CEO compensation and corporate social responsibility

Accepted Manuscript

Title: CEO Compensation and Corporate Social Responsibility

Author: Ming Jian Kin-Wai Lee

PII: S1042-444X(14)00059-0DOI: http://dx.doi.org/doi:10.1016/j.mulfin.2014.11.004Reference: MULFIN 462

To appear in: J. of Multi. Fin. Manag.

Received date: 15-10-2014Accepted date: 25-11-2014

Please cite this article as: Jian, M., Lee, K.-W.,CEO Compensation and CorporateSocial Responsibility, Journal of Multinational Financial Management (2014),http://dx.doi.org/10.1016/j.mulfin.2014.11.004

This is a PDF file of an unedited manuscript that has been accepted for publication.As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain.

http://dx.doi.org/doi:10.1016/j.mulfin.2014.11.004

http://dx.doi.org/10.1016/j.mulfin.2014.11.004

Page 1 of 49

Accep

ted

Man

uscr

ipt

CEO Compensation and Corporate Social Responsibility

Highlights

CEO compensation is negatively associated with CSR, especially in better governed

firms.

We decompose CSR investment into a normal component and an abnormal

component.

CEO compensation is positively (negatively) associated with normal (abnormal) CSR.

Thus, CEOs receive lower compensation level for excessive investment in CSR.

Better governed firms penalize abnormal CSR more than poorly governed firms.

Highlights (for review)

Page 2 of 49

Accep

ted

Man

uscr

ipt


Ming Jian

Nanyang Technological University

email: [email protected]

Kin-Wai Lee*

S3-B2A-19 Nanyang Avenue

Nanyang Business School

Nanyang Technological University

Singapore 639798

email : [email protected]

Tel : (65) 6790-4663

Fax : (65) 6792-4217

Version : 14 October 2014

*Corresponding author.

1) Title Page (WITH Author Details)

mailto:[email protected]

mailto:[email protected]

Page 3 of 49

Accep

ted

Man

uscr

ipt

1


14 October 2014

Abstract

We examine the association between CEO compensation and corporate social responsibility

(CSR). We find that CEO compensation is negatively associated with CSR investment. We find

CEO compensation is positively associated with normal CSR, suggesting that CEO is rewarded

for investing in optimal level of CSR. The positive association between CEO compensation and

normal CSR is more pronounced in firms with stronger corporate governance. However, CEO

compensation level is negatively associated with abnormal CSR, suggesting that when CSR

investment deviates from its optimal level, CEOs receive lower compensation level for excessive

CSR investments. Firms with good corporate governance penalize abnormal CSR.

Keywords: corporate social responsibility (CSR), CEO compensation, corporate governance

JEL classifications: G30, G34, M1, M2

*2) Blinded Manuscript (WITHOUT Author Details)

http://ees.elsevier.com/mulfin/viewRCResults.aspx?pdf=1&docID=819&rev=0&fileID=11533&msid=9B618104-0E50-47CB-AE19-DD69145F47A2

Page 4 of 49

Accep

ted

Man

uscr

ipt

2

1. Introduction

Conventional wisdom suggests that firms should reward CEO for undertaking corporate

social responsibility (CSR) which improves firm performance. Corporate governance and CSR

advocates such as Global Reporting Initiatives and Corporate Register recommend that

compensation of top level management should reflect CSR. For instance, the 2013 joint report by

the Investor Responsibility Research Center and the Sustainable Investments Institutes suggest

that 43% of the Fortune 500 firms tie executive compensation to CSR.1 The overall empirical

evidence on the association between CEO compensation and CSR is inconclusive. On one hand,

Berrone and Gomez-Mejia (2009) find that in polluting industries, good environmental

performance increases CEO compensation2. On the other hand, contrary to conventional wisdom,

other studies find that CEO compensation level is negatively associated with CSR (Coombs and

Gilley, 2005; Russo and Harrison, 2005; Stanwick and Stanwick, 2001).

While the preceding studies differ in sample, time period and method, they do not

explicitly consider the heterogeneity in CSR investments. For example, Borghesi et al (2014,

page 164) find that “in some instances, CSR investments enhance shareholder value. However,

in other cases, altruistic managers or managers who privately benefit from the positive attention

arising from these activities may choose to make CSR investments even if they are not value

enhancing.” By introducing the concepts of normal (value increasing) CSR and abnormal (value

decreasing) CSR, our paper revisits the association between CEO compensation and CSR

investment and aims to provide new insights to the puzzling findings of the previous papers.

We posit that the association between CEO compensation and CSR depends on whether

CSR investments are value increasing or value decreasing. Furthermore, it is plausible that

1 http://www.csrhub.com/blog/2013/05/top-companies-tie-compensation-to-sustainability.html

2 The authors focus on firms from industries subject to reporting under the Environmental Protection Agency‟s

Toxics Release Inventory, a program that requires facilities exceeding a threshold level to report their emissions.

Page 5 of 49

Accep

ted

Man

uscr

ipt

3

association between CEO compensation and CSR may vary systematically across different

corporate governance structures. Accordingly, we examine whether there is an interplay between

corporate governance structures and CSR investments in affecting CEO compensation.

We begin our analysis by considering two alternative views on the association between

CEO compensation and CSR. Under the first view, greater CSR investment enhances

shareholders‟ value because better CSR investment is associated with better retention of high

quality employees (Greening and Turban, 2000), higher demand for the firm‟s products

(Navarro, 1988), higher customer loyalty (Maignan et al., 1999; Sen and Bhattacharya, 2001)

and higher access to valuable resources (Cochran and Wood, 1984). Other studies find that CSR

is associated with better non-financial performance such as higher operational efficiencies

(Sharma and Vredenburg, 1998) and higher product quality (Johnson and Greening, 1999).

These studies draw extensively from the stakeholder value maximization theory (Cornell and

Shapiro, 1987; Hill and Jones, 1992; Jensen and Meckling, 1976), the theme of which is that a

firm is a nexus of contracts between shareholders and other stakeholders (such as customers,

suppliers and employees). Each group of stakeholders supplies the firm with critical resources in

exchange for claims outlined in explicit contracts (e.g., wage contracts and product warranties)

or suggested in implicit contracts (e.g., promises of job security to employees and continued

service to customers). If higher CSR investment is associated with greater firm-specific focus on

the interests of other stakeholders (such as customers, suppliers and employees), these

stakeholders are more likely to support the firm‟s operation, which increases shareholders‟ value.

Stated differently, CSR activities have a positive effect on shareholders‟ value because focusing

on the interests of other stakeholders increases their willingness to support a firm‟s operation,

Page 6 of 49

Accep

ted

Man

uscr

ipt

4

which in turn increases shareholders‟ value3. In the context of our study, if higher CSR is

associated with higher shareholders‟ value, we expect CEO to be rewarded for his effort in

improving CSR investment. Hence, we predict a positive association between CSR and CEO

compensation. We refer to this view as the value-creation hypothesis.

Under the second view, CSR is associated with investments in negative net present value

projects that destroy shareholders‟ value. The key to this view is that managers may over-invest

in CSR that transfer wealth from shareholders to other stakeholders (such as community,

regulators and employees), as managers have incentives to use CSR investment to build their

personal reputations, but these investments may reduce shareholders‟ value in at least three

ways. First, managers‟ career concerns might make them focus excessively on short-term profit

and distort their investment decisions, such as myopic CSR investment (Narayanan, 1985) and

overinvestment (Holmstrom and Costa, 1986). For example, managers often overinvest in CSR

for private rent-seeking benefits to secure their personal reputations in the community, enhance

their personal status with stakeholders, and leave personal legacies that destroy shareholders‟

wealth (e.g., Barnea and Rubin, 2010; Cespa and Cestone, 2007; Hemingway and Maclagan,

2004). Second, career concerns may drive managers to manipulate the flow of information

relating to the resolution of uncertainty surrounding CSR investment (Hirshleifer, 1993;

Hirshleifer and Thakor, 1992)4. Third, managerial reputation building can encourage herding of

CSR investment and lead to over-investments (Scharfstein and Stein, 1990; Trueman, 1986). For

3 We acknowledge that institutional forces can often lead to symbolic rather than genuine CSR actions and policies

whereby firms may appear to engage in CSR, but these initiatives are simply intended to appease stakeholder

demands or meet the minimum requirements of standards. Under this view, if CSR is purely symbolic without any

effect on shareholders‟ value, we expect no association between CEO compensation and CSR. However, if

symbolic CSR reduces shareholders‟ value, we expect a negative association between CEO compensation and CSR.

4 Specifically, to increase his personal reputation in the managerial labour market, CEO may delay the release of bad

news about inefficient CSR investments.

Page 7 of 49

Accep

ted

Man

uscr

ipt

5

example, CEOs might invest in CSR just for cosmetic or ceremonial reasons with little or no

economic payoff accruing to the firm, but as a mechanism to enhance their stature as corporate

citizens (e.g., Surroca and Tribó, 2008). We refer to this view as the value-destruction

hypothesis. It predicts a negative association between CSR and CEO compensation.

In summary, the value-creation (value-destruction) hypothesis predicts a positive

(negative) association between CEO compensation and CSR. Ultimately, it is an empirical issue

whether the value-creation hypothesis dominates the value-destruction hypothesis in terms of the

net impact of CSR on CEO compensation.

Our results suggest that CEO compensation level is negatively associated with CSR

investment. This result holds after controlling for standard economic determinants of CEO

compensation (such as firm size, growth opportunities, operating profitability and stock returns)

and CEO characteristics (such as age and tenure). We interpret our results as consistent with the

value-destruction hypothesis.

At first glance, the negative association between that CEO compensation level and CSR

investment may seem puzzling. A natural question is that why a CEO would undertake actions to

increase firm-specific CSR investment if he is penalized for high CSR investment. To address

this question, we draw on the insight from Larcker (2003), who suggest that although firms are

continuously strive to optimize their corporate financial policies, there are many legitimate

reasons to expect firm policies and managerial decision to deviate from their optimal levels. One

reason is that firm characteristics and manager characteristics that drives the optimal levels

change with time (Core and Guay, 1999). Another reason could be changes in legal and other

institutional environment that shift the optimal level. In reality, firms and managers do exhibit

Page 8 of 49

Accep

ted

Man

uscr

ipt

6

sub-optimal decisions as organizations adapt by experimentation and imitation (Ittner and

Larcker, 2001; Milgrom and Roberts, 1992).

To shed light on the prior inconclusive findings on the association between CEO

compensation and CSR, we conjecture it is important to distinguish normal CSR (which relates

to the optimal level of CSR investment that potentially increases shareholders‟ value) and

abnormal CSR (which relate to excessive CSR investment that can potentially destroy

shareholders‟ value), and incorporate it into a CEO compensation-CSR investments framework.

It is important to distinguish normal CSR and abnormal CSR for at least two reasons.

First, Larcker (2003) argues that although firms continuously strive to optimize their

corporate financial policies, it is plausible that firm policies and managerial decisions deviate

from the optimal levels.

If one were to subscribe to the extreme optimization perspective that all firms in the

sample are optimizing with respect to their CSR investments all the time, there should be no

statistically significant association between CEO compensation and abnormal CSR (Demsetz and

Lehn, 1985; Ittner and Larcker, 2001). In such a case, any statistically significant association

between CEO compensation related to the abnormal CSR ought to occur only because of

measurement error, misspecification of functional form or an inadequate set of controls.

However, this is an extreme view of the world but not a useful framework for structuring

research. If we assume that „„all firms are optimizing all the time,‟‟ there is no way for

researchers to provide any insight into the consequences of managerial choices (Larcker, 2003).

Ittner and Larcker (2001) and Ittner et al. (2003) argue that such an extreme optimization

perspective that fails to acknowledge the possibility of any off equilibrium behavior is perhaps

unrealistic. Instead, it is likely that all organizations are dynamically learning and moving

Page 9 of 49

Accep

ted

Man

uscr

ipt

7

towards the optimal level, but a cross-sectional sample will consist of firms that are distributed

around the optimal choice. Thus, to allow for the possibility of such dynamic learning towards

the optimal choice, we are able to assess whether the abnormal CSR is associated with CEO

compensation.

Second, in the context of CSR investments, Borghesi et al. (2014) suggest that companies

evaluate socially responsible investments in a way similar to all other investments. They are

undertaken if (1) they create value for investors, and/or (2) managers perceive a personal benefit

from the investment. Hence managers might undertake CSR investments due to their career

concerns, which may be enhanced by having a reputation for championing socially responsible

investments.

We posit that in setting CEO compensation, it is important to distinguish between normal

CSR (optimal level of CSR) and abnormal CSR (deviations from optimal level of CSR). If one

were to subscribe to the extreme optimization perspective that all firms in the sample are

optimizing with respect to their CSR investments all the time, there should be no statistically

significant association between CEO compensation and abnormal CSR (Demsetz and Lehn,

1985; Ittner and Larcker, 2001). In such a case, any statistically significant association between

CEO compensation related to the abnormal CSR ought to occur only because of measurement

error, misspecification of functional form or an inadequate set of controls. However, this is an

extreme view of the world but not a useful framework for structuring research. If we assume that

„„all firms are optimizing all the time,‟‟ there is no way for researchers to provide any insight

into the consequences of managerial choices (Larcker, 2003). Ittner and Larcker (2001) and

Ittner et al. (2003) argue that such an extreme optimization perspective that fails to acknowledge

the possibility of any off equilibrium behavior is perhaps unrealistic. Instead, it is likely that all

Page 10 of 49

Accep

ted

Man

uscr

ipt

8

organizations are dynamically learning and moving towards the optimal level, but a cross-

sectional sample will consist of firms that are distributed around the optimal choice. Thus, to

allow for the possibility of such dynamic learning towards the optimal choice, we are able to

assess whether the abnormal CSR is associated with CEO compensation.

We draw on the prior literature on the determinants of CSR (Johnson and Greening,

1999; Karpoff et al., 2005; McWilliams and Siegel, 2000; Wu, 2006) to isolate normal CSR from

abnormal CSR. Extending the prior literature, we suggest that normal CSR reflects the optimal

level of CSR investment that potentially increase shareholders‟ value and that the abnormal CSR

reflects excessive investment in CSR that potentially reduce shareholders‟ value (Fazzari et al.,

1988; Hubbard, 1998; Titman et al., 2004).

Decomposing total CSR investment into a normal component and an abnormal

component generates two interesting insights. First, we find that CEO compensation level is

positively associated with normal CSR. To the extent that normal CSR reflects the optimal level

of CSR investment, this result suggests that CEO is rewarded for investing in optimal level of

CSR. Second, we find CEO compensation level is negatively associated with abnormal CSR,

which suggests that when CSR investment deviates from its optimal level, CEO is punished for

excessive investments in CSR investment.

Next, we examine the effect of corporate governance on the association between CEO

compensation and CSR. In firms with strong corporate governance, we find that CEO

compensation is negatively associated with total CSR investment. Stated differently, the negative

association between CEO compensation and CSR is more pronounced in firms with stronger

corporate governance. We interpret our result as suggesting that strong corporate governance

Page 11 of 49

Accep

ted

Man

uscr

ipt

9

curtails the incentives of CEO to over-invest in CSR. However, in firms with weak corporate

governance, we find that CEO compensation is not associated with total CSR.

We then investigate whether there are systematic differences in the association between

CEO compensation and normal CSR (abnormal CSR) across different corporate governance

structures. In firms with strong corporate governance, we find that CEO compensation is

positively associated with normal CSR investment. Stated differently, to the extent that normal

CSR reflect optimal level of CSR investment, the result suggests that the CEO is rewarded for

undertaking normal CSR investment in firms that have strong corporate governance. In contrast,

we find that CEO compensation is negatively associated with abnormal CSR investment in firms

with strong corporate governance. Hence, effective corporate governance structure curtails

agency costs by reducing CEO‟s compensation level in firm that over-invested in CSR.

Our paper makes four contributions to the literature. First, by providing evidence on the

association between CEO compensation and CSR investment, we extend CEO compensation

research that has extensively focused on the financial performance. We show that the

incremental effect of CSR investment (over and above traditional financial performance

measures such as operating profitability and stock returns) on CEO compensation. Second, we

contribute to the corporate governance literature by demonstrating that effective corporate

governance structures play an important role in monitoring and rewarding CSR investment and

thus affect the association between CEO compensation and CSR investment. Third, we provide

evidence that in setting CEO compensation, firms appear to distinguish between normal CSR and

abnormal CSR. This delineation between normal CSR and abnormal CSR highlights the

potential pitfall in the “one-size fits all” recommendation by CSR advocates such as Global

Reporting Initiatives and Corporate Register that CEO should be rewarded for all CSR activities.

Page 12 of 49

Accep

ted

Man

uscr

ipt

10

More generally, our results broadly support the findings in prior studies that compensation

committee influence the CEO compensation setting process to promote optimal contracting

incentives (Adut et al., 2003; Dechow et al., 1994). Fourth, by separating normal CSR and

abnormal CSR, our results can potentially explain the prior puzzling finding in Coombs and

Gilley (2005) and Cai et al. (2011) that CEO compensation is negatively associated with total

CSR. In this paper we document systematic differences between normal CSR and abnormal CSR

in affecting CEO compensation. Furthermore, our results suggest that the association between

abnormal CSR and CEO compensation is conditioned on corporate governance structure.

The rest of the paper is organized as follows. Section II describes the data and research

design. Section III presents our results. Section IV concludes.

2. Data and Research Design

2.1 Sample formation

We obtain CEO compensation data from Execucomp. Financial accounting information

is from the Compustat and stock return data is from CRSP. We obtain CSR data from the Kinder,

Lydenberg and Domini (KLD) Database. Our primary source of corporate governance data is the

Risk Metrics database. We drop the following observations in the screening procedures to obtain

our sample: (1) observations with CEO turnover during the year; (2) observations in which the

current CEO has served the company for strictly less than two consecutive years; (3) firms in the

regulated industries (SIC code 4900 to 4999) or in the financial industry (SIC code 6000 to

6999); and (4) observations with zero CEO compensation in the year. The final sample consists

of 12,507 firm-years for 1,680 firms for the period 1992 to 2011.

Page 13 of 49

Accep

ted

Man

uscr

ipt

11

Table 1 describes the sample distribution over years. There was a significant increase in

the number of observations per year in and after 2003. This is mainly due to the fact that KLD

more than doubled its coverage in 2003, covering not only MSCI KLD 400 Social Index but also

the 3,000 largest US companies, while in earlier years it covered only the 1,000 largest US firms.

[Table 1 here]

2.2 Measuring CSR

KLD data have been used extensively in scholarly research to operationalize the CSR

construct (e.g., Szwajkowski and Figlewicz, 1999; Turban and Greening, 1997; Waddock and

Graves, 1997). Waddock (2003) argues that the KLD data are „„the de facto research standard‟‟

for measuring CSR in scholarly research. Chatterji et al.l (2009) contend that KLD‟s social

ratings are among the most influential and the most widely accepted CSR measure used by

academics. Mattingly and Berman (2006) assert that the KLD dataset has become the standard

for quantitative measurement of corporate social actions.

KLD evaluates CSR on seven main dimensions including community, corporate

governance, diversity, employee relations, environment, human rights and product. Corporate

governance is perceived as a distinct construct from CSR and its impact on CEO compensation is

widely examined in the prior literature (e.g., Core et al., 1999). In order to disentangle the effect

of CSR and corporate governance, we construct the CSR measure based on the six remaining

dimensions, excluding corporate governance. Specifically, following prior studies (Chatterji et

al., 2009; Johnson and Greening, 1999; Waddock and Graves, 1997), we construct CSR,

measured as total strengths minus total concerns in KLD‟s six social rating categories:

community, diversity, employee relations, environment, human rights and product.

Page 14 of 49

Accep

ted

Man

uscr

ipt

12

2.3 Empirical Model

2.3.1 CEO Compensation and CSR

Following prior studies on the determinants of CEO compensation level (Core et al.,

2008; Core et al., 1999), we employ the following model to test the association between CEO

compensation and CSR:

CEO compensation = 0 + 1× CSR + 2×FIRMSIZE + 3×ROA + 4×RETURN + 5×

VOLAROA + 6× VOLARET + 7× MTB + 8× TENURE + 9× AGE + 10×BDINDEP +

11×BDOWN + 12×INSTI + INDUSTRY + YEAR (1)

where

CEO compensation = Natural logarithm of the sum of one and total CEO compensation

level comprising salary, bonus, stock options granted, restricted stocks granted, long term

incentive payouts and other annual compensation in the fiscal year.

CSR = KLD strengths less KLD concerns for community, diversity, employee relations,

environment, human rights and product.

FIRMSIZE = Natural logarithm of total assets

ROA = Operating income divided by total assets

RETURN = Stock return in the fiscal year

VOLAROA = ROA volatility in the past 5 years

VOLARET = Stock Return volatility in the past 5 years

MTB = Market value of equity divided by book equity

TENURE = Tenure of CEO in the fiscal year.

AGE = Age of CEO in the fiscal year.

Page 15 of 49

Accep

ted

Man

uscr

ipt

13

BDINDEP = percentage of independent directors on the board. Independent directors are

directors who are neither current nor former employees of the firm.

BDOWN = percentage of common stock owned by all directors.

INSTI = percentage of common stock held by institutional shareholders.

INDUSTRY = Dummy variables to control for industry fixed effects at the 2-digit SIC

level.

YEAR = Dummy variables to control for year fixed effects.

Under the value creation hypothesis, we predict 1 to be positive. Under the value

destruction hypothesis, we predict 1 to be negative.

We include the following control variables. Past studies (Rosen, 1982; Smith and Watts,

1992) find that large firms are more complex and they demand managers with more equilibrium

wages. We control for firm size with the natural logarithm of total assets (FIRMSIZE). Core et

al. (1999) argue that CEO compensation level is positively associated with firm performance.

Our proxies for firm performance are return on assets (ROA) and stock return in the fiscal year

(RETURN). High growth firms are more complex to manage compared to low growth firms

(Smith and Watts, 1992). We control for the firm‟s growth opportunities with the ratio of market

value of common equity to book value of common equity (MTB). Firms operating in more

volatile environment are typically riskier. Hence, CEOs helming riskier should be compensated

with higher compensation level (Banker and Datar, 1989). We control for uncertainty in

operation with two proxies: standard deviation of return on assets in the past five years

(VOLAROA) and standard deviation of stock returns in the past five years (VOLAROA).

Following Core, Holthausen and Larker (1999), we also control for CEO characteristics such as

CEO age in the fiscal year (AGE) and CEO tenure at the firm (TENURE). We control for the

Page 16 of 49

Accep

ted

Man

uscr

ipt

14

effect of corporate governance on CEO compensation by including the percentage of

independent directors on the board (BDINDEP), percentage of common stock owned by all

directors (BDOWN) and percentage of common stock held by institutional shareholders (INSTI).

We also include industry dummy variables and year dummy variables to control for industry and

time-series effects on CEO compensation.

2.3.2 Normal CSR and Abnormal CSR

To further distinguish between value creation hypothesis and value destruction

hypothesis, we draw on prior literature on the determinants of CSR which posits that based on

the firm‟s operating characteristics and industry factors, there is an optimal level of CSR

investment (Johnson and Greening, 1999; Karpoff et al., 2005; McWilliams and Siegel, 2000;

Wu, 2006). Under this approach, a deviation in the amount of CSR investment from this level is

suboptimal.

We draw on the theoretical insights provided by Demsetz and Lehn (1985), Ittner and

Larcker (2001), Ittner et al. (2003) and Borghesi et al. (2014) to motivate our discussion on the

decomposition of total CSR into normal CSR and abnormal CSR. As emphasized by Larcker

(2003), although firms continuously strive to optimize their corporate financial policies, there

are many legitimate reasons to expect firm policies and managerial decision to deviate from their

optimal levels. Potential reasons include changes in firm characteristics and manager

characteristics (Core and Guay, 1999) and changes in legal and other institutional environment

that shift the optimal level. Moreover, organizations adapt by experimentation and imitation

(Ittner and Larcker, 2001; Milgrom and Roberts, 1992) . In other words, firms and managers do

exhibit sub-optimal decisions in reality. We acknowledge that if one were to adopt an extreme

Page 17 of 49

Accep

ted

Man

uscr

ipt

15

optimization perspective that all firms in the sample are optimizing with respect to their CSR

investments all the time, there should be no statistically significant association between CEO

compensation and abnormal CSR (Demsetz and Lehn, 1985; Ittner and Larcker, 2001).

Following this logic, any statistically significant association found between CEO compensation

and abnormal CSR should occur only when there is measurement error or an inadequate set of

controls. Such an extreme optimization perspective is perhaps unrealistic as it fails to

acknowledge the possibility of any off equilibrium behavior (Ittner and Larcker, 2001; Ittner et

al., 2003). Instead, we follow the insights from Hanlon et al (2003), Ittner and Larcker (2001)

and Ittner et al. (2003) and we posit that it is likely that all organizations are dynamically

learning and moving towards the optimal level, while a cross-sectional sample will consist of

firms that are distributed around the optimal choice. Thus, to allow for the possibility of such

dynamic learning towards the optimal choice, we are able to assess whether the abnormal CSR is

associated with CEO compensation.

Following prior literature we decompose total CSR investment into two components: (a)

the component that can be explained by investment based factors and (b) the component that is

unrelated to investment based factors. We refer to the first component as the “normal CSR”, and

the latter as the “abnormal CSR”. Specifically, we determine the optimal CSR and deviation for

each firm by estimating the following model:

CSR = 0 + 1 × ATO + 2×PM + 3× CASH + 4× CFO + 5×LEVERAGE + 6×

MTB + 7× FIRMSIZE + 8× R&D + 9× ADV + 10× BDINDEP + 11 × BDOWN +

12× INSTI + INDUSTRY + YEAR (2)

where

Page 18 of 49

Accep

ted

Man

uscr

ipt

16

CSR = KLD strengths less KLD concerns for community, diversity, employee relations,

environment, human rights and product

ATO = Sales over total assets.

PM = Income before extraordinary items divided by sales.

CASH = Cash divided by total assets.

CFO = Cash flow from operations divided by sales.

LEVERAGE = Total debt divided by total assets.

MTB = Market value of equity divided by book equity.

FIRMSIZE = Natural logarithm of total assets.

R&D = Research and development expenditure divided by sales.

ADV = Advertising expenditure divided by sales.

BDINDEP = percentage of independent directors on the board.

BDOWN = percentage of common stock owned by all directors.

INSTI = percentage of common stock held by institutional shareholders.

INDUSTRY = Dummy variables to control for industry fixed effects at the 2-digit SIC

level.

YEAR = Dummy variables to control for year fixed effects.

We use the fitted value of CSR investment from equation (2) as a proxy for the normal

(optimal) CSR investment and the residual from equation (2) as a proxy for abnormal CSR (i.e.

the deviation from the optimal CSR).

Drawing from previous literature that examines the major drivers of CSR investment, we

include the following variables in our model on the determinants of CSR. To control for the

effect of firm performance on CSR, we include return on assets and cash flow from operations

Page 19 of 49

Accep

ted

Man

uscr

ipt

17

divided by sales. Rather than including contemporaneous ROA, we split ROA into its constituent

components: Asset Turnover (ATO) and Profit Margin (PM), as asset turnover and profit margin

measure different aspects of profitability and are therefore likely to have different persistence

(Nissim and Penman, 2001). We include cash as a percentage of total assets and leverage to

proxy for financial flexibility because firms with more financial resources are more likely to

invest in CSR (Karpoff et al., 2005). To control for firm size, we include the natural logarithm of

total assets because larger firms have greater resources for CSR investment (Wu, 2006). Prior

studies (McWilliams and Siegel, 2000; Wieser, 2005) find that firms with higher intangibles

have higher investments in CSR due to higher sensitivity of firm value to growth opportunities.

We control for intangibles by including advertising intensity (advertising expenditure divided by

sales) and research and development intensity (research and development expenditure divided by

sales). Leverage and market-to-book equity are included as stable firms with lower risk are more

likely to invest in CSR (Cochran and Wood, 1984; Orlitzky and Benjamin, 2001). Stronger

corporate governance is associated with greater effectiveness of CSR investment (Johnson and

Greening, 1999). Thus, we include various corporate governance attributes such as the

percentage of independent directors on the board (BDINDEP), percentage of common stock

owned by all directors (BDOWN) and percentage of common stock held by institutional

shareholders (INSTI). Lastly, we include industry fixed effects and year fixed effects.

We take the fitted and residual values from equation (2), and test the association between

CEO compensation and these components of CSR investment using the following specification:

CEO compensation = λ0 + λ1×NORMAL_CSR + λ2× ABNORMAL_CSR +

CONTROLS (3)

Page 20 of 49

Accep

ted

Man

uscr

ipt

18

The value creation hypothesis predicts that CEO compensation level be positively

associated with normal CSR and abnormal CSR. Thus, the value creation hypothesis implies that

λ1 > 0 and λ2 > 0. In contrast, the value destruction hypothesis, predicts that (i) CEO

compensation level be positively associated with normal CSR because CEO should be rewarded

for optimal investments in CSR (i.e., λ1 > 0); and (ii) CEO compensation level be negatively

associated with abnormal CSR because CEO should be penalized for deviation from the optimal

investments in CSR (i.e., λ2 < 0) .

3. Results

3.1 Descriptive Statistics

Table 2 presents the descriptive statistics. Mean (median) CEO total compensation is $

5.552 ($3.329) million. Mean CSR is 0.280. On average, the sample firms are profitable with a

mean return on assets (ROA) of 0.106. The mean buy-and-hold gross stock return during the

fiscal year is 1.0665. Mean standard deviation of return on assets in the past five years

(VOLAROA) is 0.039 and mean standard deviation of stock return in the past five years

(VOLARET) is 0.384. The mean market to book value of common equity (MTB) is 1.710. Mean

CEO age is 56 years and the mean CEO tenure is 12.875 years. Mean board independence

(BDINDEP) is 68%. Mean percentage of common stock owned by the board of directors

(BDOWN) is 4%. Mean percentage of common stock owned by the institutional shareholders is

63%.

[Table 2 here]

5 This implies that $100 invested at the beginning of the year will be worth $106.6 at the end of the year, generating

a net return of 6.6%.

Page 21 of 49

Accep

ted

Man

uscr

ipt

19

Table 3 presents the Spearman correlation among selected variables. Several results are

noteworthy. First, total CEO compensation level is positively associated with CSR investment at

the 5% significance level. Second, we find that CSR is positively associated with firm size,

operating profitability and market-to-book ratio. Third, CSR is negatively associated with

volatility in profitability, volatility in stock return, CEO age, board independence, directors‟

stock ownership and institutional ownership. We caution that these simple correlations are

univariate correlations that do not control for differences in other firm characteristics such as

firm size. We conduct a formal multivariate analysis below where we control for these other

important differences.

[Table 3 here]

3.2 Regressions of CEO Compensation on CSR

Table 4 presents the regressions of CEO compensation on CSR. The dependent variable is

CEO total compensation comprising salary, bonus, stock options granted, restricted stocks

granted, long term incentive pay and other annual compensation. In Column (1), the estimated

coefficient on CSR is negative and significant at the 1% level, which suggests that CEO total

compensation is negatively associated with CSR investment. In other words, after controlling for

standard economic determinants of CEO compensation such as firm size, profitability, stock

return, firm‟s operating risk and growth opportunities, and corporate governance measures such

as board independence, board ownership and institutional ownership, CEO receive lower total

compensation in firms with higher CSR investment. This result is consistent with the value

destruction hypothesis of CSR investment.

Page 22 of 49

Accep

ted

Man

uscr

ipt

20

In column (2), we construct a composite measure of corporate governance strength

(GOVSTRENGTH) based on the principal components analysis of the percentage of



Higher scores of GOVSTRENGTH denote higher corporate governance strength. After

controlling for firm-level corporate governance strength, we still find CEO total compensation is

negatively associated with CSR investment. In column (2), the estimated coefficient on CSR is -

0.024 (significant at the 1% level).

Based on the estimated coefficients model (1), if the firm increases its CSR from the 25th

percentile to the 75th percentile, CEO compensation decreases by 4.29%. 6

Hence, the result is

economically significant.

[Table 4 here]

3.3 Determinants of CSR

Table 5 presents the regression results of the determinants of CSR based on equation (2).

We conduct a number of robustness checks on the model. In column (1) we include all firm

characteristics in addition to industry and year fixed effects. In column (2), we include all firm

characteristics, the corporate governance measures as well as industry and year fixed effects. In

column (3), we use the composite index of corporate governance strength (GOVSTRENGTH) to

replace the three governance measures: the percentage of independent directors on the board

(BDINDEP), percentage of common stock owned by all directors (BDOWN) and percentage of

common stock held by institutional shareholders (INSTI). In general, results are qualitatively

consistent. Consistent with prior studies (Johnson and Greening, 1999; Karpoff et al., 2005;

6 We hold the control variables at their sample means.

Page 23 of 49

Accep

ted

Man

uscr

ipt

21

McWilliams and Siegel, 2000; Wu, 2006), we find that CSR is positively associated with

operating cash flow, firm size, research and development divided by sales, and advertising

divided by sales. On the other hand, CSR is negatively associated with corporate governance

strength.

[Table 5 here]

In our next set of tests, we use the fitted values of CSR based on estimated coefficients in

Table 5 column (1) to compute the normal CSR (optimal level of CSR based on firm

characteristics). Thus, the residual values from the regression in column (1) Table 5 reflect the

abnormal CSR (deviations from the optimal level of CSR). 7

3.4 Regressions of CEO Compensation on Normal CSR and Abnormal CSR

In Table 6, we examine whether there are systematic differences between normal CSR and

abnormal CSR in affecting CEO compensation. We repeat the regressions reported in Table 4

but decompose CSR into normal and abnormal component using the model shown in Table 5

column (1). In all specifications we find similar results. First, we find CEO compensation level is

positively associated with normal CSR at the 1% significance level. To the extent that normal

CSR reflects the optimal level of CSR investment, this result suggests that CEO is rewarded for

investing in optimal level of CSR. Second, we find CEO compensation level is negatively

associated with abnormal CSR at the 1% significance level. This result suggests that when CSR

investment deviates from its optimal level, CEO is punished for excessive investments in CSR.

For example, based on the estimated coefficients in model 4 of Table 6 and holding all other

variables at their means, when abnormal CSR increases from the 25th

percentile to the 75th

7 The fitted and residual values for CSR investment from each of the other columns in Table 5 produce qualitatively

similar results in subsequent analysis.

Page 24 of 49

Accep

ted

Man

uscr

ipt

22

percentile, the CEO‟s total compensation decreases 2.17%. Hence, the results are economically

significant.

[Table 6 here]

Collectively, our results highlight the systematic differences between normal CSR and

abnormal CSR in affecting CEO compensation. By separating normal CSR and abnormal CSR,

our results can potentially explain the prior puzzling finding in Coombs and Gilley (2005) and

Cai et al. (2011) that CEO compensation is negatively associated with total CSR.

3.5 Does Corporate Governance Affect the Association between CEO Compensation and

CSR?

In Table 7, we examine the effect of corporate governance on the association between CEO

compensation and CSR investment. We construct a composite measure of corporate governance

strength (GOVSTRENGTH) based on the principal components analysis of the percentage of



Higher scores of GOVSTRENGTH denote higher corporate governance strength. Using the

GOVSTRENGTH, we partition the observations into the sub-sample of firms with strong

corporate governance and the sub-sample of firms with weak corporate governance. In firms

with strong corporate governance (column (1)), we find that CEO compensation is negatively

associated with CSR. However, in firms with weak corporate governance (column (2)), we find

that CEO compensation is not associated with CSR. The Chi-square to test the difference in

coefficient on CSR between strongly-governed firms and weakly-governed firms is 6.85,

Page 25 of 49

Accep

ted

Man

uscr

ipt

23

significant at 1% level. This result suggests that in strongly-governed firms, CEO compensation

is lower when total CSR is higher.

In column (3) and (4), we employ Gompers, Ishii and Metrick (2003) G-score as an

alternative measure of corporate governance. In column (3), in firms with strong corporate

governance (i.e. low G-score), CEO compensation is negatively associated with CSR. In

contrast, the results in column (4) indicate that in firms with weak corporate governance (i.e.

high G-score), CEO compensation is not associated with CSR. In column (5) and (6), we employ

Bebchuk, Cohen and Ferrell (2009) entrenchment index (E-index) as an alternative measure of

corporate governance. Results are qualitatively similar. Specifically, the difference in coefficient

on CSR between strongly-governed firms and weakly-governed firms is significant at 1% level.

Thus, in strongly-governed firms, the negative association between CEO compensation and total

CSR is stronger.

[Table 7 here]

In Table 8, we examine whether corporate governance affects the association between CEO

compensation and normal (abnormal) CSR. Using the composite corporate governance score

(GOVSTRENGTH), we partition the firms into the sub-sample of firms with strong corporate

governance (column (1)) and the sub-sample of firms with weak corporate governance (column

(2)).

In firms with strong corporate governance (column (1)), we find that CEO total

compensation is positively associated with normal CSR, suggesting CEO is rewarded for optimal

level of CSR investment. On the other hand, in well-governed firms CEO total compensation is

negatively associated abnormal CSR, suggesting CEO is punished for excessive CSR

Page 26 of 49

Accep

ted

Man

uscr

ipt

24

investment. Our results underscore the importance of separating normal CSR and abnormal CSR

in setting CEO compensation.

In firms with weak corporate governance (column (2)), we find that CEO total

compensation is positively associated with normal CSR, suggesting CEO is rewarded for optimal

level of CSR investment. More importantly, in firms with weak corporate governance, there is no

association between CEO compensation and abnormal CSR. This result suggests that in poorly

monitored firms, CEO is not punished for excessive investment in CSR.

The Chi-square that tests the difference in coefficients on normal CSR in the two

subsamples (strong versus weak corporate governance) is 0.005 (insignificant), while the Chi-

square that tests the difference in coefficients on abnormal CSR in the two subsamples (strong

versus weak corporate governance) is 8.672, significant at 1% level. Taken together, our results

suggest that it is important to recognize the interplay among corporate governance structures,

normal CSR and abnormal CSR in affecting CEO compensation.

We obtain qualitatively similar results using Gompers, Ishii and Metrick (2003) G-score

(column (3) and (4)) as an alternative measure of corporate governance to partition the sample.

Our main inferences are also similar using Bebchuk, Cohen and Ferrell (2009) entrenchment

index (E-index). In summary, in firms with stronger corporate governance structures, CEOs

receive lower compensation when abnormal CSR is higher.

[Table 8 here]

3.6 Robustness Tests

3.6.1 Alternative measures of CSR

Page 27 of 49

Accep

ted

Man

uscr

ipt

25

In a sensitivity test, as in McWilliams and Siegel (2000), we also use an alternative

measure of CSR, CSR_DSI400, an indicator variable that takes a value of 1 if the firm is

included in the DSI400 in a given year (for having passed the „„social screen‟‟), and 0 otherwise.

In order to be eligible for the DSI 400, a firm must derive less than 2% of its gross revenue from

the production of military weapons, have no involvement in nuclear power, gambling, tobacco,

and alcohol, and have a positive record in each of the remaining six categories. Our results are

qualitatively similar.

In another supplementary test, we use U.S. firms in the FTSE4Good Index as an

alternative sample of socially responsible firms. The FTSE4Good series measures the

performance of firms that meet social and environmental criteria in five categories: (1)

environmental sustainability, (2) human rights, (3) countering bribery, (4) supply chain labor

standards, and (5) climate change. The FTSE4Good Index consists of 600 to 700 socially

responsible firms around the world from 2001 to 2011, of which approximately 200 firms are

U.S. listed firms. After merging the U.S. firms in the FTSE4Good Index with the U.S. firms in

the Execucomp Database, Compustat, and CRSP, we have a sample with 170 US firms.

To examine whether FTSE4Good U.S. firms compensate their CEOs differently from

non- FTSE4Good U.S. firms, we use a propensity score matching approach. We choose

matching firms (i.e. non-FTSE4Good U.S firms) from the merged database of Execucomp and

Compustat database using firm size, leverage, market-to-book ratio, industry dummies and year

dummies. Untabulated results suggest the CEO compensation is approximately 1% lower for

FTSE4Good U.S. firms than for non-FTSE4Good U.S. firms. Since FTSE4Good U.S. firms are

determined independently from KLD firms, the results using FTSE4Good U.S. firms lend

additional support to the value destruction view.

Page 28 of 49

Accep

ted

Man

uscr

ipt

26

3.6.2 Financial Constraints

Another concern is that financially constrained firms have less resource to invest in CSR.

We employ several measures of “financial constraints” such as the Kaplan and Zingales (1997)

measure, the Whited and Wu (2006) size-age measure, firm size and dividends payout. Our main

results on the association between CEO compensation and CSR are robust across alternative

measures of “financial constraints”.

3.6.3 Endogeneity

Although using an extensive list of control variables (in equation 1) helps to reduce

omitted variables bias in estimating the relation between a firm‟s CSR investment and CEO

compensation, the results from the regression could still suffer from endogeneity bias caused by

unobservable omitted variables. To address the potential endogeneity problem, we perform two-

stage-least-squares (2SLS) regression analysis using religion rank and a blue state dummy as

instrumental variables for the CSR investment (Deng et al., 2013). Religion rank measures the

religion ranking of the state in which the firm‟s headquarters is located, which ranges between 1

and 50. The ranking is based on the ratio of the number of religious adherents in the firm‟s state

to the total population in that state in 20008. A higher ranking indicates more religiosity.

Angelidis and Ibrahim (2004) find that the degree of religiousness is positively correlated with

attitudes toward CSR. This finding suggests that the religion rank variable is likely to be

positively correlated with a firm‟s CSR, thus satisfying the relevance requirement of

8 The Association of Religion Data Archive provides information on religiosity every decade. We use 2000 data

since it covers the middle of our sample period (1992-2011). As a robustness test, we recalculate religion ranks for

the periods 1992-1999 and 2000-2007 as the average religion rank based on 1990 and 2000 data and the average

religion rank based on 2000 and 2010 data, respectively. Our results are qualitatively similar.

Page 29 of 49

Accep

ted

Man

uscr

ipt

27

instrumental variables. However, to the extent that the construction of the religion rank variable

is based on the state in which a firm is located, it is unlikely that this variable has a significant

effect on the CEO compensation, satisfying the exclusion condition of instrumental variables.

Blue state is a dummy variable that equals one if a firm‟s headquarters is located in a

blue/democratic state and zero otherwise9. Rubin (2008) finds that firms with high CSR ratings

tend to be located in democratic or blue states. We therefore expect this dummy variable to be

highly correlated with our sample firms‟ CSR. It is unlikely, however, that the choice of locating

in a blue or red state could has a direct significant effect on CEO compensation except via its

effect on CSR.

Using religion rank and a blue state dummy as instrumental variables for the CSR

investment, results (not tabulated) from 2SLS regression analyses indicate that our main

inferences on the association between CEO compensation and CSR are qualitatively similar.

4. Conclusions

We examine whether CEO compensation is associated with corporate social responsibility

(CSR) investment. The value creation (value destruction) hypothesis predicts a positive

(negative) association between CEO compensation and CSR. Consistent with the value

destruction hypothesis, we find that CEO total compensation level is negatively associated with

CSR investment. In firms with strong corporate governance, we find that the negative association

9 We obtain the list of blue states from http://en.wikipedia.org/wiki/Red_states_and_blue_states. We define the state

as a blue state if it is listed as the blue state in the website. In untabulated tests, we redefine the blue state dummy in

a certain year as an indicator that takes the value of one if a firm‟s headquarters is located in a state in which the

Democratic Party wins both previous and next presidential elections of that year. Our results are qualitatively

similar.

http://en.wikipedia.org/wiki/Red_states_and_blue_states

Page 30 of 49

Accep

ted

Man

uscr

ipt

28

between CEO compensation and CSR is more pronounced than in firms with weak corporate

governance.

Decomposing total CSR investment into a normal component and an abnormal component

generates two interesting insights. First, we find CEO compensation level is positively associated

with normal corporate social responsibility. To the extent that normal CSR reflects the optimal

level of CSR investment, this result suggests that CEO is rewarded for investing in optimal level

of CSR. We also document that the positive association between CEO compensation and normal

CSR is more pronounced in firms with stronger corporate governance. Second, we find CEO

compensation level is negatively associated with abnormal corporate social responsibility. This

result suggests that when CSR investment deviates from its optimal level, CEOs receive lower

compensation for excessive investments in CSR. Strong corporate governance structure

penalizes CEO more by reducing CEO compensation if CEOs over-invest in CSR.

Page 31 of 49

Accep

ted

Man

uscr

ipt

29

References

Adut D, Cready WH, Lopez TJ Restructuring charges and CEO cash compensation: A

reexamination. Accounting Review. 2003; 78; 169.

Angelidis J, Ibrahim N An exploratory study of the impact of degree of religiousness upon an

individual's corporate social responsiveness orientation. Journal of Business Ethics. 2004; 51;

119-128.

Banker RD, Datar SM Sensitivity, precision, and linear aggregation of signals for performance

evaluation. Journal of Accounting Research. 1989; 27; 21-21.

Barnea A, Rubin A Corporate Social Responsibility as a Conflict Between Shareholders. Journal

of Business Ethics. 2010; 97; 71-86.

Bebchuk L, Cohen A, Ferrell A What matters in corporate governance? Review of Financial

Studies. 2009; 22; 783-827.

Berrone P, Gomez-mejia LR Environmental performance and executive compensation: an

integrated agency-institutional perspective. Academy of Management Journal. 2009; 52; 103-

126.

Borghesi R, Houston JF, Naranjo A Corporate socially responsible investments: CEO altruism,

reputation, and shareholder interests. Journal of Corporate Finance. 2014; 26; 164-181.

Cai Y, Jo H, Pan C Vice or Virtue? The Impact of Corporate Social Responsibility on Executive

Compensation. Journal of Business Ethics. 2011; 104; 159-173.

Cespa G, Cestone G Corporate Social Responsibility and Managerial Entrenchment. Journal of

Economics & Management Strategy. 2007; 16; 741-771.

Chatterji AK, Levine DI, Toffel MW How well do social ratings actually measure corporate

social responsibility? Journal of Economics & Management Strategy. 2009; 18; 125.

Cochran PL, Wood RA Corporate social responsibility and financial performance. Academy of

Management Journal (pre-1986). 1984; 27; 42-42.

Coombs JE, Gilley KM Stakeholder management as a predictor of CEO compensation: main

effects and interactions with financial performance. Strategic Management Journal. 2005; 26;

827-827.

Core J, Guay W The use of equity grants to manage optimal equity incentive levels. Journal of

Accounting and Economics. 1999; 28; 151-184.

Core JE, Guay W, Larcker DF The power of the pen and executive compensation. Journal of

Financial Economics. 2008; 88; 1-25.

Core JE, Holthausen RW, Larcker DF Corporate governance, chief executive officer

compensation, and firm performance. Journal of Financial Economics. 1999; 51; 371-406.

Cornell B, Shapiro AC Corporate stakeholders and corporate finance. Financial Management.

1987; 16; 5-5.

Dechow PM, Huson MR, Sloan RG The effect of restructuring charges on executives' cash

compensation. The Accounting Review. 1994; 69; 138-138.

Page 32 of 49

Accep

ted

Man

uscr

ipt

30

Demsetz H, Lehn K The Structure of Corporate Ownership: Causes and Consequences. Journal

of Political Economy. 1985; 93; 1155.

Deng X, Kang J-k, Low BS Corporate social responsibility and stakeholder value maximization:

Evidence from mergers. Journal of Financial Economics. 2013; 110; 87-109.

Fazzari S, Hubbard RG, Petersen BC Financing Constraints and Corporate Investment. National

Bureau of Economic Research Working Paper Series. 1988; No. 2387.

Gompers PA, Ishii JL, Metrick A Corporate governance and equity prices. Quaterly Journal of

Economics. 2003; 118; 107-156.

Greening DW, Turban DB Corporate social performance as a competitive advantage in attracting

a quality workforce. Business and Society. 2000; 39; 254-280.

Hanlon M, Rajgopal S, Shevlin T Are executive stock options associated with future earnings?

Journal of Accounting and Economics. 2003; 36; 3-43.

Hemingway CA, Maclagan PW Managers' Personal Values as Drivers of Corporate Social

Responsibility. Journal of Business Ethics. 2004; 50; 33-44.

Hill CWL, Jones TM Stakeholder-agency theory. The Journal of Management Studies. 1992; 29;

131-131.

Hirshleifer D Managerial reputation and corporate investment decisions. Financial Management.

1993; 22; 145.

Hirshleifer D, Thakor AV Managerial conservatism, project choice, and debt. The Review of

Financial Studies. 1992; 5; 437-470.

Holmstrom B, Costa JRI Managerial incentives and capital management. Quarterly Journal of

Economics. 1986; 101; 835-860.

Hubbard RG Capital--market imperfections and investment. Journal of Economic Literature.

1998; 36; 193-225.

Ittner CD, Larcker DF Assessing empirical research in managerial accounting: a value-based

management perspective. Journal of Accounting and Economics. 2001; 32; 349-410.

Ittner CD, Larcker DF, Randall T Performance implications of strategic performance

measurement in financial services firms. Accounting, Organizations and Society. 2003; 28; 715-

741.

Jensen MC, Meckling WH Theory of the firm: Managerial behavior, agency costs and ownership

structure. Journal of Financial Economics. 1976; 3; 305-360.

Johnson RA, Greening DW The effects of corporate governance and institutional ownership

types on corporate social performance. Academy of Management Journal. 1999; 42; 564-576.

Kaplan SN, Zingales L Do investment-cash flow sensitivities provide useful measures of

financing constraints? Quarterly Journal of Economics. 1997; 112; 169-215.

Karpoff JM, Lott JR, Jr., Wehrly EW The reputational penalities for environmental violations:

Empirical evidence. Journal of Law and Economics. 2005; 48; 653-653+.

Larcker DF Discussion of “are executive stock options associated with future earnings?”. Journal

of Accounting and Economics. 2003; 36; 91-103.

Page 33 of 49

Accep

ted

Man

uscr

ipt

31

Maignan I, Ferrell OC, Hult GTM Corporate citizenship: Cultural antecedents and business

benefits. Academy of Marketing Science. Journal. 1999; 27; 455-469.

Mattingly JE, Berman SL Measurement of corporate social action: Discovering taxonomy in the

Kinder Lydenburg Domini ratings data. Business and Society. 2006; 45; 20-46.

McWilliams A, Siegel D Corporate social responsibility and financial performance: Correlation

or misspecification? Strategic Management Journal. 2000; 21; 603-609.

Milgrom PR, Roberts J Economics, Organization, and Management. Prentice-Hall, Englewood

Cliffs, NJ.; 1992.

Narayanan MP Managerial incentives for short-term results. Journal of Finance. 1985; 40; 1469-

1484.

Navarro P Why do corporations give to charity? The Journal of Business. 1988; 61; 65-65.

Nissim D, Penman S Ratio Analysis and Equity Valuation: From Research to Practice. Review

of Accounting Studies. 2001; 6; 109-154.

Orlitzky M, Benjamin JD Corporate social performance and firm risk: A meta-analytic review.

Business and Society. 2001; 40; 369-396.

Rosen S Authority, control, and the distribution of earnings. Bell Journal of Economics. 1982;

13; 311-311.

Rubin A Political views and corporate decision making: The case of corporate social

responsibility. The Financial Review. 2008; 43; 337.

Russo MV, Harrison NS Organizational design and environmental performance clues from the

electronics industry. Academy of Management Journal. 2005; 48; 582-593.

Scharfstein DS, Stein JC Herd behavior and investment. American Economic Review. 1990; 80;

465.

Sen S, Bhattacharya CB Does doing good always lead to doing better? Consumer reactions to

corporate social responsibility. JMR, Journal of Marketing Research. 2001; 38; 225-243.

Sharma S, Vredenburg H Proactive corporate environmental strategy and the development of

competitively valuable organizational capabilities. Strategic Management Journal (1986-1998).

1998; 19; 729-729.

Smith CW, Jr., Watts RL The investment opportunity set and corporate financing, dividend, and

compensation policies. Journal of Financial Economics. 1992; 32; 263-263.

Stanwick PA, Stanwick SD CEO compensation: does it pay to be green? Business Strategy and

the Environment. 2001; 10; 176-182.

Surroca J, Tribó JA Managerial Entrenchment and Corporate Social Performance. Journal of

Business Finance & Accounting. 2008; 35; 748-789.

Szwajkowski E, Figlewicz RE Evaluating corporate performance: A comparison of the fortune

reputation survey and the socrates social rating database. Journal of Managerial Issues. 1999; 11;

137-154.

Titman S, Wei KCJ, Xie F Capital Investments and Stock Returns. Journal of Financial &

Quantitative Analysis. 2004; 39; 677-700.

Page 34 of 49

Accep

ted

Man

uscr

ipt

32

Trueman B The relationship between the level of capital expenditures and firm value. Journal of

Financial & Quantitative Analysis. 1986; 21; 115-129.

Turban DB, Greening DW Corporate social performance and organizational attractiveness to

prospective employees. Academy of Management Journal. 1997; 40; 658-672.

Waddock S Stakeholder performance implications of corporate responsibility. International

Journal of Business Performance Management. 2003; 5; 114-124.

Waddock SA, Graves SB The corporate social performance-financial performance link. Strategic

Management Journal. 1997; 18; 303-303.

Whited TM, Wu G Financial constraints risk. Review of Financial Studies. 2006; 19; 531-559.

Wieser R Research and development productivity and spillovers: Empirical evidence at the firm

level. Journal of Economic Surveys. 2005; 19; 587-621.

Wu M-L Corporate social performance, corporate financial performance, and firm size: A meta-

analysis. Journal of American Academy of Business, Cambridge. 2006; 8; 163-171.

Page 35 of 49

Accep

ted

Man

uscr

ipt

33

Table 1

Sample

This table presents the sample distribution over the sample period from 1992 through 2011.

Year Number of

firm-years Percentage (%)

1992 138 1.1

1993 286 2.29

1994 302 2.41

1995 291 2.33

1996 308 2.46

1997 312 2.49

1998 321 2.57

1999 320 2.56

2000 313 2.5

2001 458 3.66

2002 493 3.94

2003 906 7.24

2004 933 7.46

2005 924 7.39

2006 957 7.65

2007 1,036 8.28

2008 1,022 8.17

2009 1,080 8.64

2010 1,079 8.63

2011 1,028 8.22

Total 12,507 100

Page 36 of 49

Accep

ted

Man

uscr

ipt

34

Table 2

Descriptive statistics

This table presents summary statistics of the variables used in later analysis. Our sample consists of 12,705 firm-year observations from 1992

through 2011. CEO compensation is the total CEO compensation level (million USD) comprising salary, bonus, stock options granted, restricted

stocks granted, long term incentive payouts and other annual compensation in the fiscal year. CSR is computed as the KLD strengths less KLD

concerns for community, diversity, employee relations, environment, human rights and product. FIRMSIZE is the natural logarithm of total

assets. ROA is calculated as the operating income divided by total assets. RETURN is the firm‟s gross stock return in the fiscal year.

VOLAROA measures the ROA volatility in the past 5 year. VOLARET is the firm‟s stock return volatility in the past 5 years. MTB is computed

as the market value of equity divided by book equity of the firm. TENURE represents the tenure of CEO in the fiscal year. AGE denotes the age

of CEO in the fiscal year. R&D is calculated as the research and development expenditure divided by sales. ADV is the advertising expenditure

divided by sales. BDINDEP is the percentage of independent directors on the board. Independent directors are directors who are neither current

nor former employees of the firm. BDOWN is the percentage of common stock owned by all directors. INSTI is the percentage of common

stock held by institutional shareholders. ATO is sales over total assets. PM is calculated as income before extraordinary items divided by sales.

CASH is computed as cash divided by total assets. CFO represents cash flow from operations divided by sales. LEVERAGE is computed as

total debt divided by total assets. MTB is market value of equity divided by book equity. FIRMSIZE is the natural logarithm of total assets.

R&D is calculated as the research and development expenditure divided by sales. ADV is the advertising expenditure divided by sales.

Page 37 of 49

Accep

ted

Man

uscr

ipt

35

Table 2 (continued)

Variable Mean Standard

Deviation

25th

percentile Median

75th

percentile

CEO Compensation (USD

mil) 5.552 9.691 1.679 3.329 6.45

CSR 0.280 2.268 -1 0 1

FIRMSIZE 7.647 1.455 6.588 7.546 8.594

ROA 0.106 0.08 0.062 0.101 0.149

VOLAROA 0.039 0.038 0.015 0.027 0.049

RETURN 1.066 0.364 0.834 1.015 1.23

VOLARET 0.384 0.309 0.2 0.294 0.447

MTB 1.71 1.133 0.988 1.372 2.019

TENURE 12.875 11.297 4 9 20

AGE 55.692 6.832 51 56 60

BDINDEP 68.13% 18.37% 36.00% 70.00% 82.50%

BDOWN 4.17% 2.59% 0.72% 1.63% 2.94%

INSTI 63.12% 27.91% 38.52% 60.19% 71.62%

ATO 1.126 0.681 0.655 0.966 1.406

PM 0.058 0.105 0.025 0.058 0.101

CASH 0.103 0.105 0.025 0.069 0.147

CFO 0.114 0.072 0.069 0.109 0.154

LEVERAGE 0.202 0.158 0.056 0.195 0.308

R&D 0.042 0.072 0 0.005 0.05

ADV 0.012 0.027 0 0 0.011

Page 38 of 49

Accep

ted

Man

uscr

ipt

36

Table 3

Spearman correlations

This table presents Spearman correlation between variables used in later analysis later. Our sample consists of 12,705 firm-year observations

from 1992 through 2011. CEO compensation (log) is the natural logarithm of the sum of one and total CEO compensation level, which

comprising salary, bonus, stock options granted, restricted stocks granted, long term incentive payouts and other annual compensation in the

fiscal year. CSR is computed as the KLD strengths less KLD concerns for community, diversity, employee relations, environment, human rights

and product. FIRMSIZE is the natural logarithm of total assets. ROA is calculated as the operating income divided by total assets. RETURN is

the firm‟s gross stock return in the fiscal year. VOLAROA measures the ROA volatility in the past 5 year. VOLARET is the firm‟s stock return

volatility in the past 5 years. MTB is computed as the market value of equity divided by book equity of the firm. TENURE represents the tenure

of CEO in the fiscal year. AGE denotes the age of CEO in the fiscal year. BDINDEP is the percentage of independent directors on the board.

Independent directors are directors who are neither current nor former employees of the firm. BDOWN is the percentage of common stock

owned by all directors. INSTI is the percentage of common stock held by institutional shareholders. * denotes significance at 5% level.

Page 39 of 49

Accep

ted

Man

uscr

ipt

37

Table 3 (cont’d)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

(1) CEO

Compensation 1

(2) CSR 0.0933* 1

(3) FIRMSIZE 0.6466* 0.1364* 1

(4) ROA 0.1673* 0.1236* 0.0580* 1

(5) VOLAROA -0.1058* -0.0531* -0.2821* -0.0995* 1

(6) RETURN 0.1028* -0.0336* 0.0134 0.1561* -0.0115 1

(7) VOLARET -0.1098* -0.1256* -0.3268* -0.1415* 0.4005* 0.0954* 1

(8) MTB 0.1002* 0.1616* -0.0830* 0.6095* 0.1035* 0.2523* 0.0354* 1

(9) TENURE -0.0427* 0.0375* 0.0535* 0.0981* -0.0634* 0.0005 -0.0723* 0.0711* 1

(10) AGE 0.0084 -0.0381* 0.0733* 0.0315* -0.0725* -0.0083 -0.1104* -0.0471* 0.3231* 1

(11) BDINDEP 0.0219* -0.1352* 0.0682* 0.0393* -0.0293* 0.0181 -0.0435* 0.0527* 0.0198* 0.0025 1

(12) BDOWN 0.1096* -0.2175* 0.1625* 0.0121 -0.1399* 0.0088 -0.1719* -0.0839* -0.0716* 0.0169 0.0731* 1

(13) INSTI 0.0253* -0.2807* -0.0201* -0.0479* -0.0729* 0.01 -0.0409* -0.1221* -0.1037* 0.0035 0.0014 0.7174*

Page 40 of 49

Accep

ted

Man

uscr

ipt

38

Table 4

Regression of CEO compensation on CSR.

This table presents the OLS regression results of CEO compensation on CSR. Our sample

consists of 12,705 firm-year observations from 1992 through 2011. Standard errors are

clustered by firm. Coefficients are presented with t-statistics below in parentheses. ***, **

and * indicate statistical significance at the 1%, 5% and 10% level, respectively.

The dependent variable is CEO compensation. It is computed as the natural logarithm of the

sum of one and total CEO compensation level, which comprising salary, bonus, stock options

granted, restricted stocks granted, long term incentive payouts and other annual compensation

in the fiscal year. CSR is computed as the KLD strengths less KLD concerns for community,

diversity, employee relations, environment, human rights and product. FIRMSIZE is the

natural logarithm of total assets. ROA is calculated as the operating income divided by

total assets. RETURN is the firm‟s gross stock return in the fiscal year. VOLAROA measures

the ROA volatility in the past 5 year. VOLARET is the firm‟s stock return volatility in the

past 5 years. MTB is computed as the market value of equity divided by book equity of the

firm. TENURE represents the tenure of CEO in the fiscal year. AGE denotes the age of CEO

in the fiscal year. BDINDEP is the percentage of independent directors on the board.

Independent directors are directors who are neither current nor former employees of the firm.

BDOWN is the percentage of common stock owned by all directors. INSTI is the percentage

of common stock held by institutional shareholders.

Page 41 of 49

Accep

ted

Man

uscr

ipt

39

Table 4 (continued)

(1) (2)

CSR -0.021*** -0.024***

(2.673) (2.851)

FIRMSIZE 0.337*** 0.464***

(9.838) (32.299)

ROA 1.710*** 2.261***

(10.747) (11.777)

VOLAROA 0.405 0.096

(1.192) (0.259)

RETURN 0.087*** 0.311***

(4.301) (11.904)

VOLARET 0.042 0.098**

(0.926) (1.974)

MTB 0.060*** 0.084***

(2.917) (4.429)

TENURE -0.003** -0.001

(2.517) (0.397)

AGE 0.003 0.012***

(1.631) (6.108)

BDINDEP -0.029*

(1.960)

BDOWN -0.052**

(2.070)

INSTI -0.107***

(2.410)

GOVSTRENGTH

-0.185**

(2.370)

Constant 4.319*** 2.827***

(13.325) (10.095)

YEAR Yes Yes

INDUSTRY Yes Yes

Adjusted R2 0.718 0.473

Page 42 of 49

Accep

ted

Man

uscr

ipt

40

Table 5

Determinants of CSR investment

This table presents the OLS regressions of the determinants of CSR. Our sample consists of

12,705 firm-year observations from 1992 through 2011. Standard errors are clustered by firm.

Coefficients are presented with t-statistics below in parentheses. ***, ** and * indicate

statistical significance at the 1%, 5% and 10% level, respectively.

The dependent variable is CSR, which is computed as the KLD strengths less KLD

concerns for community, diversity, employee relations, environment, human rights and

product. ATO is sales over total assets. PM is calculated as income before extraordinary

items divided by sales. CASH is computed as cash divided by total assets. CFO represents

cash flow from operations divided by sales. LEVERAGE is computed as total debt divided

by total assets. MTB is market value of equity divided by book equity. FIRMSIZE is the

natural logarithm of total assets. R&D is calculated as the research and development

expenditure divided by sales. ADV is the advertising expenditure divided by sales.

BDINDEP is the percentage of independent directors on the board. Independent directors are

directors who are neither current nor former employees of the firm. BDOWN is the

percentage of common stock owned by all directors. INSTI is the percentage of common

stock held by institutional shareholders.

Page 43 of 49

Accep

ted

Man

uscr

ipt

41

Table 5 (continued)

(1) (2) (3)

ATO 0.127 0.134 0.099

(1.492) (1.578) (1.194)

PM 0.358 0.508** 0.405

(1.349) (1.981) (1.590)

CASH 0.128 0.095 0.074

(0.333) (0.252) (0.206)

CFO 2.525*** 2.382*** 2.310***

(4.638) (4.531) (4.703)

LEVERAGE -0.309 -0.232 -0.184

(0.943) (0.714) (0.599)

MTB 0.053 0.04 0.042

(1.322) (0.998) (1.135)

FIRMSIZE 0.469*** 0.417*** 0.375***

(9.272) (8.492) (8.441)

R&D 3.940*** 4.078*** 4.011***

(5.359) (5.439) (5.607)

ADV 8.754*** 9.485*** 8.336***

(4.427) (4.807) (4.624)

BDINDEP

-0.172**

(1.885)

BDOWN

-0.209**

(2.061)

INSTI

-0.286***

(2.492)

GOVSTRENGTH

-1.454***

(13.561)

Constant -5.085*** -4.529*** -3.556***

(8.535) (7.393) (4.881)

INDUSTRY Yes Yes Yes

YEAR Yes Yes Yes

Adjusted R2 0.179 0.187 0.226

Page 44 of 49

Accep

ted

Man

uscr

ipt

42

Table 6

Regression of CEO compensation on normal CSR and abnormal CSR.

This table presents the OLS regression results of CEO compensation on normal and abnormal

CSR. Our sample consists of 12,705 firm-year observations from 1992 through 2011.

Standard errors are clustered by firm. Coefficients are presented with t-statistics below in

parentheses. ***, ** and * indicate statistical significance at the 1%, 5% and 10% level,

respectively.




in the fiscal year. NORMAL_CSR is the predicted value of CSR from Table 5 model 4.

ABNORMAL_CSR is the residual value of CSR from Table 5 model 4. FIRMSIZE is the






in the fiscal year. BDINDEP is the percentage of independent directors on the board.

Independent directors are directors who are neither current nor former employees of the firm.

BDOWN is the percentage of common stock owned by all directors. INSTI is the percentage

of common stock held by institutional shareholders.

Page 45 of 49

Accep

ted

Man

uscr

ipt

43

Table 6 (continued)

(1) (2) (3) (4)

NORMAL_CSR 0.055*** 0.057*** 0.209*** 0.212***

(5.513) (5.701) (10.623) (10.754)

ABNORMAL_CSR -0.012*** -0.010*** -0.012*** -0.010***

(3.733) (-3.201) (3.575) (3.033)

FIRMSIZE 0.462*** 0.464*** 0.399*** 0.400***

(69.101) (69.423) (41.258) (41.379)

ROA 0.920*** 0.926*** 0.793*** 0.799***

(9.333) (9.412) (8.002) (8.076)

VOLAROA 0.905*** 0.837*** 0.586*** 0.517***

(4.588) (4.252) (2.935) (2.593)

RETURN 0.116*** 0.113*** 0.127*** 0.124***

(6.478) (6.346) (7.071) (6.947)

VOLARET 0.164*** 0.163*** 0.146*** 0.145***

(6.731) (6.702) (6.012) (5.987)

MTB 0.044*** 0.047*** 0.017** 0.020**

(5.762) (6.108) (2.137) (2.441)

TENURE

-0.006***

-0.006***

(9.143)

(9.171)

AGE

0.003***

0.003***

(3.238)

(3.495)

BDINDEP

-0.035 -0.033*

(1.612) (1.872)

BDOWN

-0.055** -0.049**

(2.217) (2.019)

INSTI

-0.128** -0.135**

(2.035) (2.209)

Constant 3.602*** 3.561*** 4.125*** 4.072***

(20.587) (19.465) (22.412) (21.332)

Year dummies Yes Yes Yes Yes

Industry dummies Yes Yes Yes Yes

Adjusted R2 0.522 0.525 0.524 0.527

Page 46 of 49

Accep

ted

Man

uscr

ipt

44

Table 7

Regression of CEO compensation on CSR: Strong Corporate Governance versus Weak

Corporate Governance

This table presents the OLS regression results of CEO compensation on CSR in the sub-

samples of strong and weak corporate governance. Our sample consists of 12,705 firm-year

observations from 1992 through 2011. Standard errors are clustered by firm. Coefficients are

presented with t-statistics below in parentheses. ***, ** and * indicate statistical significance

at the 1%, 5% and 10% level, respectively. The sample is partitioned into two subsamples

according to the median of three corporate governance measures: GOVSTRENGTH (column

1 and 2), G-score (column 3 and 4) and E-index (column 5 and 6) respectively.

GOVSTRENGTH is a corporate governance index derived from the principal components

analysis of the board independence (BDINDEP), board ownership (BDOWN) and

institutional investors ownership (INSTI). G-score is the G-score for corporate governance

from Gompers, Ishii and Metrick (2003), E-index is the corporate governance measure from

Bebchuk, Cohen and Ferrell (2009).




in the fiscal year. CSR is computed as the KLD strengths less KLD concerns for community,

diversity, employee relations, environment, human rights and product. FIRMSIZE is the






in the fiscal year. R&D is calculated as the research and development expenditure divided by

sales. ADV is the advertising expenditure divided by sales.

Page 47 of 49

Accep

ted

Man

uscr

ipt

45

Table 7 (continued)

GOVSTRENGTH G-score E-index

Strong CG

Weak

CG

Strong

CG Weak CG

Strong CG Weak CG

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

CSR -0.010** 0.015

-0.023*** -0.006 -0.020*** -0.005

(2.277) (0.835)

(4.676) (1.464)

(4.879) (1.045)

FIRMSIZE 0.328*** 0.489***

0.476*** 0.489***

0.466*** 0.514***

(17.560) (3.937)

(61.862) (70.977)

(66.700) (66.422)

ROA 1.753*** 1.685***

1.224*** 0.691***

1.067*** 0.844***

(14.490) (2.854)

(8.508) (5.233)

(7.981) (5.936)

VOLAROA 0.508* 1.349

0.918*** 0.812***

0.932*** 0.938***

(1.954) (1.064)

(3.321) (2.891)

(3.515) (3.236)

RETURN 0.079*** 0.073

0.080*** 0.149***

0.105*** 0.105***

(4.635) (0.912)

(3.068) (6.295)

(4.300) (4.135)

VOLARET 0.077** -0.108

0.200*** 0.121***

0.179*** 0.161***

(2.349) (0.577)

(6.004) (3.375)

(5.474) (4.494)

MTB 0.066*** 0.072

0.038*** 0.084***

0.041*** 0.086***

(6.407) (1.311)

(3.492) (8.027)

(4.124) (7.614)

TENURE -0.003*** 0.001

-0.007*** -0.003***

-0.008*** -0.001*

(3.205) (0.128)

(7.994) (3.331)

(9.088) (1.688)

AGE 0.003** 0.019**

0.002 0.004***

0.002* 0.004***

(2.275) (2.352)

(1.378) (3.164)

(1.745) (2.991)

Constant 4.133*** 2.560**

3.306*** 3.228***

3.810*** 2.701***

(22.989) (2.439)

(10.328) (15.329)

(14.816) (11.465)

INDUSTRY Yes Yes

Yes Yes

Yes Yes

YEAR Yes Yes

Yes Yes

Yes Yes

Adjusted R2 0.141 -0.906

0.487 0.566 0.495 0.557

Test the difference in coefficients on CSR between two subsamples

Chi-square

6.85

8.78

7.18

p value (0.005) (0.003) (0.007)

Page 48 of 49

Accep

ted

Man

uscr

ipt

46

Table 8

Regression of CEO compensation on normal CSR and abnormal CSR: Strong Corporate

Governance versus Weak Corporate Governance

This table presents the OLS regression results of CEO compensation on normal and abnormal

CSR in the sub-samples of strong and weak corporate governance. Our sample consists of

12,705 firm-year observations from 1992 through 2011. Standard errors are clustered by firm.

Coefficients are presented with t-statistics below in parentheses. ***, ** and * indicate

statistical significance at the 1%, 5% and 10% level, respectively. The sample is partitioned

into two subsamples according to the median of three corporate governance measures:

GOVSTRENGTH (column 1 and 2), G-score (column 3 and 4) and E-index (column 5 and 6)

respectively. GOVSTRENGTH is a corporate governance index derived from the principal

components analysis of the board independence (BDINDEP), board ownership (BDOWN)

and institutional investors ownership (INSTI). G-score is the G-score for corporate

governance from Gompers, Ishii and Metrick (2003), E-index is the corporate governance

measure from Bebchuk, Cohen and Ferrell (2009).




in the fiscal year. NORMAL_CSR is the predicted value of CSR from Table 5 model 4.

ABNORMAL_CSR is the residual value of CSR from Table 5 model 4. FIRMSIZE is the






in the fiscal year.

Page 49 of 49

Accep

ted

Man

uscr

ipt

47

Table 8 (continued)

GOVSTRENGTH G-score E-index

Strong CG Weak CG

Strong CG Weak CG

Strong CG Weak CG

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

NORMAL_CSR 0.105*** 0.096**

0.070*** 0.032** 0.058*** 0.057***

(6.747) (2.545)

(4.543) (2.486)

(4.354) (3.575)

ABNORMAL_CSR -0.020*** 0.003

-0.025*** -0.008*

-0.023*** -0.009*

(5.955) (0.277)

(4.733) (1.807)

(5.018) (1.823)

FIRMSIZE 0.439*** 0.451***

0.446*** 0.474***

0.442*** 0.490***

(46.928) (21.707)

(43.651) (53.140)

(48.295) (48.249)

ROA 0.888*** 0.669**

1.217*** 0.573***

1.029*** 0.731***

(8.362) (2.317)

(8.406) (4.297)

(7.667) (5.099)

VOLAROA 0.640*** 1.428***

0.795*** 0.787***

0.773*** 0.870***

(2.947) (2.727)

(2.857) (2.795)

(2.910) (3.003)

RETURN 0.106*** 0.257***

0.086*** 0.148***

0.110*** 0.103***

(5.589) (4.262)

(3.284) (6.261)

(4.518) (4.021)

VOLARET 0.176*** 0.029

0.194*** 0.118***

0.172*** 0.159***

(6.758) (0.403)

(5.791) (3.283)

(5.237) (4.422)

MTB 0.036*** 0.016

0.021* 0.080***

0.036*** 0.078***

(4.375) (0.611)

(1.939) (7.511)

(3.536) (6.756)

TENURE -0.003** -0.001

-0.008*** -0.003***

-0.008*** -0.002*

(2.547) (0.082)

(8.688) (3.471)

(9.671) (1.907)

AGE 0.003* 0.014

0.002 0.005***

0.002 0.005***

(1.797) (1.317)

(1.406) (3.303)

(1.406) (3.163)

Constant 3.766*** 3.726***

3.580*** 3.391***

4.114*** 2.911***

(18.220) (8.713)

(10.937) (15.518)

(14.965) (12.036)

INDUSTRY Yes Yes

Yes Yes

Yes Yes

YEAR Yes Yes

Yes Yes

Yes Yes

Adjusted R2 0.523 0.498

0.495 0.569 0.506 0.56

Test the difference in coefficients on NORMAL_CSR between two subsamples

Chi-square

0.005

3.91

0

p value

(0.820)

(0.048)

(0.965)

Test the difference in coefficients on ABNORMAL_CSR between two subsamples

Chi-square

8.672

7.65

5.02

p value (0.005) (0.006) (0.025)

Documents

CEO compensation and corporate social responsibility