a CEO worked for: A CEO who worked for multiple firms is likely to have
better developed managerial skills transferable across firms. The CEO is likely to have greater
opportunities on the labor market.
Number of industries in which a CEO worked: A CEO who worked for multiple industries
is likely to have been exposed to different industrial environments. The CEO’s managerial skills
are likely to be broadly demanded in the labor market.
Experience as a CEO of another firm: A CEO who worked at a top managerial position of
another firm is likely to have developed managerial skills are not specific to any given
organization. These skills are applicable across firms.
Conglomerate experience: A CEO who worked for a multi-division firm is likely to have been
exposed to complex business environment. This experience may help the CEO to accumulate
her/his general managerial skills.
MBA degree: A CEO who earned a MBA degree is likely to have learned a body of
knowledge in multiple disciplines, such as economics, management science, finance, and
accounting. The knowledge is not firm-specific, and MBA education is likely to support a CEO
in developing her/his general managerial skills.
CEO-Chair: A CEO who also holds chairman position is likely to have a deep
understanding of the set of skills to manage a modern corporation, which are applicable across
firms.
CEO Age: An elder CEO is likely to be more experienced and better developed her/his
general managerial skills.
An alternative way to measure CEOs’ marketable managerial capital is to construct the
General Ability Index (GAI) suggested by CFM (2012). The GAI is a one-dimensional index
based on five variables that proxy for CEOs’ general managerial capital. The five variables are
Number of firms a CEO worked for(X1), Number of Positions a CEO worked(X2), Number of
Industries in which a CEO worked(X3), CEO Experience Dummy(X4), and Conglomerate Experience
Dummy(X5). Using principal component analysis, we can extract common components from five
variables, then create GAI. Specifically, the GAI index for CEO i in year t is calculated by applying
the scoring coefficients to the five standardized components.
𝐺𝐴𝐼𝑖,𝑡 = 0.319𝑋1𝑖,𝑡 + 0.290𝑋2𝑖,𝑡 + 0.301𝑋3𝑖,𝑡 + 0.183𝑋4𝑖,𝑡 + 0.216𝑋5𝑖,𝑡
The GAI index gives relative higher weight to Number of firms a CEO worked for (X1) and
relative lower weight to CEO Experience Dummy(X4). A higher value of GAI suggests a higher
level of transferability of managerial capital.
Table 2 presents the summary statistics of our variables and the GAI index. We compute
the statistics both for our sample-firm CEOs (6,297 observations) and for all the observations of
selected peer CEOs (122,133 observations). The numbers shows that on average sample-firm
CEOs are largely similar to the selected peer CEOs in both CEO characteristics and firm
characteristics. This similarity provides preliminary evidences that actual compensation peers
are selected based on CEOs’ marketable managerial capital. However, the table 2 also shows that
on average the peer CEOs are more transferable and working for larger and better performing
firms. For example, the number of positions a CEO worked (10.3), the number of firms a CEO
worked for (3.2), and the number of industries a CEO worked (2.2) of the firm CEOs are all
better than those of the sample CEOs, which are 9.6, 3.0 and 2.1 respectively. The average GAI
index of peer CEOs (0.344) is higher than the average GAI of the firm CEOs (0.264). In addition,
the average size of peer-CEO firms is larger than the size of sample CEO firms ($13.7 billion vs.
$10.9 billion by Sale). The average performance of peer-CEO firms is better than the
performance of sample CEO firms (4.32% vs. 3.98%, by return on assets). These systematic
differences raise concerns that actual peers are biasedly selected. In next section, we will further
look into the issue of biased peer selection by estimating a peer choice model.
4. Peer Selection
4.1. General Managerial Ability and CEO Pay
The positive relation between general managerial ability and CEO pay has been first
documented in CFM (2013). According to CFM (2013), generalist CEOs earn a wage premium
of 19% more than specialist CEOs. Our data sample, a panel of CEOs in the 2006-2012 period,
is largely different from CFM’s (2013) sample, which consists of a panel of CEOs in the 1993-
2007 period. To ensure that we can refer to the main finding of CFM (2013), we replicate their
main test with our sample.
Table 3 reports our replication of CFM’s (2013) main finding of a strong positive relation
between CEO pay and the General Ability Index (Table 5 in CFM (2013)) In the panel
regressions, the dependent variable is the logarithm of total pay of sample CEOs and peer CEOs
in the 2006–2012 periods. Independent variables include the measures of firm characteristics and
CEO characteristics, as well as the constructed General Ability Index for each CEO-year. We
include year and industry (two-digit SIC codes) fixed-effects to control for any variation in CEO
pay across industries and over time. The robust t-statistics are adjusted for firm-level clustering.
Consistent with CFM (2013), CEOs with higher general managerial ability earn a wage
premium. In column 3 of table 4, CEOs who are one standard deviation higher in the General
Ability Index distribution earn 10.0% higher in annual pay, compared with 11.7% as reported in
CFM (2013) (Table 5 Column 3). We also construct a General Ability Index dummy variable that
takes a value of one with a CEO’s General Ability Index above the yearly median and zero
otherwise. In column 5 of table 3, a generalist CEO earns about 17.9% more than a specialist
CEO, compared with 18.6% as reported in CFM (2013) (Table 5 Column 6). In the summary,
despite using a different test sample, our estimates are quantitatively close to those reported in
CFM (2013). The regressions reconfirm that CEOs with higher general managerial ability
receive higher pay. Moreover, the estimates will help us to assess the economic significance of
peer selection biases that will be reported in the following sections 4.2 and 4.3.
4.2. Managerial Human Capital and Peer Group Selection
In our main test, we estimate multivariate logit regressions to identify the variables that
drive the choice of peers. Specifically, we use the method of multivariate logit regressions
developed by BLN (2011).
,
N
ij n n ij ijPeer Selection S
where the dependent variable 𝑃𝑒𝑒𝑟 𝑆𝑒𝑙𝑒𝑐𝑡𝑖𝑜𝑛𝑖𝑗 takes the value of one if CEO 𝑗 is
chosen as a peer of CEO 𝑖 and zero otherwise. The pool of potential peer CEOs in each year
includes all sample CEOs and all disclosed peer CEOs in that year. The independent variable
𝑆𝑛,𝑖𝑗 measures the difference between CEO i and CEO j in the value of variable n, and 𝑆𝑛,𝑖𝑗 is
normalized. As we discussed in Section 3, variable n is a proxy for the marketable managerial
capital of CEOs. Accordingly, the independent variables 𝑆𝑛,𝑖𝑗 captures the similarity between
CEO i and CEO j in their marketable managerial capital. Specifically, 𝑆𝑛,𝑖𝑗 covers the differences
in firm characteristics, such as the difference in firm sales, the difference in ROA, the difference
in market-to-book, and the same industry dummy, etc. 𝑆𝑛,𝑖𝑗 also covers the differences in CEO
lifetime experience, such as the difference in the General Ability Index, the difference in the number
of worked positions, the difference in the number of worked firms, whether CEO i and CEO j
both have conglomerate experience, etc. Table 4 provides a complete list of the independent
variables 𝑆𝑛,𝑖𝑗, as well as the method how we construct each variable. The hypothesis is that a
larger difference 𝑆𝑛,𝑖𝑗 between CEO i and CEO j would lead to a smaller probability of CEO j
being chosen as a peer of CEO i.
More importantly, for each difference 𝑆𝑛,𝑖𝑗 between CEO i and CEO j, we further
construct a pair of asymmetric measures 𝑆𝑛,𝑖𝑗+ and 𝑆𝑛,𝑖𝑗
− . The first variable 𝑆𝑛,𝑖𝑗+ is equal to the
difference 𝑆𝑛,𝑖𝑗 when the difference is positive and is set to equal to zero otherwise. The second
variable 𝑆𝑛,𝑖𝑗− is equal to the difference 𝑆𝑛,𝑖𝑗 when the difference is negative, and zero otherwise.
The hypothesis is that, in the absence of biases in the choice of peers, a positive difference 𝑆𝑛,𝑖𝑗
and a negative difference 𝑆𝑛,𝑖𝑗 with the same absolute value would identically reduce the
probability of CEO j being chosen as a peer of CEO i. Hence, the pair of asymmetric measures,
𝑆𝑛,𝑖𝑗+ and 𝑆𝑛,𝑖𝑗
− , would allow us to identify possible selection biases by examining whether the
sign of 𝑆𝑛,𝑖𝑗 may affect the choice of peers in distinguishing ways.
The size of peer group differs across different CEOs. So the unconditional probability
of a potential CEO j being selected as a peer for a particular CEO i would depend on the size of
CEO i’s peer group. Our regressions include CEO fixed-effects to account for such probability
difference.
Table 4 presents the coefficient estimates in logit regressions. The p-values are reported
in parentheses below, and the marginal effects are reported in brackets. Table 4 shows that nearly
all the coefficient estimates are statistically significant. The coefficient estimates on firm
characteristics indicates that firms tend to select CEO compensation peers in the same industry,
from firms with similar size, similar accounting performance, similar market-to-book ratio,
similar credit rating, and same S&P 500 membership. Moreover, the estimates on firm
characteristics also show asymmetric effects of relative firm size and firm performance on peer
selection. According to our definitions in Section 2, this is a type-two bias, where a firm
systematically chooses peer CEOs whose managerial capital is more productive. This type of
selection bias is the main findings in BLN (2011).
The main new findings from our regressions are the coefficient estimates on CEO
characteristics. From the model 2 in table 4, coefficient estimates on the General Ability Index (GAI)
are statistically significant. Conditional on the potential peer CEO’s GAI being smaller than the
firm CEO’s, one standard deviation increase in relative GAI decreases the probability of the CEO
being chosen as a compensation peer by 6.1% (table 4, coefficient 𝛼12). However, conditional on
the potential peer CEO’s GAI being higher than the firm CEO’s, one standard deviation increase
in relative GAI actually increases, rather than decreases, the probability of being chosen by 0.7%
(table 4, coefficient 𝛼11). The coefficient estimate on positive GAI difference is surprisingly
significant positive, rather than having the expected negative sign. This positive sign means that,
all else equal, if CEO j’s GAI is higher than the GAI of CEO I, there would be a higher
probability of CEO j being chosen as a peer of CEO i. The effects of relative GAI on peer
selection are very asymmetric.
An alternative approach is to classify each CEO as either a generalist CEO or a specialist
CEO according to the median of GAI distribution in each year. Then we define two dummy
variables. The first dummy variable takes a value of one if a specialist CEO is chosen as a peer
for a generalist CEO, and zero otherwise. The second dummy variable takes a value of one if a
generalist CEO is chosen as a peer for a specialist CEO, and zero otherwise. In the model 3 in
table 4, we present the regression estimates after replacing the positive and negative GAI
differences with the two dummy variables. The dummy coefficients are both significant, but the
coefficient estimate on the second dummy (table 4, coefficient 𝛼14) has a positive sign rather than
a negative sign. All else equal, a specialist CEO are 8.7% less likely to be chosen as a compensation
peer for a generalist CEO (table 4, coefficient 𝛼13), but a generalist CEO are 3.5% more likely to
be chosen as a compensation peer for a specialist CEO (table 4, coefficient 𝛼14). This one-sided
preference to generalist CEOs indicate that firms tend to choose the peer CEOs with higher
general managerial ability. To the best of our knowledge, we are the first to document this type
of peer selection biases. According to our definitions in Section 2, this is a type-one bias, where
a firm systematically chooses peer CEOs whose managerial capital is more marketable. This type
of selection bias is orthogonal to the biases documented in prior papers, which are basically
mismatches in firm characteristics.
A following important question is whether our newly documented peer-selection bias is
economically significant. As shown in table 4 model 3, the marginal probability is 3.5% (table 4,
coefficient 𝛼14) for a specialist choosing a generalist, versus a marginal probability -8.7% (table
4, coefficient 𝛼13) for a generalist choosing a specialist. To balance this asymmetry, the marginal
probability for a specialist choosing a generalist would need to reduce by 12.2%. From our
estimate results in section 4.1, a generalist earns about 18% more than a specialist. Accordingly,
a 12.1% reduction in the marginal probability for a specialist choosing a generalist would lead to
an about 2.2% reduction in CEO pay. At first glance, a 2.2% bias in CEO pay may seem not very
economically large. However, because firms need to select peers to construct compensation peer
groups every year, even a modest systematic bias in each selection can lead to a ratcheting of
CEO pays over years, and therefore considerably inflate CEO pays in the long run. For example,
over a period of 30 years, a 2.2% annual inflation would eventually lead to nearly 92% inflation
in CEO pay.
In table 4 model 1, similar asymmetric effects also show up for the individual variables
that construct the GAI (coefficients 𝛼15-𝛼23). For example, conditional on a potential peer CEO
j’s worked positions being less than a CEO i’s, all else equal, one standard deviation increase of
the relative difference in worked positions decreases the marginal probability of the CEO j being
chosen as a compensation peer by 5.6%. However, conditional on a potential peer CEO j’s worked
positions being more than a CEO i’s, one standard deviation increase of the relative difference in
worked positions increases, rather than decrease, the marginal probability of the CEO j being
chosen by 0.5%. Again, conditional on a potential peer CEO j’s worked industries being less than
a CEO i’s, one standard deviation increase of the relative difference in worked industries
decreases the marginal probability of the CEO j being chosen by 1.8%. However, conditional on
a potential peer CEO j’s worked industries being more than a CEO i’s, all else equal, one standard
deviation increase of the relative difference in worked industries increases, rather than decrease,
the marginal probability of the CEO j being chosen by 1.6%. The positive difference and negative
difference have asymmetric effects on peer selection. The asymmetry indicates that firms prefer
to select peer CEOs with more position experience and more industry experience.
Other noteworthy findings are that a CEO holding chairman position is 9% more likely
to be chosen as compensation peers for another chairman CEO (table 4, 𝛼27). However, a non-
chairman CEO is not more likely to be chosen as peer for another non-chairman CEO and the
coefficient estimate is statistically insignificant (table 4, 𝛼28 ). All else equal, one standard
deviation change in relative age when the potential peer CEO j is elder (younger) than the CEO
i decreases the marginal probability of the potential CEO j being chosen as a peer by 1.7% (3.8%).
While both positive and negative age differences decrease the probability of being chosen as peer,
firms are more reluctant to choose a younger CEO. Moreover, a CEO who has (doesn’t have)
conglomerate experience is more likely to be chosen as compensation peer for another CEO with
(without) conglomerate experience (𝛼21, 𝛼22). A CEO who has no previous CEO experience is
more likely to be chosen as compensation peer for another CEO with no pervious CEO experience
(𝛼24). A CEO who holds MBA degree is more likely to be chosen as compensation peers for
another CEO holding MBA degree (𝛼29). A CEO who graduated from Ivy League schools is
more likely to be chosen as compensation peers for another CEO with Ivy League degree (𝛼30).
We also estimate the regressions for the subsample in each year from 2006 to 2012, rather
than for the full sample. The SEC disclosure rules require that firms report the peer groups they
use to set up executives compensation. While the rule was implemented at the end of 2006, most
of the 2006 peer groups were constructed early in the fiscal year. Every year thereafter, firms
constructed their peer groups with being aware of the new disclosure rule. The policy change
may raise a question whether the disclosure rule mitigated peer selection biases. In unreported
results, we find that peer selection biases show up for each subsample year from 2006 to 2012.
We don’t find any evidences that the selection biases diminish over the years following the SEC
disclosure requirements of 2006.
We then address the question whether the peer selection biases are largely driven by small
firms. We do multivariate analysis separately for S&P 500 firms and non-S&P 500 firms, and we
report the results in Table 5. The asymmetric effects of relative CEO characteristics 𝑆𝑛,𝑖𝑗+ and
𝑆𝑛,𝑖𝑗− on peer selections show up for both groups of firms. For example, for S&P500 firms,
conditional on the potential peer CEO’s GAI being larger than the firm CEO’s, one standard
deviation increase in relative GAI actually increases, rather than decreases, the probability of the
CEO being chosen as a peer by 1.6%. Conditional on the potential peer CEO’s GAI being smaller
than the firm CEO’s one standard deviation decrease in relative GAI actually decreases the
probability of the CEO being chosen as a peer by 4.3%. For non-S&P 500 firms, conditional on
the potential peer CEO’s GAI being larger (smaller) than the firm CEO’s, one standard deviation
difference in relative GAI actually decreases the probability of the CEO being chosen as a peer
by 0.4% (7.9%). Generally, the asymmetric effects are statistically and economically significant
for both non-S&P 500 firms and S&P 500 firms.
We also separately estimate the regressions for some subgroups of CEOs. In unreported
results, we find the asymmetric effects of similar magnitudes to chairman CEOs and non-
chairman CEOs, to externally hired CEOs and internally promoted CEOs. I don’t find empirical
evidence that peer selections biases are driven by a particular type of CEOs or firms.
Overall, consistent with our hypothesis in Section 2, the multivariate analysis indicates
that the characteristics of firms and the characteristics of CEO lifetime experiences are both
significant determinants of peer choice. Firms construct compensation peer groups largely by
selecting peer CEOs whose marketable managerial capitals are similar to the firm CEOs’.
Nevertheless, the asymmetric effects indicate some opportunistic behavior in peer selections.
Firms tend to misrepresent the labor market competition by favoring more productive (type II
biases) and more marketable (type I biases) CEOs as compensation peers.
4.3. Managerial Human Capital and Peer Group’s Median
In addition to the results from multivariate analysis above, we adopt another approach to
investigate peer selection biases by examining the characteristics the peer group’s median. In
practice, firms generally target CEO pay at the 50th percentile of peer pays or above.9 Hence, it
is informative to investigate possible mismatches by comparing the characteristics of a firm CEO
with the median value of his/her peer CEOs. Unfortunately, this approach will not be able to
precisely measure peer selection biases, because many firms target above the 50th percentile of
peer CEOs. Nevertheless, the peer group’s median is informative in the sense that it provides the
lower limit of actual biases in peer selection. This approach is also a robustness check for our
main findings in section 4.2.
In table 6, we report the differences between sample CEOs and the median of each CEO’s
peer group. The characteristics of interest include firm size, firm accounting performance, CEO’s
General Ability Index, the number of worked positions, the number of worked firms, and the
numbers of worked industries. We reports results for all sample CEOs, as well as for CEOs from
S&P 500 firms and non-S&P 500 firms separately. Overall, we find that the peer-group-median
firm are larger and performing better. The peer-group-median firm is 5.0% bigger (in log sales)
than the sample CEO’s firm. The accounting performance of peer-group-median firm is
marginally better than the performance of the sample CEO’s firm. Table 5 also shows that the
peer-group-median CEO has higher general managerial ability than the sample CEO. The
median GAI of peer CEOs is 0.042 higher than a sample CEO’s. These mismatches are consistent
with the selection biases that we find from our logit analysis in section 4.2. Especially, table 5
shows that the mismatches are larger for non-S&P 500 firms. For the non-S&P 500 firm, peer-
group-median firm is 8.8% bigger (in log sales) than the sample CEO’s firm, and peer-group-
median CEO is 0.104 higher in GAI than the sample CEO. From our results in section 4.1, the
loading of log CEO pay on log firm sales is 0.343, and the loading of log CEO pay on GAI is
0.115.10 Accordingly, a 8.8% difference in firm size could lead to about a 3.0% increment in CEO
pays, and a 0.104 difference in GAI could lead to about a 1.2% increment in CEO pays. The biases
may seem not economically very large. Nevertheless, the mismatches provide only the lower limit
of actual selection biases. More importantly, as we point out before, even small systematic biases
9 For example, BLN(2011) reports that among their sample firms disclosing a specific pay target, 71.5% of them
target at the 50th percentile of peer pays, 28.5% of them target above the 50th percentile, and 0% of them targets
below the 50th percentile. 10 As reference, in Gabaix, Landier, and Sauvagnat (2014), the loading of log CEO pay on log firm sales is 0.364. In
CFM (2013), the loading of log CEO pay on GAI is 0.117.
can lead to a ratcheting of CEO compensations over years, and considerably inflate CEO pays in
the long run.
5. Conclusion
In this paper, we suggest a new criterion on compensation peer selection. We argue that
a CEO’s peer is another CEO, not a firm. A compensation peer group should be a group of other
CEOs, not a group of firms. Firms need to select the CEOs with similar marketable managerial
capital, rather than similar firms, to construct peer groups. Then we attempt to test the role of
CEO managerial capital in actual peer selections. We consider a broad set of proxy variables that
may reflect transferable CEO managerial capital, and examine the roles of these variables in
determining the choice of peers. We find that our proxy variables, including both firm
characteristics and CEO experience characteristics, are important determinants of peer choice.
Firms tend to choose the peer CEOs from firms of same industry, similar size, similar
performance, and similar market-to-book ratios. Firms also tend to choose the peer CEOs with
similar ages, similar educational backgrounds, similar conglomerate experience, and similar
multi-firm experiences, etc.
However, biased peer selection would lead to mismatches in managerial capital between a
CEO and selected peers. There are two types of possible selection biases: mismatches in
transferability and mismatches in productivity. Our empirical analysis shows that actual peer
groups are biasedly structured in both ways. Firms are more likely to choose peer CEOs whose
managerial capital is more marketable and peer CEOs who work for larger and better performing
firms.
Dramatic data explosion has been seen as one of most exciting progresses in recent years.
Data at the level of individuals is helping to reshape the business practices in many areas. In the
area of CEO contract design, it has become much easier to assess the detailed information on
CEO lifetime experiences. We might be at the doorstep to an era when CEO compensation design
can be highly personalized and deeply relies on the detailed record of each individual CEO’s past
experiences.
References
Albuquerque, A, G. De Franco, and R. Verdi, 2013. Peer choice and CEO compensation, Journal
of Financial Economics 108, 160-181.
Becker, G., 1962. Investment in human capital: A theoretical analysis. Journal of Political
Economy 70, 9-49.
Bertrand, M., Schoar, A., 2003. Managing with style: The effect of managers on firm policies,
Quarterly Journal of Economics 68, 1169-1208.
Bizjak, J., M. Lemmon, and L. Naveen, 2008. Does the use of peer groups contribute to higher
pay and less efficient compensation? Journal of Financial Economics 90 (2) 152-168.
Bizjak, J., M. Lemmon, and T. Nguyen, 2011. Are all CEOs above average? An empirical
analysis of compensation peer groups and pay design. Journal of Financial Economics 100, 538–
555.
Cademan, B. and M. Carter, 2013. Compensation peer groups and the relation with CEO
compensation. Journal of Management Accounting Research
Cunat, V., Guadalupe, M., 2009b. Globalization and the provision of incentives inside the firm:
The effect of foreign competition. Journal of Labor Economics 27, 179-212.
Custodio, C., Ferreira, M.A., and Matos, P., 2013. Generalists versus Specialists: Lifetime Work
Experience and CEO Pay., Journal of Financial Economics.
Faulkender, M. and J. Yang, 2010. Inside the black box: The role and composition of
compensation peer groups. Journal of Financial Economics.
Faulkender, M. and J. Yang, 2013. Is Disclosure an Effective Cleansing Mechanism? The
Dynamics of Compensation Peer Benchmarking. Review of Financial Studies 26, 806-839
Fee, C., Hadlock, C., 2003. Raids, rewards, and reputations in the market for managerial talent.
Review of Financial Studies 16, 1315–1357.
Frydman, C., Jenter, D., 2010. CEO compensation. Annual Review of Financial Economics 2, 75-
102.
Frydman, C., 2005. Rising through the ranks: The evolution of the market for corporate
executives, 1936-2003. Unpublished working paper, MIT Sloan School.
Gabaix, X., Landier, A., 2008. Why has CEO pay increased so much? Quarterly Journal of
Economics 123, 49-100.
Graham, J., Li, S., Qiu, J., 2012. Managerial attributes and executive compensation. Review of
Financial Studies 25, 144–186.
Holmstrom, B., Kaplan, S., 2003. The state of US corporate governance: what’s right and what’s
wrong? Journal of Applied Corporate Finance 15, 8–20.
Linck, J., Netter, J., Yang, T., 2008. The determinants of board structure. Journal of Financial
Economics 87, 308–328.
Marin, D. and Verdier, T., 2004. Globalization and the empowerment of talent, CEPR Discussion
Paper, No. 4129.
Murphy, K. J., Zabojnik, J., 2007. Managerial capital and the market for CEOs. Unpublished
working paper, University of Southern California.
Murphy, K. J., Zabojnik, J., 2004. CEO pay and appointments: A market-based explanation for
recent trends. American Economic Review Papers and Proceedings 94, 192-196.
Table A Variable Definitions
Variable Description and Data Source
Panel A: CEO Compensation
Total Compensation Total Compensation in thousands $ = (Salary + Bonus + Other annual + Restricted Stock Grants +
LTIP Payouts + All Other + Value of Options Granted for 2005 fiscal year; and Salary + Bonus+
Non-Equity Incentive Plan Compensation + Value of Options Granted + Grant-Date Fair Value of
Stock Awards + Deferred Compensation Earnings Reported as Compensation + Other Compensation
for 2006 fiscal year), (Execucomp).
Panel B: CEO Characteristics
Number of Positions Number of positions CEO has had based on past work experience in publicly traded firms (BoardEx).
Number of Firms Number of firms where CEO has worked based on past work experience in publicly traded firms
(BoardEx).
Number of Industries Number of industries (four-digit SIC) where CEO has worked based on past work experience in
publicly traded firms (BoardEx).
Prior Top-Manager
Experience
Dummy variable that takes a value of one if CEO held a CEO position at another company based on
past work experience in publicly traded firms, and zero otherwise (BoardEx).
Conglomerate
Experience
Dummy variable that takes a value of one if CEO worked at multi-segment company based on past
work experience in publicly traded firms, and zero otherwise (BoardEx).
External Hiring Dummy variable that takes a value of one if CEO was hired from outside the firm, and zero
otherwise (BoardEx).
CEO Age Age of CEO in years (BoardEx).
CEO Tenure Number of years as CEO in the current position (BoardEx)
CEO-Chair Dummy Dummy variable that takes a value of one if CEO is also chair of the board, and zero otherwise
(BoardEx).
First Year as CEO
Dummy
Dummy variable that takes a value of one if CEO is in the first year of the job, and zero otherwise
(BoardEx).
MBA Dummy Dummy variable that takes a value of one if CEO has a MBA degree, and zero otherwise (BoardEx).
IVY League Dummy Dummy variable that takes a value of one if CEO attended an Ivy League school (Brown University,
Columbia University, Cornell University, Dartmouth College, Harvard University, Princeton
University, University of Pennsylvania, and Yale University) at any academic level, and zero
otherwise (BoardEx).
Panel C: Firm Characteristics
Sale Sales (Compustat)
Log (Sale) Log (Sales Revenue)
ROA Income Before Extraordinary Items / total assets (Compustat)
Volatility Annualized standard deviation of monthly stock returns (CRSP)
Firm Age Number of years since a firm listed its shares (CRSP).
Diversification Dummy Dummy variable that takes a value of one if a firm has more than one business segment, and zero
otherwise (Compustat).
Market-To-Book (market equity + total debt + preferred stock liquidating value – deferred taxes and investment tax
credits)/ book assets (Compustat).
Credit Rating Dummy variable that takes a value of one if S&P Domestic Long Term Issuer Credit Rating is
between BBB- and AAA, two if the credit rating is between CC and BB+, three if the credit
rating is D or SD , and four if the credit rat0ing is missing (Compustat).
Table A Variable Definitions (continued)
Panel D: Dependent Variables (Logit Regressions)
General ability index
difference (positive)
= (the GAI of Peer CEO firm – the GAI of sample CEO firm) if the GAI of Peer CEO firm > the
GAI of sample CEO firm; = 0 otherwise
General ability index
difference (negative)
= (the GAI of Peer CEO firm – the GAI of sample CEO firm) if the GAI of Peer CEO firm < the
GAI of sample CEO firm; = 0 otherwise
Generalist being
chosen for specialist
Dummy equal to one if a generalist CEO is chosen as peer for a specialist CEO, zero otherwise.
Specialist being chosen
for generalist
Dummy equal to one if a specialist CEO is chosen as peer for a generalist CEO, zero otherwise.
Num. of positions
difference (positive)
= (the number of positions of Peer CEO – the number of positions of sample CEO) if the number
of positions of Peer CEO > the number of positions of sample CEO; = 0 otherwise
Num. of positions
difference (negative)
= (the number of positions of Peer CEO – the number of positions of sample CEO) if the number
of positions of Peer CEO < the number of positions of sample CEO; = 0 otherwise
Num. of industries
difference (positive)
= (the number of industries of Peer CEO – the number of industries of sample CEO) if the number
of industries of Peer CEO > the number of industries of sample CEO; = 0 otherwise
Num. of industries
difference (negative)
= (the number of industries of Peer CEO – the number of industries of sample CEO) if the number
of industries of Peer CEO < the number of industries of sample CEO; = 0 otherwise
Num. of firms
difference (positive)
= (the number of firms of Peer CEO – the number of firms of sample CEO) if the number of firms
of Peer CEO > the number of firms of sample CEO; = 0 otherwise
Num. of firms
difference (negative)
= (the number of firms of Peer CEO – the number of firms of sample CEO) if the number of firms
of Peer CEO < the number of firms of sample CEO; = 0 otherwise
Age difference
(positive)
= (the age Peer CEO – the age sample CEO) if the age of Peer CEO > the age of sample CEO; = 0
otherwise
Age difference
(negative)
= (the age Peer CEO – the age sample CEO) if the age of Peer CEO < the age of sample CEO; = 0
otherwise
CEO-Chair Dummy equal to one if both sample CEO and peer CEO are chairman of the board of directors
Not CEO-Chair Dummy equal to one if both sample CEO and peer CEO are not chairman of the board of directors
Conglomerate
experience
Dummy equal to one if both sample CEO and peer CEO have conglomerate experience
No conglomerate
experience
Dummy equal to one if both sample CEO and peer CEO have no conglomerate experience
CEO experience Dummy equal to one if both sample CEO and peer CEO had CEO experience at another firm
No CEO experience Dummy equal to one if both sample CEO and peer CEO had no CEO experience at another firm
MBA degree Dummy equal to one if both sample CEO and peer CEO have a MBA degree
IVY League Dummy equal to one if both sample CEO and peer CEO graduated from an IVY league school
Same Industry Dummy equal to one if both CEO and peer are in the same FF-49 industry
Industry return
correlation
Correlation between CEO’s industry return and potential peer CEO’s industry return
Sale difference
(positive)
= (the sale of Peer CEO firm – the sale of sample CEO firm) if the sale of Peer CEO firm > the sale
of sample CEO firm; = 0 otherwise
Sale difference
(negative)
= (the sale of Peer CEO firm – the sale of sample CEO firm) if the sale of Peer CEO firm < the sale
of sample CEO firm; = 0 otherwise
ROA difference
(positive)
= (the ROA of Peer CEO firm – the ROA of sample CEO firm) if the ROA of Peer CEO firm > the
ROA of sample CEO firm; = 0 otherwise
ROA difference
(negative)
= (the ROA of Peer CEO firm – the ROA of sample CEO firm) if the ROA of Peer CEO firm < the
ROA of sample CEO firm; = 0 otherwise
ME/BE difference
(positive)
= (the ME/BE of Peer CEO firm – the ME/BE of sample CEO firm) if the ME/BE of Peer CEO
firm > the ME/BE of sample CEO firm; = 0 otherwise
ME/BE difference
(negative)
= (the ME/BE of Peer CEO firm – the ME/BE of sample CEO firm) if the ME/BE of Peer CEO
firm < the ME/BE of sample CEO firm; = 0 otherwise
Table 1
Summary Statistics for the Size of Compensation Peer Groups
Compensation peer group data comes from the Incentive Lab Compensation and Metrics Data. The sample covers
the S&P1500 firms that reported their compensation peer groups in the fiscal years 2006 through 2012. This table
presents the number of S&P1500 firms that reported peer groups, as well as descriptive statistics for the number of
peers in each reported peer group. The numbers are reported for the entire sample, as well as for the subsample in
each fiscal year. Year Firms Number of Peers
Mean Std. 1st 5th 25th 50th 75th 95th 99th
2006 878 17.5 11.4 3 6 11 16 21 34 52
2007 1061 21.1 24.4 3 7 12 16 23 47 116
2008 1047 22.2 25.4 4 7 13 17 22 51 144
2009 1035 25.8 58.2 5 8 13 17 22 54 172
2010 1038 23.5 40.8 5 9 13 17 22 51 143
2011 1016 24.5 50.8 5 8 13 17 22 47 172
2012 91411 21.2 33.8 5 9 14 17 20 40 110
All 6989 22.4 38.5 4 7 13 17 22 45 137
11 Our sample data doesn’t cover all the firms that reported compensation peers in the fiscal year 2012.
Table 2
Summary Statistics for Proxy Variables
The variables are proxy variables for the marketable managerial human capital of CEOs. The variables belong to two
categories: the characteristics of a CEO’s current employer firm, and the characteristics of a CEO’s lifetime work
and education experiences. We compute the summary statistics both for our sample firm-CEOs and for all
observations of selected peer-CEOs. In addition, the general ability index (GAI) is also presented. The sample covers
fiscal years 2006 to 2012. CEO characteristics data are from BoardEx. Firm characteristics data are from Compustat.
Variable definitions are provided in Table A.
Firm CEOs Peer CEOs
Mean Median Std. Mean Median Std.
CEO characteristics:
Number of positions worked
9.6 8 6.5 10.3 9 6.6
Number of firms worked for 3.0 3 2.1 3.2 3 2.1
Number of industries worked 2.1 2 1.2 2.2 2 1.2
Prior CEO experience 14.5% - - 14.5% - -
Conglomerate experience 42.5% - - 43.2% - -
CEO-Chair 57.3% - - 60.6% - -
MBA degree 35.7% - - 36.8% - -
IVY league degree 25.7% - - 25.9% - -
Age 55.8 56 6.5 56.2 56 6.5
General Ability Index
(GAI) 0.264 0.909 0.092 0.344 0.924 0.156
Firm characteristics:
Sale ($B) 10.9 26.8 3.4 13.7 29.2 4.4
ROA 3.98% 11.19% 4.30% 4.32% 10.81% 4.78%
Market-to-book 1.41 1.17 1.23 1.39 1.05 1.13
Credit rating b/w BBB- and AAA 50.0% - - 56.1% - -
S&P 500 51.0% - - 58.4% - -
Table 3
CEO Pay and General Managerial Ability
The panel regressions of the logarithm of CEO total pay on firm and CEO control variables, as well as the General
Ability Index and the general ability index dummy which takes the value of one if the index is above the annual
median. The regressions include year and industry (two-digit SIC codes) fixed effect. Model (1), (2), (3) and (5) present
OLS regressions. Model (4) and (6) include firm fixed effects. The test sample consists of a pool of CEOs and peer
CEOs for which compensation data are available from EXECUMP in the 2006-2012 period. Heteroskedasticity-robust
t-statistics are adjusted for firm level clustering. ***, ** and * denote significance at 1%, 5% and 10% levels.
(1) (2) (3) (4) (5) (6)
OLS OLS OLS Firm fixed
effects OLS
Firm fixed
effects
Number of Positions 0.005* 0.005
Number of Firms 0.049*** 0.038**
Number of industries -0.011 -0.053*
Prior Top-Manager Experience -0.091** -0.033
Conglomerate experience 0.046 -0.012
General Ability Index 0.100***
General Ability Index Dummy 0.179*** 0.070*
CEO Tenure -0.005 -0.003 -0.002 0.002 -0.003 0.001
MBA Dummy 0.096*** 0.075** 0.079*** 0.080* 0.080*** 0.077*
First Year As CEO Dummy -0.041 -0.036 -0.032 0.004 -0.033 0.007
CEO-Chair Dummy 0.113*** 0.115*** 0.113*** 0.021 0.116*** 0.021
CEO Age -0.006** -0.007** -0.008** 0.000 -0.007** 0.000
Stock Return 0.008 0.009 0.009 0.014* 0.008 0.014*
Stock Return (t-1) -0.000 -0.000 -0.000 -0.000*** -0.000 -0.000***
Firm Age 0.002* 0.002 0.002 -0.000 0.002 -0.000
Log(Sale) 0.377*** 0.359*** 0.359*** 0.272*** 0.362*** 0.272***
Market to Book 0.059** 0.056** 0.057** 0.061** 0.059** 0.061**
ROA 0.099 0.115 0.104 0.487** 0.079 0.498**
ROA (t-1) -0.546** -0.504* -0.492* -0.107 -0.486* -0.090
Diversification Dummy 0.000 0.002 0.004 -0.031 0.008 -0.031
Volatility -1.658*** -1.681*** -1.715*** -1.187*** -1.694*** -1.187***
Observations 8,172 8,171 8,171 8,176 8,171 8,176
R-squared (%) 35.55 36.45 36.18 76.42 36.13 76.39
Table 4
Logit Analysis of Peer Selection
Logit regressions of the proxy variables for CEOs’ marketable managerial capital. The dependent variable is one if
a potential peer CEO is chosen as a peer by the sample firm and zero otherwise. In each year, the set of potential
peer-CEOs includes the union of sample firm-CEOs and the chosen peer-CEOs. The logit regression contains all the
sample years 2006-2012. P-values are given in parentheses and p-values for symmetry tests on each pair of
coefficients are in brackets. Non-dummy variables are normalized. Regression model 1 includes CEO characteristics
as dependent variables. Regression model 2 adopts CEO general ability index (GAI) as a dependent variable. Detailed
variable definitions are provided in the Table A.
Dependent variable is one if a potential peer-
CEO is chosen as a peer by the sample firm
and zero otherwise
Model 1 Model 2 Model 3
Firm characteristics
𝛼1 Dummy equal to one if both firm-CEO and peer-CEO are
in the same Fama-French industry
3.822
(0.000)
[0.944]
3.819
(0.000)
[0.944]
3.817
(0.000)
[0.944]
𝛼2 Dummy equal to one if both firm-CEO and peer-CEO
work for firms with the same credit rating
0.385
(0.000)
[0.153 ]
0.388
(0.000)
[0.154]
0.389
(0.000)
[0.154]
𝛼3 Dummy equal to one if both firm-CEO and peer-CEO
work for S&P 500 firms
1.063
(0.000)
[0.405]
1.075
(0.000)
[0.409]
1.074
(0.000)
[0.409]
𝛼4 Dummy equal to one if both firm-CEO and peer-CEO
don’t work for S&P 500 firms
-0.243
(0.000)
[-0.097]
-0.249
(0.000)
[0.099]
-0.246
(0.000)
[0.098]
𝛼5 Log peer sales – Log firm sales when peer sales > firm
sales, =0 otherwise
-1.041
(0.000)
[-0.397]
-1.038
(0.000)
[-0.396]
-1.041
(0.000)
[-0.397]
𝛼6 Log peer sales – Log firm sales when peer sales < firm
sales, =0 otherwise
2.383
(0.000)
[0.767]
2.388
(0.000)
[0.768]
2.386
(0.000)
[0.767]
𝛼7 Peer ROA – firm ROA when peer ROA > firm ROA, =0
otherwise
-0.167
(0.000)
[-0.067]
-0.163
(0.000)
[-0.067]
-0.162
(0.000)
[-0.064]
𝛼8 Peer ROA – firm ROA when peer ROA < firm ROA, =0
otherwise
0.077
(0.000)
[0.031]
0.086
(0.000)
[0.034]
0.089
(0.000)
[0.035]
𝛼9 Peer MTB – firm MTB when peer MTB > firm MTB, =0
otherwise
-0.076
(0.000)
[-0.030]
-0.079
(0.000)
[-0.032]
-0.078
(0.000)
[-0.031]
𝛼10 Peer MTB – firm MTB when peer MTB < firm MTB, =0
otherwise
0.917
(0.000)
[0.353]
0.917
(0.000)
[0.353]
0.918
(0.000)
[0.354]
Table 4 (continued)
Dependent variable is one if a potential peer-
CEO is chosen as a peer by the sample firm
and zero otherwise
Model 1 Model 2 Model 3
CEO characteristics
𝛼11 Peer-CEO’s GAI – firm-CEO’s GAI when peer-CEO’s
GAI > firm-CEO’s GAI, =0 otherwise
0.018
(0.000)
[0.007]
𝛼12 Peer-CEO’s GAI – firm-CEO’s GAI when peer-CEO’s
GAI < firm-CEO’s GAI, =0 otherwise
0.154
(0.000)
[0.061]
𝛼13 Dummy equal to one if a specialist CEO is chosen as a
peer for a generalist CEO, =0 otherwise
-0.218
(0.000)
[-0.087]
𝛼14 Dummy equal to one if a generalist CEO is chosen as a
peer for a specialist CEO, =0 otherwise
0.089
(0.000)
[0.035]
𝛼15 Peer-CEO’s positions – firm-CEO’s positions when peer-
CEO’s positions > firm-CEO’s positions, =0 otherwise
0.012
(0.025)
[0.005]
𝛼16 Peer-CEO’s positions – firm-CEO’s positions when peer-
CEO’s positions < firm-CEO’s positions, =0 otherwise
0.146
(0.000)
[0.058]
𝛼17 Peer-CEO’s firms – firm-CEO’s firms when peer-CEO’s
firms > firm-CEO’s firms, =0 otherwise
-0.030
(0.000)
[-0.012]
𝛼18 Peer-CEO’s firms – firm-CEO’s firms when peer-CEO’s
firms < firm-CEO’s firms, =0 otherwise
0.024
(0.000)
[0.010]
𝛼19 Peer-CEO’s industries – firm-CEO’s industries when peer-
CEO’s industries > firm-CEO’s industries, =0 otherwise
0.040
(0.000)
[0.016]
𝛼20 Peer-CEO’s industries – firm-CEO’s industries when peer-
CEO’s industries < firm-CEO’s industries, =0 otherwise
0.044
(0.000)
[0.018]
𝛼21 Dummy equal to one if both firm-CEO and peer-CEO
have conglomerate experience, =0 otherwise
0.034
(0.003)
[0.014]
𝛼22 Dummy equal to one if both firm-CEO and peer-CEO
don’t have conglomerate experience, =0 otherwise
0.057
(0.000)
[0.023]
𝛼23 Dummy equal to one if both firm-CEO and peer-CEO
have previous experience as CEO, =0 otherwise
-0.051
(0.052)
[-0.020]
𝛼24 Dummy equal to one if both firm-CEO and peer-CEO
don’t have previous experience as CEO, =0 otherwise
0.096
(0.000)
[0.038]
Table 4 (continued)
Dependent variable is one if a potential peer-
CEO is chosen as a peer by the sample firm
and zero otherwise
Model 1 Model 2 Model 3
CEO characteristics (continued)
𝛼25 Peer-CEO’s age – firm-CEO’s age when peer-CEO’s age >
firm-CEO’s age, =0 otherwise
-0.043
(0.000)
[-0.017]
-0.046
(0.000)
[-0.018]
-0.052
(0.000)
[-0.021]
𝛼26 Peer-CEO’s age – firm-CEO’s age when peer-CEO’s age <
firm-CEO’s age, =0 otherwise
0.096
(0.000)
[0.038]
0.100
(0.000)
[0.040]
0.097
(0.000)
[0.038]
𝛼27 Dummy equal to one if both firm-CEO and peer-CEO
hold chairman positions, =0 otherwise
0.223
(0.000)
[0.089]
0.238
(0.000)
[0.095]
0.236
(0.000)
[0.094]
𝛼28 Dummy equal to one if both firm-CEO and peer-CEO
don’t hold chairman positions, =0 otherwise
-0.019
(0.079)
[-0.008]
-0.030
(0.072)
[-0.012]
-0.022
(0.047)
[-0.009]
𝛼29 Dummy equal to one if both firm-CEO and peer-CEO
have MBA degrees, =0 otherwise
0.061
(0.000)
[0.024]
0.063
(0.000)
[0.025]
0.057
(0.000)
[0.023]
𝛼30 Dummy equal to one if both firm-CEO and peer-CEO
attended Ivy League schools, =0 otherwise
0.077
(0.000)
[0.031]
0.077
(0.000)
[0.031]
0.077
(0.000)
[0.031]
Table 5
SP500 Firms and Non-SP500 firms
Logit regressions of the proxy variables for CEOs’ marketable managerial capital for SP500 firms and Non-SP500
firms. The dependent variable is one if a potential peer CEO is chosen as a peer by the sample firm and zero otherwise.
In each year, the set of potential peer-CEOs includes the union of sample firm-CEOs and the chosen peer-CEOs.
The logit regression contains all the sample years 2006-2012. P-values are given in parentheses and p-values for
symmetry tests on each pair of coefficients are in brackets. Non-dummy variables are normalized. Detailed variable
definitions are provided in the Table A.
Dependent variable is one if a potential peer-CEO is chosen as a peer by the
sample firm and zero otherwise
Regression Model (1) Regression Model (2) Regression Model (3)
SP500 Non-SP500 SP500 Non-SP500 SP500 Non-SP500
Firm characteristics
𝛼1 Dummy equal to one if in
the same Fama-French
industry
3.855
(0.000)
[0.946]
3.813
(0.000)
[0.943]
3.850
(0.000)
[0.946]
3.813
(0.000)
[0.943]
3.848
(0.000)
[0.946]
3.812
(0.000)
[0.943]
𝛼2 Dummy equal to one if
with the same credit
rating
0.450
(0.000)
[0.178]
0.304
(0.000)
[0.121]
0.455
(0.000)
[0.180]
0.304
(0.000)
[0.121]
0.455
(0.000)
[0.180]
0.304
(0.000)
[0.121]
𝛼3 Dummy equal to one if
both S&P 500 firms 0.986
(0.000)
[0.378]
- 0.995
(0.000)
[0.381]
- 0.990
(0.000)
[0.380]
-
𝛼4 Dummy equal to one if
both not S&P 500 firms - -0.367
(0.000)
[-0.146]
- -0.379
(0.000)
[-0.150]
- -0.378
(0.000)
[-0.150]
𝛼5 Log peer sales – Log firm
sales if peer sales > firm
sales, or=0
-0.965
(0.000)
[-0.371]
-1.115
(0.000)
[-0.423]
-0.962
(0.000)
[-0.369]
-1.114
(0.000)
[0.422]
-0.964
(0.000)
[-0.370]
-1.117
(0.000)
[0.424]
𝛼6 Log peer sales – Log firm
sales when peer sales <
firm sales, or=0
2.541
(0.000)
[0.796]
2.110
(0.000)
[0.709]
2.546
(0.000)
[0.797]
2.114
(0.000)
[0.709]
2.542
(0.000)
[0.796]
2.113
(0.000)
[0.709]
𝛼7 Peer ROA – firm ROA
when peer ROA > firm
ROA, =0 otherwise
-0.098
(0.000)
[-0.039]
-0.202
(0.000)
[-0.080]
-0.092
(0.000)
[-0.037]
-0.200
(0.000)
[-0.080]
-0.091
(0.000)
[-0.036]
-0.198
(0.000)
[-0.079]
𝛼8 Peer ROA – firm ROA
when peer ROA < firm
ROA, =0 otherwise
0.117
(0.000)
[0.047]
0.070
(0.000)
[0.028]
0.127
(0.000)
[0.051]
0.076
(0.000)
[0.030]
0.131
(0.000)
[0.052]
0.078
(0.000)
[0.031]
𝛼9 Peer MTB – firm MTB
when peer MTB > firm
MTB, =0 otherwise
-0.105
(0.000)
[-0.042]
-0.068
(0.000)
[-0.027]
-0.106
(0.000)
[-0.042]
-0.070
(0.000)
[-0.028]
-0.105
(0.000)
[-0.042]
-0.070
(0.000)
[-0.028]
𝛼10 Peer MTB – firm MTB
when peer MTB < firm
MTB, =0 otherwise
1.065
(0.000)
[0.406]
0.793
(0.000)
[0.308]
1.067
(0.000)
[0.406]
0.791
(0.000)
[0.308]
1.067
(0.000)
[0.406]
0.792
(0.000)
[0.308]
Table 5 (continued)
Dependent variable is one if a potential peer-CEO is chosen as a peer by the
sample firm and zero otherwise
Regression Model (1) Regression Model (2) Regression Model (3)
SP500 Non-SP500 SP500 Non-SP500 SP500 Non-SP500
CEO characteristics
𝛼11 Peer GAI – firm GAI
when peer GAI > firm
GAI, =0 otherwise
0.041
(0.000)
[0.016]
-0.009
(0.220)
[-0.004]
𝛼12 Peer GAI – firm GAI
when peer GAI < firm
GAI, =0 otherwise
0.107
(0.000)
[0.043]
0.199
(0.000)
[0.079]
𝛼13 Dummy equal to one if a
specialist CEO is chosen
for a generalist CEO, =0
otherwise
-0.221
(0.000)
[-0.088]
-0.205
(0.000)
[-0.082]
𝛼14 Dummy equal to one if a
generalist CEO is chosen
for a specialist CEO, =0
otherwise
0.119
(0.000)
[0.047]
0.064
(0.000)
[0.026]
𝛼15 Peer-CEO’s positions –
firm-CEO’s positions
when the difference is
positive, =0 otherwise
0.002
(0.788)
[0.001]
0.017
(0.026)
[0.007]
𝛼16 Peer-CEO’s positions –
firm-CEO’s positions
when the difference is
negative, =0 otherwise
0.117
(0.000)
[0.047]
0.184
(0.000)
[0.073]
𝛼17 Peer-CEO’s firms – firm-
CEO’s firms when the
difference is positive, =0
otherwise
-0.034
(0.000)
[-0.014]
-0.018
(0.000)
[-0.007]
𝛼18 Peer-CEO’s firms – firm-
CEO’s firms when the
difference is negative, =0
otherwise
0.033
(0.024)
[0.013]
0.097
(0.024)
[0.039]
𝛼19 Peer-CEO’s industries –
firm-CEO’s industries
when the difference is
positive, =0 otherwise
0.061
(0.000)
[0.024]
0.011
(0.150)
[0.004]
𝛼20 Peer-CEO’s industries –
firm-CEO’s industries
when the difference is
negative, =0 otherwise
0.055
(0.000)
[0.022]
0.016
(0.206)
[0.006]
Table 5 (continued)
Dependent variable is one if a potential peer-CEO is chosen as a peer by the
sample firm and zero otherwise
Regression Model (1) Regression Model (2) Regression Model (3)
SP500 Non-SP500 SP500 Non-SP500 SP500 Non-
SP500
CEO characteristics
𝛼21 Dummy equal to one if
both have conglomerate
experience, =0 otherwise
0.059
(0.000)
[0.024]
0.020
(0.232)
[0.008]
𝛼22 Dummy equal to one if
both no conglomerate
experience, =0 otherwise
-0.007
(0.615)
[-0.003]
0.112
(0.000)
[0.045]
𝛼23 Dummy equal to one if
both have previous
experience as a CEO, =0
otherwise
-0.060
(0.099)
[0.024]
-0.052
(0.169)
[-0.021]
𝛼24 Dummy equal to one if
both don’t have previous
experience as a CEO, =0
otherwise
0.081
(0.000)
[0.032]
0.127
(0.000)
[0.051]
𝛼25 Peer-CEO’s age – firm-
CEO’s age when the
difference is positive, =0
otherwise
-0.090
(0.000)
[-0.036]
-0.099
(0.000)
[-0.039]
-0.097
(0.000)
[-0.039]
-0.100
(0.000)
[-0.040]
-0.089
(0.000)
[-0.036]
-0.101
(0.000)
[-0.040]
𝛼26 Peer-CEO’s age – firm-
CEO’s age when the
difference is negative, =0
otherwise
0.023
(0.006)
[0.009]
0.070
(0.000)
[0.028]
0.026
(0.002)
[0.010]
0.073
(0.000)
[0.029]
0.032
(0.002)
[0.013]
0.076
(0.000)
[0.030]
𝛼27 Dummy equal to one if
both hold chairman
positions, =0 otherwise
0.259
(0.000)
[0.103]
0.180
(0.000)
[0.072]
0.273
(0.000)
[0.109]
0.196
(0.000)
[0.078]
0.270
(0.000)
[0.107]
0.193
(0.000)
[0.077]
𝛼28 Dummy equal to one if
both hold chairman
positions, =0 otherwise
-0.034
(0.041)
[-0.014]
-0.001
(0.925)
[-0.000]
-0.043
(0.010)
[-0.017]
-0.014
(0.364)
[-0.006]
-0.034
(0.038)
[-0.014]
-0.007
(0.642)
[-0.003]
𝛼29 Dummy equal to one if
both have MBA degrees,
=0 otherwise
0.104
(0.000)
[0.041]
-0.003
(0.888)
[-0.001]
0.108
(0.000)
[0.043]
-0.003
(0.863)
[-0.001]
0.100
(0.000)
[0.040]
-0.006
(0.763)
[-0.002]
𝛼30 Dummy equal to one if
both attended Ivy League
schools, =0 otherwise
0.050
(0.019)
[0.020]
0.118
(0.000)
[0.047]
0.051
(0.016)
[0.020]
0.118
(0.000)
[0.047]
0.048
(0.022)
[0.019]
0.121
(0.000)
[0.048]
Table 6
Firm CEO and the Median of Peer Group
The sign test for the difference in firm size, firm performance, and CEO experience between sample CEO and the
median of peer group. ***, ** indicate significance at 1% and 5% confidence levels. The three columns report the
difference for all sample firms, S&P 500 sample firms, and non-S&P 500 sample firms respectively. Firm size is
measured by log of sales revenue. Firm performance is measured by ROA. CEO experience variables includes general
ability index (GAI), worked positions, worked firms, and worked industries. All variables are normalized.
Peer Median Minus Sample
CEO
(Non-S&P 500 firms)
Peer Median Minus Sample
CEO
(S&P 500 firms)
Peer Median Minus Sample
CEO
(All sample firms)
Size(log sale)
ROA
GAI
Positions
Firms
Industries
0.088***
0.001
0.104***
1.000***
0.000**
0.000
0.013
0.002***
-0.024
0.000
0.000
-0.000
0.050***
0.001***
0.042***
0.500***
0.000**
-0.000