Customer Concentration and Managerial Agency Costs * Taeyeon Kim Korea Advanced Institute of Science and Technology (KAIST) [email protected]Hyun-Dong Kim † Sogang University [email protected]Kwangwoo Park Korea Advanced Institute of Science and Technology (KAIST) [email protected]This version: February 15, 2019 * This research was supported by the Sogang University Research Grant of 2018 (201810007.01). † Corresponding author: Professor of International Finance, Graduate School of International Studies, Sogang University; 35 Baekbeom-ro, Mapo-gu, Seoul 04107, South Korea; Tel: +82-2-705-8682; Email: [email protected]
53
Embed
Customer Concentration and Managerial Agency Costs · 2019. 4. 24. · Customer Concentration and Managerial Agency Costs* Taeyeon Kim Korea Advanced Institute of Science and Technology
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Customer Concentration and Managerial Agency Costs*
Taeyeon Kim
Korea Advanced Institute of Science and Technology (KAIST)
that a firm needs to make a commitment not to act opportunistically to exploit its
counterparties’ quasi-rents.3 This idea is consistent with the view of Johnson, Karpoff, and Yi
(2015) that a supplier’s commitment is important for maintaining a supplier–customer
relationship. In the relationship between customer-dependent suppliers and major customers,
implicit contracts to engage in this type of firm commitment appear to be valuable in
3 Quasi-rents occur when a counterparty engages in a relationship-specific investment that may lose value if the
firm adjusts its policies and decisions (Johnson et al., 2015).
4
resolving hold-up problems (Gillan, Hartzell, and Parrino, 2009; Johnson et al., 2015).4 Such
implicit contracts are informally implemented by managers’ personal connections and
reputations with their major counterparties (Klein and Leffler, 1981). Johnson et al. (2015)
note that CEOs’ personal connections and reputations at customer-dependent suppliers are
essential in retaining long-term relationships with customers.5
In line with this thought, managers of customer-dependent suppliers are hired for their
personal commitments to their customers (Shleifer and Summers, 1988; Intintoli et al., 2017).
Such connections and reputations make it possible that suppliers do not behave themselves
like opportunists to exploit their customers’ quasi-rents (Johnson et al., 2015). Accordingly,
personal commitments encourage customers to consistently engage in relationship-specific
investment. However, if the managers are removed, existing personal commitments are no
longer effective, and suppliers fail to retain their stable relationship with customers (Shleifer
and Summers, 1988). In a similar vein, Johnson et al. (2015) suggest that takeover defenses
become important in maintaining implicit commitments because they decrease the likelihood
of managerial replacement. Intintoli et al. (2017) further show that the replacement of
managers disrupting supplier–customer relationships has a negative effect on the financial
performance of suppliers. Taken together, incumbent CEOs of customer-dependent suppliers
who have personal connections and reputations are most likely to continue close relationships
with major customers. This implies that replacing these CEOs is costly to suppliers with
concentrated customers.
Since the seminal work by Jensen and Meckling (1976), extensive literature argues that
4 In the classical Fisher Body–General Motors example described in Klein, Crawford and Alchian (1978),
managers of Fisher Body (supplier) could promise not to increase prices to appropriate General Motors’
(customer) quasi-rents. See Klein et al. (1978), Johnson et al. (2015), and Gillan et al. (2009) for more detailed
descriptions of the Fisher Body–General Motors example. 5 Johnson et al. (2015) provide an example of the Pemstar–International Business Machines case in their study.
See Johnson et al. (2015) for more detailed descriptions on the case.
5
an agency conflict between a principal (a shareholder) and an agency (a manager) occurs due
to an irreversible investment from their unique relationship (Williamson, 1975; Klein et al.,
1978; Grossman and Hart, 1986). In particular, Shleifer and Vishny (1989) model how
managers can entrench themselves by engaging in manager-specific investments. Managers
have an incentive to engage in businesses associated with their skills and expertise. Specially,
managers focus on investing in assets specific to their skills, and the value of such assets
would be higher than if they were controlled by alternative managers without the required set
of skills. Such investments will make themselves valuable to shareholders and they are less
likely to be replaced even after a poor performance. As a result, managers can be entrenched
and pursue perquisites by wasting free cash flow.
Returning to the issue of dependent suppliers on concentrated customers, we conjecture
that CEOs having implicit commitments with concentrated customers are more likely to make
manager-specific investments associated with supplier–customer relationships. So, replacing
CEOs of customer-dependent suppliers imposes a substantial cost to these firms. Hence,
CEOs with major customers can entrench themselves by engaging in excessive investments
in their own specific assets. In this regard, managerial agency problems resulting from
manager-specific investments appear to be significantly prevalent in suppliers with highly
concentrated customers.
As an example of agency costs in practice, extensive studies discuss that managers waste
corporate resources by using the resources to their own devices. To capture the inefficient
uses of corporate resources and the possible value destruction resulting from agency
problems, we focus on cash reserves.6 Firms should hold a certain amount of cash to pay out
6 Three reasons are suggested by extant literature for using cash reserves as a proxy for agency costs (e.g.,
Dittmar and Mahrt-Smith, 2007; Frésard and Salva, 2010; Faulkender and Wang, 2006). First, managers can
easily access cash reserves with little monitoring, and also have much discretion on their use. Thus, cash may
provide managers with resources to invest in non-positive net present value (NPV) projects, destroying
6
funds for day-by-day operation, and to provide a buffer against unexpected events or the cost
of external financing for their investments. However, if cash reserves are held in excess of the
amount committed for operations and investments, they can be exploited as resources for
managers’ private benefits (Myers and Rajan, 1998). In this case, holding excessive cash may
harm firm value, suggesting that the market value of one dollar cash reserves may not be
equivalent to the value of a dollar. To justify this thought, Dittmar and Mahrt-Smith (2007)
and Frésard and Salva (2010) examine how investors value excess cash holdings of a firm
with weak governance. When investors recognize that managers may use cash inefficiently,
the market value of those resources is discounted. We thus hypothesize that excess cash
holdings of suppliers with more concentrated customers will be valued lower by investors
because managerial agency problems are expected to be more severe for such suppliers.
Using a comprehensive sample of U.S. firms spanning 1977 to 2016, we examine the
relationship between suppliers’ customer concentration and their managerial agency costs.
We find that a higher level of customer concentration is related to a lower market value of
suppliers’ excess cash, implying that managerial agency problems are expected to be more
severe in suppliers with concentrated customers. Our baseline results may be subject to
endogeneity concerns related to measurement errors, omitted variable bias, and reverse
causality. Hence, we conduct various tests designed to mitigate potential endogeneity issues.
Our results are qualitatively similar when we employ alternative measures for the market
value of excess cash and also control for time-invariant omitted CEO and firm characteristics,
as well as corporate governance. Moreover, the results still hold when a propensity score
matching procedure and two-stage least square regressions are run.
shareholders’ wealth. Second, firms reserve substantial amounts of cash, and the value of cash holdings accounts
for a significant proportion of their wealth. Third, while a supplier–customer level is quite sticky, the degree of
cash holdings substantially varies over time. This variation in cash provides us with an optimal setting to test the
effect of customer concentration on the value of cash in suppliers.
7
In addition, we investigate how customer concentration affects CEO compensation and
acquisition decisions to confirm the existence of agency problems. Our results show that
managers in suppliers with higher customer concentration receive greater compensation than
vice versa. Such suppliers also experience lower abnormal stock returns after a mergers and
acquisitions (M&A) announcement upon acquiring a target firm. Thus, when a business
relationship with customers is more important, managers extract higher benefits from
shareholders, in the form of receiving a higher compensation and investing in value-
destroying deals.
Next, we examine the possible channel through which higher customer concentration
leads to managerial agency costs of suppliers. Because supplier–customer relationships
become more important for suppliers who highly depend on a few major customers, these
suppliers are likely to hire managers who are better able to retain customer relationships.
Thus, supplier–customer relationships are most valuable under current managers and
replacing managers is very costly to suppliers with a highly concentrated customer base. As
expected, we find that customer-related CEOs are more prevalent in suppliers with higher,
than lower, customer concentration. Our finding also shows that such managers are less
forced out, allowing them greater job security.
We further identify potential circumstances wherein customer concentration is more
closely associated with managerial agency problems. We find that the negative relationship
between customer concentration and the market value of excess cash is more pronounced in
suppliers that are dependent on major customers who can easily switch their suppliers. This
result shows that the negative market value of excess cash from customer concentration
occurs particularly under circumstances wherein the management of supplier–customer
relationships and the role of the supplier’s manager are more important.
8
Our study contributes to the extant literature in numerous ways. First, to the best of our
knowledge, we are the first to directly examine managerial agency costs of suppliers that
result from their concentrated customer base. Prior studies investigate the governance role of
customers in suppliers, and postulate that major customers have incentives to monitor the
managers of suppliers (Wang, 2012; Johnson et al., 2015; Kang, Liu, Yi, and Zhang, 2015).
Shifting our attention from the literature on customers’ role in monitoring supplier managers,
we focus on managerial agency conflicts with shareholders in customer-dependent suppliers
and suggest a mechanism by which managerial agency problems occur.7 Second, our study
adds to the literature that investigates management entrenchment due to manager-specific
investments (e.g., Shleifer and Vishny, 1989). Our findings show that managers for suppliers
with major customers can entrench themselves by engaging in specific investments
associated with supplier–customer relationships. Third, our findings also contribute to the
existing literature on the value of cash holdings (e.g., Dittmar and Mahrt-Smith, 2007;
Faulkender and Wang, 2006). These extant studies mainly address how corporate governance
and corporate financial policy affect the value of cash. We further show that the marginal
value of excess cash declines with higher customer concentration.
The rest of the paper is organized as follows. Section 2 describes the sample and variables
used in this study, and provides the descriptive statistics. Next, section 3 discusses the
where Market Value/Assetsi,t is the market value of supplier i scaled by total assets at time t,
which is calculated as the sum of market value of equity and book value of short- and long-
term debt divided by total assets. Customer Concentrationi,t and Xcashi,t are supplier i’s
customer concentration and excess cash holdings in period t, respectively.
Following Fama and French (1998), we control for suppliers’ earnings, research and
development (R&D) expenditures, interest expenses, and dividend payouts. To control for
investors’ expectation, we also include two-year lagged (from year t-2 to year t) and forward
(from year t to year t+2) changes of these control variables, as well as two-year forward
changes in the market value of the firm. All controls are normalized by total assets, and year
and industry fixed effects are included in the regressions.
2.4 Descriptive Statistics
Panel A of Table 1 presents the descriptive statistics for our variables. The mean value of
Major Customer is 0.345, which indicates that 34.5% of suppliers in our sample have at least
one major customer. The average supplier’s sales to major customers account for 13.9% of
total sales, and the mean value of Customer HHI is 0.047. Suppliers with at least one major
customer sell 40.2% of their total sales to those customers on average, and the mean value of
their Customer HHI is 0.136. The mean excess cash held by suppliers is 0.867, and cash
reserves account for 19.8% of total assets on average. The mean total assets and net assets of
13
suppliers are 2,261 and 1,975 million dollars, respectively. On average, the market value of
the firm to total assets is 1.159, and earnings ratio to total assets is 0.087. The average R&D
expenditure and interest expense amount to 3.8% and 2.7% of total assets, respectively.
Moreover, the mean value of dividend ratio is 0.010. To mitigate the outlier effect, all
continuous variables are winsorized at the 1% level.
[Insert Table 1 here]
We conduct univariate tests to obtain preliminary insights on the relationship between
customer concentration and excess cash. In Panel B of Table 1, we split the sample into two
groups according to whether or not suppliers disclose at least one major customer. We
perform the difference-in-mean and difference-in-median tests between these two groups.
Suppliers with at least one major customer hold more excess cash than those without a major
customer. Specifically, the mean value of excess cash holdings is 0.899 for suppliers that
have at least one major customer, whereas it is 0.849 for those with no major customer. The
difference is statistically significant at the 1% confidence level. The results of the univariate
tests suggest that agency problems are more severe in suppliers with more concentrated
customers.9 In addition, suppliers with at least one major customer have smaller total assets
and net assets; higher cash, market value, and R&D expenditure; and lower interest expense
and dividend payouts.10 These results are consistent with the extant literature.11
9 Since suppliers with more concentrated customers are likely to face high risks (Itzkowitz, 2013; Bae and
Wang, 2015; Dhaliwal et al., 2016; Campello and Gao, 2017), they may hold more cash reserves. To focus on
the agency problem, rather than the precautionary purpose of cash holdings, we investigate the market value of
excess cash holdings using positive excess cash, as explained in section 2.3. 10 Campello and Gao (2017) find that higher customer concentration increases interest rate spreads since the
risk of firms with concentrated customers is higher than the risk for other firms. Such firms suffer from higher
costs of debt and may face difficulties in raising capital from debt financing, thus yielding lower leverage and
interest expense. 11 See Wang (2012), Itzkowitz (2013), Dhaliwal et al. (2016), Campello and Gao (2017), and Krolikowski and
Yuan (2017).
14
3. Baseline Regression Analysis
To investigate the effect of customer concentration on the market value of excess cash
holdings, we estimate equation (3) by using ordinary least square (OLS) regressions. Table 2
presents our baseline results. In column (1), we use Major Customer as a customer
concentration measure, and the coefficient on excess cash (Xcash), -0.253, is significantly
negative at the 1% level. Given suppliers with positive excess cash are only used in our
analysis, negative coefficient on excess cash appears to be reasonable because investors may
be concerned about excessive cash reserves that can potentially result in an agency problem.
The coefficient on interaction term between customer concentration and excess cash (Major
Customer*Xcash here), our main variable of interest, is -0.116, with statistical significance at
the 1% level. This result suggests that the market value of excess cash is lower in suppliers
with at least one major customer than in those with no major customer. In the perspective of
economic significance, the market value of suppliers with at least one major customer
decreases by $ 0.369 (=0.253+0.116) as they hold an additional one dollar of excess cash,
whereas the market value of those with no major customer reduces by $0.253 with an
additional one dollar of excess cash.
[Insert Table 2 here]
Columns (2) and (3) of Table 2 present the regression results using Major Customer Sales
and Customer HHI as the measures of customer concentration, respectively. The coefficients
on Xcash and Customer Concentration*Xcash are significantly negative at the 1% level.
Given that the means of Major Customer Sales and Customer HHI are 0.139 and 0.047 in
Panel A of Table 1, respectively, the market value of an additional one dollar on excess cash
is -$ 0.298 (=-0.249-0.354*0.139) and -$0.300 (=-0.259-0.875*0.047), respectively. If Major
Customer Sales and Customer HHI increase by one standard deviation (0.379 for Major
15
Customer Sales and 0.160 for Customer HHI in Panel A of Table 1), the market value of an
additional one dollar of excess cash is -$ 0.383 (=-0.249-0.354*0.379) and -$ 0.399 (=-0.259-
0.875*0.160), respectively. Accordingly, the results of columns (2) and (3) show that the one-
standard-deviation increases in Major Customer Sales and Customer HHI lead to around 30%
drop in the market value of excess cash. These results are qualitatively similar when we re-
estimate equation (3) using a subset of suppliers that report at least one major customer in
columns (4) and (5). Overall, our results of Table 2 show that the negative impact of customer
concentration on the market value of excess cash holdings is economically significant,
thereby providing empirical evidence on the presence of managerial agency problems in
suppliers with a concentrated customer base.
4. Endogeneity Issue
In section 3, we have shown a negative relationship between customer concentration and
the market value of excess cash holdings. However, our results might be subject to various
types of endogeneity problems, such as measurement error, omitted variable bias, and reverse
causality. Although we measure our main variables and control for other determinants by
following prior literature, a measurement error and an omitted variables bias can still
influence both customer concentration and the market value of excess cash holdings. Such
issues would make our observed relationship suspicious. Because exogenous variations in
customer concentration measures appear to be insufficient, reverse causality may also arise in
the baseline results. Thus, we perform additional tests to mitigate endogeneity concerns.
4.1 Alternative Measures for the Market Value of Excess Cash Holdings
In Table 3, we address issues on measurement errors by employing the market value of
16
cash ratio and the forecasted market value of excess cash holdings, instead of the market
value of excess cash holdings. We re-estimate our baseline specification in Panel A of Table 3
using cash ratio. In columns (1)–(3), the coefficients on Cash Ratio are significantly positive.
Given that cash holdings can be used as buffers for new investments or against financial
constraints, the positive sign seems to be reasonable. Furthermore, the coefficients on the
interaction term between customer concentration and cash ratio are negative, with statistical
significance at the 1% level. These results are consistent with our baseline results in Table 2,
indicating that the market value of cash ratio is lower for suppliers with more concentrated
customers.
[Insert Table 3 here]
In our normal cash regressions used to calculate excess cash, we control for the market
value of assets normalized by total assets, but this variable is included again as a dependent
variable in equation (3). This measurement process may lead to endogeneity related to the
market value of assets. To address endogeneity, we use the forecasted market value of assets.
The market value of assets forecasted by analysts is measured as the sum of the forecasted
market value of equity and the book value of short- and long-term debt. In addition,
forecasted market value of equity is defined as the product of the common shares outstanding
and the average target price predicted by analysts. To calculate the average target stock price,
we use analysts’ prediction, whose forecast horizon is 12 months; the prediction is announced
within three months before and after the fiscal year end. Because analysts are more informed
than other types of investors, their predicted market value of excess cash would better reflect
governance issues, including agency problems.
In Panel B of Table 3, we re-run the regressions with the forecasted market value
normalized by total assets. The coefficients on the interaction term between customer
17
concentration and excess cash remain significantly negative in all columns, suggesting that
analysts poorly evaluate excess cash holdings of suppliers that depend on a few major
customers. Taken together, our main results hold unaffected even when we use alternative
measures for the market value of excess cash holdings.
4.2 Propensity Score Matched Sample Analysis
We have so far controlled for various firm-level characteristics, but our estimation may
still suffer from omitted variables that correlate with both customer concentration and the
market value of excess cash holdings. To address any omitted variable bias, we employ a
propensity score matching procedure to enable a closer comparison between firms sharing
similar characteristics in all respects, with the only exception of customer concentration.
Using a logit regression model, we first regress Major Customer on our control variables
used in equation (3), and estimate the propensity score—that is, the probability that a supplier
has at least one major customer. Based on a nearest-neighbor propensity score matching
procedure, we match each supplier with at least one major customer to a supplier with no
major customer, but with the closest propensity score. Finally, we construct a smaller
subsample of suppliers with at least one major customer and matched suppliers with no major
customer. We then run our main regression again for this subsample in Table 4.
[Insert Table 4 here]
The results using a subsample obtained from propensity score matching are comparable to
our baseline result in Table 2. The coefficients on Xcash’s interaction terms with customer
concentration are significantly negative in columns (1)–(3). This is clear indication that our
earlier finding is not likely to be a by-product of omitted variable issues.
18
4.3 CEO and Firm Fixed Effects Regressions
To additionally alleviate any omitted variable bias, we re-estimate baseline regressions
with CEO and fixed effects in Table 5. Columns (1)–(3) report the results including the CEO
fixed effects, and the other columns present those with firm fixed effects regression.12 All
coefficients on interaction terms between customer concentration and excess cash are
negative, with statistical significance at the 1% level. These results suggest that omitted
variables related to CEO or firm characteristics are not likely to drive our baseline finding.
[Insert Table 5 here]
4.4 Two-stage Least Squares Regressions
Furthermore, we run two-stage least squares (2SLS) regressions with instrumental
variables in order to mitigate primary endogeneity issues, including omitted variable bias and
reverse causality. Particularly, we address endogeneity concerns that may arise because the
exogenous variations in our customer concentration variables are not sufficient. An
instrumental variable used in our 2SLS regressions should capture a variation in customer
concentration (i.e., inclusion restriction), but be exogenous to the market value of assets (i.e.,
exclusion restriction). As in Campello and Gao (2017), we employ M&A activities in major
customers’ industries (horizontal M&A), denoted as Customer M&A, as an instrumental
variable. Extant studies (e.g., Fee and Thomas, 2004; Bhattacharyya and Nain, 2011;
Campello and Gao, 2017) show that suppliers tend to engage in business with more
concentrated customers after M&A waves in their customers’ industries because the number
of customers decreases by their horizontal M&A. Therefore, M&A activities in major
12 We use ExecuComp database to retrieve information on CEOs. Because ExecuComp database starts from
1993, our sample substantially decreases in the regressions with CEO fixed effects.
19
customers’ industries are positively associated with the level of customer concentration,
which satisfies the inclusion restriction. On the other hand, Customer M&A appears to affect
suppliers’ market value of assets only through the supplier–customer link—that is, a shock of
customers’ industries is unlikely to directly influence supplier characteristics, which satisfies
the exclusion restriction.
For the M&A data, we focus on observations with the deals that are initiated and
completed by acquirers in the same two-digit SIC code as their targets. We use the deal value
of M&A normalized by the acquirer’s total sales as a proxy for acquisition activity. Then, the
extent of industry-level acquisition is defined as the average acquisition activities of firms in
a given industry over the past five years. The five-year window is used to prevent a few big
deals from driving the total deal value. Next, we manually match the reported major
customers’ name with the historical company name listed on the Compustat database.13,14
Finally, after matching major customers’ industries with industry-level acquisition activities,
we measure Customer M&A as the weighted sum of acquisition activities across industries in
which the supplier’s major customers operate.15 To meet the exclusion condition, we exclude
the observations that a supplier operates in the same industries with its customers.
Our main regressions include two potential endogenous variables, Customer
Concentration and Customer Concentration*Xcash. Thus, we run two first-stage regressions
with two instrumental variables, Customer M&A and Customer M&A*Xcash, following the
methodology of Benmelech and Frydman (2015). Our two instrumental variables are
included in each first-stage regression. In addition, suppliers that report their major customers
13 While the Compustat segment customer file provides suppliers’ identification number, customers’
identification numbers are not available. 14 Our sample significantly reduces because we use only suppliers whose major customers are matched with
Compustat’s historical company data. 15 Since a supplier may have major customers operating in different industries, we use the weight that is defined
as the supplier’s percentage sales to each major customer.
20
are only useful because we should know the major customers’ names to measure Customer
M&A. Under this construction rule, a sample of Major Customer equal to one only remains,
making Major Customer useless in the 2SLS regression. We therefore employ Major
Customer Sales and Customer HHI as customer concentration variables. Table 6 reports the
results on 2SLS regressions.
[Insert Table 6 here]
Columns (1)–(4) present the results of first-stage regressions, and columns (5) and (6)
report those of the second-stage regressions. Major Customer Sales is used in columns (1),
(2), and (5), while Customer HHI is used in columns (3), (4), and (6). Major Customer Sales
and Customer HHI are included in the first-stage regression as dependent variables in
columns (1) and (3), respectively. In addition, Major Customer Sales*Xcash and Customer
HHI*Xcash are used as dependent variables in columns (2) and (4), respectively. On the other
hand, the market value of assets normalized by total assets is included in the second-stage
regressions as a dependent variable. The coefficients on Customer M&A (Customer
M&A*Xcash) in columns (1) and (3) (columns (2) and (4)) are significantly positive. The F-
statistics of the first-stage regressions are also sizable, exceeding the rule-of-thumb value of
10 for the weak instrument test. In the second-stage regressions, the coefficients on the
interaction terms between customer concentration and excess cash (Major Customer
Sales*Xcash and Customer HHI*Xcash) are negative, with statistical significance at the 1%
level. The results on the F-statistics obtained from the Wu–Hausman test also confirm that
our customer concentration measures are exogenous by themselves. In short, our main
finding, namely, the negative impact of customer concentration on managerial agency
problems, remains significant, suggesting that the result is not likely to be driven by
endogeneity issues
21
5. Further Robustness Tests
5.1 Alternative Measures of Managerial Agency Costs
We have far argued that excess cash held by suppliers with more customer concentration
is poorly valued by the capital market because of the potential agency problem. In this
section, we perform additional robustness tests to confirm the presence of an agency problem
for suppliers with a concentrated customer base.
Following Masulis, Wang and Xie (2009) and Chen, Harford and Lin (2015), we first
focus on CEO compensation. Since CEO compensation can be the most direct way to shift
shareholders’ wealth to managers, a rich body of literature on corporate governance considers
a higher level of CEO compensation, relative to comparable firms, as a by-product of a
managerial agency problem (e.g., Bertrand and Mullainathan, 1999; Core, Holthausen, and
Larcker, 1999; Masulis et al., 2009; Chen et al., 2015). Thus, we expect that managers of
suppliers with a highly concentrated customer base will receive a higher level of
compensation customer concentration.
Panel A of Table 7 shows the regression results for the effect of customer concentration
on CEO compensation. We employ two different CEO compensation variables, ln(CEO Total
Compensation) and CEO Excess Compensation, as dependent variables. ln(CEO Total
Compensation) is the natural logarithm of total annual compensation of CEO, while CEO
Excess Compensation is the residuals from the regression of the natural logarithm of CEO
total compensation on the natural logarithm of total market value of the firm. Following
Masulis et al. (2009), firm size (ln(Assets)), Tobin’s q (Tobin’s Q), return on assets (ROA),