+ All Categories
Home > Documents > Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the...

Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the...

Date post: 23-Dec-2016
Category:
Upload: marty
View: 215 times
Download: 2 times
Share this document with a friend
13
Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States Li Sun Marty Stuebs Received: 9 November 2011 / Accepted: 19 November 2012 / Published online: 2 December 2012 Ó Springer Science+Business Media Dordrecht 2012 Abstract Prior research suggests that participating in corporate social responsibility (CSR) activities can lead to higher future productivity. However, the empirical evi- dence is still scarce. The purpose of this study is to examine the relationship between CSR and future firm productivity in the U.S. chemical industry. Specifically, this study examines the relationship between CSR in year t and firm productivity in year (t ? 1), (t ? 2), and (t ? 3). We use Data Envelopment Analysis, a non-parametric method, to measure firm productivity. Results from the regression analysis support a significantly positive rela- tionship between CSR and future firm productivity, sug- gesting that CSR can lead to higher productivity in the chemical industry. The findings add to the validity of the proposition in prior research. Keywords Corporate social responsibility Firm productivity Chemical industry Introduction There are 170 major chemical companies (e.g., DuPont, Dow Chemicals) in the United States operating interna- tionally with more than 2,800 facilities abroad. Hurston (2011) states that the chemical industry is among the largest industries in the U.S., an approximately $720 billion enterprise, accounting for 26 percent of the nation’s Gross Domestic Product (GDP). The U.S. chemical industry also represents about 19 percent of the global chemicals output. Chandler (2005) suggests that chemical firms in the U.S. must have a strong commitment to a healthy environment, because they transform raw materials into products useful to our society. The U.S. chemical industry has always drawn attention from different stakeholders, including the government, communities, and customers, because this industry manu- factures products that directly affect the environment and the quality of human life. Anecdotal evidence (e.g., Huston 2011) suggests that corporate social responsibility (CSR) has become an important part of business practice in the chemical industry. For example, according to the American Chemistry Council 1 (ACC), more and more chemical firms in the U.S. during the past decade have participated in a global CSR program known as Responsible Care. This program encourages chemical firms to engage in more CSR activities including (1) promoting mutual trust between chemical firms and their stakeholders; (2) supporting edu- cation and research on the health, safety, and environ- mental issues surrounding chemical production processes and products; (3) promoting cooperation with the govern- ment and other organizations in the development and implementation of effective regulations; and (4) enhancing accountability of chemical firms. Due to high pressure and constant monitoring from the public, chemical firms in the U.S. have improved their CSR performance substantially. Thus, the chemical industry serves as a good candidate to explore the impact of CSR activities on the performance of chemical firms in this study. L. Sun (&) Department of Accounting, Miller College of Business, Ball State University, Muncie, IN 47306, USA e-mail: [email protected] M. Stuebs Department of Accounting and Business Law, Hankamer School of Business, Baylor University, Waco, TX 76798, USA e-mail: [email protected] 1 http://responsiblecare.americanchemistry.com/. 123 J Bus Ethics (2013) 118:251–263 DOI 10.1007/s10551-012-1579-9
Transcript
Page 1: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

Corporate Social Responsibility and Firm Productivity: Evidencefrom the Chemical Industry in the United States

Li Sun • Marty Stuebs

Received: 9 November 2011 / Accepted: 19 November 2012 / Published online: 2 December 2012

� Springer Science+Business Media Dordrecht 2012

Abstract Prior research suggests that participating in

corporate social responsibility (CSR) activities can lead to

higher future productivity. However, the empirical evi-

dence is still scarce. The purpose of this study is to

examine the relationship between CSR and future firm

productivity in the U.S. chemical industry. Specifically,

this study examines the relationship between CSR in year

t and firm productivity in year (t ? 1), (t ? 2), and (t ? 3).

We use Data Envelopment Analysis, a non-parametric

method, to measure firm productivity. Results from the

regression analysis support a significantly positive rela-

tionship between CSR and future firm productivity, sug-

gesting that CSR can lead to higher productivity in the

chemical industry. The findings add to the validity of the

proposition in prior research.

Keywords Corporate social responsibility �Firm productivity � Chemical industry

Introduction

There are 170 major chemical companies (e.g., DuPont,

Dow Chemicals) in the United States operating interna-

tionally with more than 2,800 facilities abroad. Hurston

(2011) states that the chemical industry is among the

largest industries in the U.S., an approximately $720 billion

enterprise, accounting for 26 percent of the nation’s Gross

Domestic Product (GDP). The U.S. chemical industry also

represents about 19 percent of the global chemicals output.

Chandler (2005) suggests that chemical firms in the U.S.

must have a strong commitment to a healthy environment,

because they transform raw materials into products useful

to our society.

The U.S. chemical industry has always drawn attention

from different stakeholders, including the government,

communities, and customers, because this industry manu-

factures products that directly affect the environment and

the quality of human life. Anecdotal evidence (e.g., Huston

2011) suggests that corporate social responsibility (CSR)

has become an important part of business practice in the

chemical industry. For example, according to the American

Chemistry Council1 (ACC), more and more chemical firms

in the U.S. during the past decade have participated in a

global CSR program known as Responsible Care. This

program encourages chemical firms to engage in more CSR

activities including (1) promoting mutual trust between

chemical firms and their stakeholders; (2) supporting edu-

cation and research on the health, safety, and environ-

mental issues surrounding chemical production processes

and products; (3) promoting cooperation with the govern-

ment and other organizations in the development and

implementation of effective regulations; and (4) enhancing

accountability of chemical firms. Due to high pressure and

constant monitoring from the public, chemical firms in the

U.S. have improved their CSR performance substantially.

Thus, the chemical industry serves as a good candidate to

explore the impact of CSR activities on the performance of

chemical firms in this study.

L. Sun (&)

Department of Accounting, Miller College of Business,

Ball State University, Muncie, IN 47306, USA

e-mail: [email protected]

M. Stuebs

Department of Accounting and Business Law, Hankamer School

of Business, Baylor University, Waco, TX 76798, USA

e-mail: [email protected] 1 http://responsiblecare.americanchemistry.com/.

123

J Bus Ethics (2013) 118:251–263

DOI 10.1007/s10551-012-1579-9

Page 2: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

Vilanova et al. (2009) argue that CSR is related to firm

competitiveness through a learning and innovation cycle.

First, learning takes place when a company integrates CSR

activities into its business process. Then, learning generates

innovative ideas and practices. Finally, the innovative

practices lead to firm competitiveness. In Vilanova et al.

(2009), firm competitiveness includes five dimensions:

(1) financial performance, (2) quality of product/service,

(3) productivity, (4) innovation, and (5) image/reputation.

This study focuses on the productivity dimension of firm

competitiveness. If Vilanova et al. (2009) is valid, we

would expect to observe a positive relationship between

CSR and firm productivity.

The purpose of this study is to examine the relationship

between CSR and firm productivity in the chemical industry

(SIC = 2800–2899). Specifically, we examine the rela-

tionship between CSR in year t and firm productivity in year

(t ? 1), (t ? 2), and (t ? 3), respectively. We obtain CSR

data (for the period 1998–2009) from Kinder, Lydenberg,

and Domini’s (KLD’s) database, and financial data (for the

period 1999–2010) from Compustat database. We merge the

two samples based on the first six digits of the Committee on

Uniform Security Identification Procedures (CUSIP) code.

Since this study focuses on chemical firms, we restrict the

sample to only chemical firms. Regression analysis reveals

that CSR in year t is positively related to firm productivity in

year (t ? 1) and year (t ? 2) at a significant level. Overall,

the results support a positive relationship between CSR and

firm productivity, suggesting that engaging in CSR activities

can make chemical firms more productive in future periods.

Additional results find the impact of CSR on firm produc-

tivity is greater for chemical firms facing intense market

competition.

We use Data Envelopment Analysis (DEA) to measure a

firm’s productivity. Given a collection of points in a mul-

tidimensional space, DEA fits a piecewise linear envelope

or frontier to the given data. The envelope indicates a

normative ideal. Points located on the envelope are opti-

mally efficient, while points below the envelope are inef-

ficient. DEA evaluates all points with respect to their

deviation from the frontier. The values of the points on the

frontier equal one and the values of other points which

operate beneath the frontier are between zero and one.

This study makes several contributions. First, the finding

in this study is consistent with the propositions in Vilanova

et al. (2009) and Porter and Kramer (2006). That is, doing

CSR activities can increase a firm’s competitiveness,

including productivity. Thus, this study provides empirical

evidence to support the validity of the propositions. Second,

Beurden and Gossling (2008, p. 420) recently issued a call

for industry-specific studies to advance the usefulness of

CSR research by stating that ‘‘to continue to have value for

management practice and for the improvement of the

business world, future studies should focus on segments of

groups of firms that practice CSR. In this respect, research in

different industries may be helpful.’’ By focusing on the

chemical industry, this study answers that call. Third,

although many studies examine the relationship between

CSR and financial performance, very few studies examine

the relationship between CSR and firm productivity. Thus,

this paper contributes to the CSR literature. Last, from a

practical perspective, the results should be of interest to

managers who contemplate engaging in socially responsible

activities, investors and financial analysts who assess firm

performance, and policy makers who design and implement

guidelines on CSR.

The remainder of the paper is organized as follows.

‘‘Review of Prior Research and Hypothesis Development’’

section reviews prior research and develops the hypotheses.

‘‘Research Design’’ section describes the research design,

including measurement of dependent variable, primary

independent variable, and empirical specification. ‘‘Sample

Selection and Descriptive Statistics’’ section describes

sample selection and descriptive statistics, while ‘‘Results’’

section reports the results from the regression analysis.

‘‘Conclusion’’ section summarizes the study.

Review of Prior Research and Hypothesis Development

Corporate social responsibility is defined as ‘‘the voluntary

integration of social and environmental concerns into business

operations and into their interaction with stakeholders’’

(European Commission 2002). Vilanova et al. (2009) pro-

pose that the definition of CSR consists of five dimensions,

including vision, community relations, workplace, account-

ability, and marketplace. For example, vision includes

CSR conceptual development, codes, and value within the

organization. Community relations include partnerships with

different stakeholders such as customers, suppliers, etc.

Workplace includes human rights and labor practices within

the organization. Accountability includes the transparency in

communication and financial reporting. Marketplace includes

the relationship between CSR and core business processes

such as sales, purchasing, etc.

The topic of CSR has received increasing attention in

recent years. The practice of CSR is still controversial since it

requires firms to undertake additional investments in CSR.

These CSR investments are often examined through the

economic cost-benefit analytical lens. Some (e.g., Karnani

2010) argue that CSR activities increase costs without suf-

ficient offsetting benefits, hurt performance and compete

with value-maximizing activities. Examples of these addi-

tional costs include making charitable donations, developing

plans for community improvement, and establishing proce-

dures to reduce pollution.

252 L. Sun, M. Stuebs

123

Page 3: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

The majority of CSR studies focus on testing the link

between CSR and financial performance of a firm. Empirical

results are somewhat mixed. For example, Aupperle et al.

(1985) use survey to assess Chief Executive Officer (CEO)’s

perspectives on CSR activities and find a significant and

negative association between CSR and accounting-based

performance measures. Moore (2001) uses eight main

companies in the United Kingdom (U.K.) supermarket

industry and finds a negative relationship between CSR and

financial performance. Nelling and Webb (2009) use the

KLD index as the measure of CSR and return on assets

(ROA) to measure financial performance. They find no evi-

dence that CSR is related to a firm’s financial performance.

Many other CSR-financial performance studies document

a positive relationship between CSR and financial perfor-

mance. Early work by Cochran and Wood (1984) and

McGuire et al. (1988) find a positive link between CSR and

financial performance. Waddock and Graves (1997) use KLD

data to measure CSR performance, and use return on assets

(ROA), return on equity (ROE), and return on sales (ROS) to

measure financial performance, and document two major

findings: (1) better financial performance leads to better future

CSR performance, and (2) better CSR performance leads to

better future financial performance. Tsoutsoura (2004), using

firms selected from the S&P 500 Index from 1996 to 2000,

finds a significant and positive association between KLD data

and financial performance, including ROA, ROE, and ROS. A

recent literature review in Beurden and Gossling (2008)

concludes that the consensus appears to be that a positive

relationship exists between CSR and financial performance,

suggesting that firms that care about their social responsibil-

ities may perform well in today’s society.

Porter and Kramer (2006) argue that CSR has become an

inescapable priority for companies in every country, since it

is much more than a cost or a charitable deed. CSR can bring

opportunities, innovations, and competitive advantages to

companies. If a company establishes an affirmative CSR

agenda and incorporates the agenda into its business practice,

this can generate maximum social and financial benefits to

the company. Vilanova et al. (2009) argue that CSR is related

to competitiveness through a learning and innovation cycle.

First, learning takes place when a company integrates CSR

activities into its business process. Then, learning generates

innovative ideas and practices. Finally, the innovative

practices lead to competitiveness. Vilanova et al. (2009)

propose that a firm’s competitiveness can be grouped into

five dimensions, including (1) financial performance (Hamel

and Prahalad 1989), including conventional measures such

as return on assets, net income; (2) quality of product/service

(Barney 1991); (3) productivity (Porter and Kramer 2006), in

terms of higher outputs and lower inputs; (4) innovation in

product, service or management process (Mintzber 1993);

and (5) image/reputation (Kay 1993).

While most CSR studies examine the impact of CSR on

the first competitiveness dimension—financial performance,

research exploring the impact of CSR on other dimensions is

still limited. This study focuses on the productivity dimen-

sion of the firm competitiveness. If the propositions in Porter

and Kramer (2006) and Vilanova et al. (2009) are valid, we

would expect that participating in CSR activities can lead

to higher productivity. The main hypothesis is stated as

follows:

H1 Corporate social responsibility is positively related to

firm productivity.

Luo and Bhattacharya (2006, p. 2) state that there is no

simple and unconditional relationship between CSR and firm

performance, because firms are not the same in doing CSR

activities. Luo and Bhattacharya (2006, 2009) suggest that

the relationship between CSR and firm performance is better

understood in a firm-specific context or industry-specific

context (e.g., market competition). Following Luo and

Bhattachaya (2006, 2009), this study provides an industry-

specific context (measured as market competition) to

examine the relationship between CSR and firm performance

(productivity). That is, we add another hypothesis with the

moderating role of market competition. For example, in a

highly competitive industry, firms are continuously pressed

to outperform their peers. As a result, they are more likely to

engage in CSR activities, which may bring them future

benefits. We use the Herfindahl competition index to capture

market competition. The Herfindahl index is the sum of

squared market shares of the firms in the industry. For

example, for a market consisting of five firms with market

shares of 30, 30, 20, 10, and 10 percent, the Herfindahl index

is 2,400 (302 ? 302 ? 202 ? 102 ? 102 = 2,400). The

lower the Herfindahl index, the higher the market competi-

tion in the industry.

H2 The impact of CSR on firm’s productivity is greater

for firms facing higher market competition.

Research Design

Measurement of the Dependent Variable—Firm

Productivity

In this study, we use DEA to measure the productivity of a

firm. DEA is a nonparametric model that measures the

productivity of a decision-making unit (DMU) such as a

firm or a school. Charnes et al. (1978, p. 429) describe

DEA as ‘‘a mathematical programing model applied to

observational data that provides a new way of obtaining

empirical estimates of relations that are cornerstones of

modern economics.’’ Cooper et al. (2000) describe some

Corporate Social Responsibility and Firm Productivity 253

123

Page 4: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

characteristics of DEA. First, DEA produces a more flex-

ible and adaptable measure of firm productivity, since DEA

does not require either a prescribed functional form such as

the Cobb–Douglas production function or pre-assigned

weights to each input and output. Second, unlike the typical

parametric approach that compares each DMU to an

average DMU, DEA compares each DMU to the ‘‘best’’

DMU. The term ‘‘best’’ is used here to mean that the

(outputs/inputs) ratio for each DMU is maximized, relative

to all other DMUs. Third, DEA does not require the user to

prescribe weights to be attached to each input and output

variable. DEA determines the ‘‘best’’ input and output

variable weights for each DMU against all other DMUs.

Each DMU’s weights may differ from other DMUs.

Figure 1 shows a simple DEA example with two inputs

(91, 92) and one output (y) for 4 DMUs (A, B, C, D). DEA

fits a piecewise linear envelope or frontier to the given data.

The envelope indicates a normative ideal. Points located on

the envelope are optimally efficient, while points below the

envelope are inefficient. In Fig. 1, zz0 represent the envelope

or productivity frontier. DMUs A and B are on the pro-

ductivity frontier, and thus, their values of the productivity

scores (measured as the output/input ratio) are one. The

values of the productivity scores for DMUs, which operate

beneath the productivity frontier, are between zero and one.

For instance, the productivity score of DMU C is OC/OC0.The productivity score of DMU D is OD/OD0.

DEA has two basic models—the CCR model and the BCC

model. The CCR model, introduced by Charnes, Cooper and

Rhodes (1978), applies only to technologies characterized by

constant returns to scale. The BCC model, proposed by

Banker et al. (1984), extends the CCR model to accommo-

date technologies that exhibit variable returns to scale.

The CCR model estimates the efficiency of DMUs by

solving the following linear program:

Min z

Subject to

Xn

i¼1

kixi� zx0

Xn

i¼1

kiyi� y0

z; ki� 0

where

xi and yi are vectors of observed inputs and outputs,

respectively, for each of n DMUs.

x0 and y0 represent inputs and outputs of the DMU to be

evaluated.

The BCC model estimates the efficiency of DMUs by

solving the following linear program:

Min z

Subject to

Xn

i¼1

kixi� zx0

Xn

i¼1

kiyi� y0

Xn

i¼1

ki ¼ 1

z; ki� 0

where

xi and yi are vectors of observed inputs and outputs,

respectively, for each of n DMUs.

x0 and y0 represent inputs and outputs of the DMU to be

evaluated.

Dyson et al. (2001) suggest using BCC model only when

variable returns to scale are clearly demonstrated in the data.

Banker et al. (2004) also suggest that DEA studies need to

take into account of the economic concept of returns to scale.

Following Dyson et al. (2001) and Banker et al. (2004), we

test the returns to scale (RTS) of the data before deciding

which DEA model (CCR or BCC) to use in this study.

In DEA, there are three alternative ways to identify the

returns to scale: (1) a primal approach (Banker 1984), (2) a

dual approach (Banker et al. 1984), and (3) a nesting

approach (Fare et al. 1985). Banker et al. (1996) examine all

three alternatives and conclude that all three are equivalent

methods for determining returns to scale in DEA. Thus, we

use the first method (Banker 1984) to test the returns to scale.

Banker (1984) shows that we can determine the returns to

scale in (x0, x0) by examining the values ofPn

i¼1 ki, obtained

from the CCR model when the sum of lambda is uniquely

determined. The Returns to Scale Theorem is as follows:Fig. 1 A basic DEA model with two inputs and one output

254 L. Sun, M. Stuebs

123

Page 5: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

IfPn

i¼1 ki ¼ 1, in any alternate optima, then constant

returns to scale prevail.

IfPn

i¼1 ki [ 1, in any alternate optima, then decreasing

returns to scale prevail.

IfPn

i¼1 ki\1, in any alternate optima, then increasing

returns to scale prevail.

Using the sample consisting of 902 observations, we run

DEA for each year and calculate the sum of lambdaPni¼1 ki

� �for each DMU. The majority (approximately

84 %) of the DMUs indicate either decreasing or increas-

ing returns to scale. Thus, we choose to use the BCC model

to compute firm productivity in this study.

Using DEA requires identifying input and output vari-

ables. Following Bowlin (1999), we use three input vari-

ables: cost of goods sold (COGS), selling, general and

administrative expenses (XSGA) and total assets (ASSETS).

Cost of goods sold (COGS) represents all costs directly

allocated to production, such as direct materials, direct labor

and overhead. Selling, general and administrative expenses

(XSGA) represents non-production expenses incurred in the

regular course of business. Total assets (ASSETS) represents

current assets plus long-term assets. We also use three output

variables: sales (SALES), operating cash flows (OCF) and

market share (MKSHARE). Sales (SALES) represents sales

after any discounts, returned sales, and allowances for which

credit is given to customers. Operating cash flows (OCF)

represents the net change in cash in the operating activities.

Market share (MKSHARE) represents the percentage of a

market (in terms of sales) accounted for by a specific com-

pany. This measure enables managers to judge not only total

market growth or decline but also trends in customers’

selections among competitors. Market share is a key indi-

cator of market competitiveness (Vilanova et al. 2009).

In summary, we use the following variables in DEA.

Input variables Output variables

Cost of goods sold (COGS) Sales (SALES)

Selling, general and administrative

expenses (XSGA)

Operating cash flows

(OCF)

Total assets (ASSETS) Market share

(MKSHARE)

In addition to Bowlin (1999), many academic studies of

business use DEA to calculate relative productivity scores.

Some studies suggest that the relative productivity score

calculated by DEA is a superior performance measure,

relative to other firm performance measures. For example,

Feroz et al. (2003) suggest that DEA can produce a con-

sistent and reliable measure of the operational efficiency of

the firms in the U.S. oil and gas industry. In Feroz et al.

(2008), the authors argue that traditional measures like

return on assets (ROA) and return on investment (ROI)

may generate inconclusive performance results, because

they are measure-specific and can be affected by non-

value-added factors. Instead, Feroz et al. (2008) suggest

that incorporating traditional variables, such as sales and

cost of goods sold, into a DEA model may produce a more

comprehensive measure of firm performance. In addition,

Dopuch et al. (2010) use both stochastic frontier efficiency

(SFE) and DEA to measure the relative productivity of the

audit service of public accounting firms. They find that SFE

is not powerful enough to detect inefficiencies of the audit

service. Dopuch et al. (2010) suggest that DEA is a better

measure to capture productivity of DMUs.

Measurement of the Primary Independent Variable—

Corporate Social Responsibility

Kinder, Lydenberg, and Domini (KLD), a Boston-based

consulting firm, has been actively providing rating data on

CSR since 1991. KLD data are an influential measure of

CSR. While many investment managers rely on KLD data

when making social screening, the KLD data are also

frequently used in the academic literature. It is ‘‘the largest

multidimensional corporate social performance database

available to the public and is used extensively in research

on corporate social performance’’ (Deckop et al. 2006,

p. 334). KLD accumulates CSR information for more firms

than other CSR data sources. It has become ‘‘the de facto

corporate social performance research standard at the

moment’’ (Waddock 2003, p. 369).

KLD provides rating data for approximately eighty

variables in seven qualitative areas for each selected firm.

The seven areas include community, corporate governance,

diversity, employee relations, environment, human rights,

and product. For each qualitative variable, positive ratings

indicate strengths, and negative ratings indicate concerns.

For example, the environment area contains six strength

items (beneficial products, pollution prevention, recycling,

clean energy, property plant and equipment, and other

strengths) and six concern items (hazardous waste, regu-

latory problems, ozone depleting chemicals, substantial

emissions, agriculture chemicals, and other concerns). In

addition to these seven qualitative areas, KLD also evalu-

ates six controversial issues that include alcohol, gambling,

firearms, military, nuclear power, and tobacco activities.

Involvement in any of these six controversial issues results

in a negative rating.

Consistent with prior research (e.g., Dhaliwal et al.

2011; Chen et al. 2008; Cho et al. 2006; Deckop et al.

2006; Nelling and Webb 2009; Ruf et al. 2001; Johnson

and Greening 1999; Griffin and Mahon 1997; Shropshire

Corporate Social Responsibility and Firm Productivity 255

123

Page 6: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

and Hillman 2007; Waddock and Graves 1997; Graves and

Waddock 1994), we subtract total concerns from total

strengths and assign equal importance/weight to each area

in calculating the KLD index score. This approach is also

suggested by KLD2. More specifically, the KLD index

score is computed as follows:

Empirical Specification

Since the values of productivity scores by DEA are between 0

and 1, we use the Tobit regression model (Luo and Donthu

2006) to test the relationship between CSR and firm produc-

tivity (DEA). The DEA score is the Tobit regression model’s

dependent variable, while the CSR performance is the inde-

pendent variable of interest. Consistent with Cochran and

Wood (1984) and other prior studies, additional variables are

included to control for firm size (natural log of total assets) and

risk (leverage ratio). We also include return on assets (ROA)

to control for firm performance, because firms with high

productivity may also have high ROA (Bowlin 1999).

We use the following three models to examine the rela-

tionship between CSR in year t and firm productivity in year

(t ? 1), (t ? 2), and (t ? 3), respectively. Model 1 tests the

relationship between CSR in year t and DEA in the sub-

sequent year. Model 2 tests the relationship between CSR in

year t and DEA in year (t ? 2), while Model 3 tests the

relationship between CSR in year t and DEA in year (t ? 3).

Model 1 : DEAi;ðtþ1Þ ¼a0þ a1 �KLDi;tþ a2 �COMPINTi;t

þ a3 �KLDi;t� COMPINTi;t

þ a4 �LATi;ðtþ1Þ þ a5 �ROAi;ðtþ1Þ

þ a6 �LEVi;ðtþ1Þ þ ei

Model 2 : DEAi;ðtþ2Þ ¼ a0þ a1 �KLDi;t

þ a2 �COMPINTi;tþ a3 �KLDi;t

� COMPINTi;tþ a4 �LATi;ðtþ2Þ

þ a5 �ROAi;ðtþ2Þ þ a6 �LEVi;ðtþ2Þ þ li

Model 3 : DEAi;ðtþ3Þ ¼ a0þ a1 �KLDi;tþ a2 �COMPINTi;t

þ a3 �KLDi;t� COMPINTi;t

þ a4 �LATi;ðtþ3Þ þ a5 �ROAi;ðtþ3Þ

þ a6 �LEVi;ðtþ3Þ þ £i

where

KLDi,t corporate social responsibility (CSR) score

of firm i in year t;

COMPINTt market competition intensity (measured by

the Herfindahl industry concentration index)

in year t. This index is the sum of squared

market shares of the firms in the industry. The

lower the Herfindahl industry concentration

index, the higher the competition intensity;

LATi,t natural log of total assets (Compustat Item

#6) of firm i in year t;

ROAi,t return on asset ratio [operating income (Comp-

ustat Item#13)/total assets (Compustat Item #6)]

of firm i in year t;

LEVi,t leverage ratio [total liabilities (Compustat

Item #9 ? Compustat Item #34)/total assets

(Compustat Item #6)] of firm i in year t

To test the main hypothesis (H1), we analyze a1, the

coefficient on KLD. To the extent that stronger CSR per-

formance leads to higher future productivity, we expect a

positive and significant coefficient (a1) on KLD. To test the

moderating hypothesis (H2), we analyze the coefficient

(a3) on the interaction of KLD and COMPINT. The coef-

ficient a3 measures the moderation effect. Since lower

KLD ¼ ðTotal strengths of Community� Total concerns of CommunityÞþ ðTotal strengths of Corporate Governance� Total concerns of Corporate GovernanceÞþ ðTotal strengths of Diversity� Total concerns of DiversityÞþ ðTotal strengths of Employee Relations� Total concerns of Employee RelationsÞþ ðTotal strengths of Environment� Total concerns of EnvironmentÞþ ðTotal strengths of Human Rights� Total concerns of Human RightsÞþ ðTotal strengths of Product� Total concerns of ProductÞ� Any concerns of Alcohol� Any concerns of Gambling� Any concerns of Firearm

� Any concerns of Military� Any concerns of Nuclear Power� Any concerns of Tobacco

2 For a complete listing of strengths and concerns of KLD variables,

please visit www.kld.com.

256 L. Sun, M. Stuebs

123

Page 7: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

COMPINT indicates higher market competition intensity,

we expect a negative and significant a3 in the regression

analysis.

Sample Selection and Descriptive Statistics

We begin our sample selection process by downloading

KLD data, including the seven major areas and six con-

troversial issues, during the period from 1998 to 2009. The

KLD sample consists of 24,941 firm-year observations.

Next, we use Compustat to obtain financial statement data,

which include total cost of goods sold, total general and

administrative expenses, total assets, total sales, total lia-

bilities, total net value of property, plant and equipment,

total gross value of property, plant and equipment, and total

cash flows from operating activities. The Compustat sam-

ple consists of 65,536 observations during the period from

1999 to 2010. We merge the two samples, based on the first

six digits of the Committee on Uniform Security Identifi-

cation Procedures (CUSIP). A few observations are lost

due to missing values. Since this study focuses on chemical

firms, we restrict the sample to chemical firms. The final

sample used in model 1 to test the relationship between

CSR in t and DEA in year (t ? 1) consists of 902 firm-year

observations. The final sample used in model 2 to test the

relationship between CSR in t and DEA in year (t ? 2)

consists of 783 firm-year observations. The final sample

used in model 3 to test the relationship between CSR in

t and DEA in year (t ? 3) consists of 660 firm-year

observations. In all three models, we run the BCC model of

DEA to calculate productivity scores for each year. Table 1

provides the sample distribution and descriptive statistics

for sample firms.

Panel A, Panel B, and Panel C of Table 1 reports the

mean values of CSR performance (KLD), market compe-

tition intensity scores (COMPINT), DEA scores (DEA),

cost of goods sold (COGS), general, selling and adminis-

trative expenses (XSGA), total assets (ASSETS), total sales

(SALES), cash flows from operating activities (OCF),

market share (MKSHARE), leverage ratio (LEV), and the

age of long-term assets (ASSETAGE) by year for model 1,

2 and 3, respectively. In model 1 (Panel A), the mean value

of the CSR performance (KLD) in 1998 is -0.68 and the

mean value of market competition intensity scores in 1998

is 278.38. The mean value of the DEA scores in 1999 is

0.95. The mean value of KLD is -0.49 for the entire period

from 1998 to 2009. In model 2 (Panel B), the mean value of

the CSR performance (KLD) in 1998 is 0 and the mean

value of market competition intensity scores in 1998 is

289.12. The mean value of the DEA scores in 2000 is 0.97.

The mean value of KLD is -0.43 for the entire period from

1998 to 2008. In model 3 (Panel C), the mean value of

KLD in 1998 is 0.03 and the mean value of market com-

petition intensity scores in 1998 is 302.32. The mean value

of the DEA scores in 2001 is 0.97. The mean value of KLD

is -0.39 for the entire period from 1998 to 2007. DEA

assigns weights to input and output variables. In all 3

models, cost of goods sold (COGS) on average has the

highest weight, relative to other input variables. For output

variables, market share (MASHARE) on average has the

highest weight, relative to other output variables.

Panel A of Table 2 provides the correlation matrices for

the key variables in Model 1 testing the relationship

between CSR in year t and DEA in year (t ? 1). Those

variables include KLDt, DEAt?1, COGSt?1, XSGAt?1,

ASSETSt?1, SALESt?1, OCFt?1, MKSHAREt?1, LEVt?1,

and ASSETAGEt?1. For each pair of variables, the Pearson

and Spearman correlation coefficients and related p values

are provided. Both Spearman and Pearson correlations

indicate a positive and significant (p \ 0.0001; p \ 0.0001)

association between KLDt and DEAt?1. Panel B of Table 2

provides the correlation matrices for the key variables in

Model 2 testing the relationship between CSR in year t

and DEA in year (t ? 2). Both Spearman and Pearson

correlations indicate a positive and significant (p \ 0.0001;

p \ 0.0001) association between KLDt and DEAt?2. Panel

C of Table 2 provides the correlation matrices for the key

variables in Model 3 testing the relationship between CSR in

t and DEA in year (t ? 3). Both Spearman and Pearson

correlations indicate a positive and significant (p \ 0.0001;

p \ 0.0001) association between KLDt and DEAt?3. The

above evidence provides initial support for Hypothesis 1. In

all panels of Table 2, input and output variables in DEA are

highly correlated. For example, COGS is significantly pos-

itively related to SALES. According to Dyson et al. (2001),

some input and output variables are naturally correlated in

DEA studies.

Results

We predict a positive and significant relationship between

CSR and firm productivity in the chemical industry. That

is, the coefficient on KLD (a1) is expected to be positive

and significant. For the moderating hypothesis (H2), we

expect a negative and significant a3 in the regression

analysis. The regression model also includes three addi-

tional variables to control for size (LAT), performance

(ROA), and risk (LEV).

We summarize the Tobit regression results from the

three models in Table 3. For Model 1, as shown in Table 3,

the coefficient on KLD (a1) is positive (0.0509) and sta-

tistically significant (p = 0.0013), supporting a positive

relationship between CSR and subsequent firm productiv-

ity. This suggests that chemical firms with better CSR

Corporate Social Responsibility and Firm Productivity 257

123

Page 8: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

Table 1 Descriptive statistics of sample firms chemical industry (SIC = 2800–2899)

Years N KLD COMPINT DEA COGS XSGA ASSETS SALES OCF MKSHAEE LEV ASSETAGE

Mean Mean Years Mean Mean Mean Mean Mean Mean Mean (%) Mean Mean

Panel A: sample distribution by year [KLDt ? DEA(t?1)]

1998 28 -0.68 278.38 1999 0.95 $3,207.13 $2,574.09 $9,119.73 $7,613.78 $1,479.11 0.58 0.58 0.53

1999 32 0.00 289.12 2000 0.97 $3,761.31 $2,593.92 $9,338.73 $8,363.64 $1,344.08 0.59 0.59 0.52

2000 35 0.09 302.32 2001 0.97 $3,605.92 $2,663.70 $10,143.38 $8,198.08 $1,435.36 1.28 0.58 0.52

2001 51 -0.16 313.84 2002 0.91 $2,556.12 $2,027.66 $7,830.31 $5,988.71 $9,62.80 0.90 0.57 0.55

2002 51 0.00 313.44 2003 0.94 $3,025.83 $2,305.15 $10,162.35 $6,929.56 $1,227.10 0.99 0.54 0.53

2003 112 -0.53 328.86 2004 0.85 $1,702.60 $1,260.98 $5,524.85 $3,341.17 $6,21.15 0.51 0.52 0.51

2004 114 -0.89 318.14 2005 0.85 $1,922.71 $1,339.26 $5,508.07 $4,207.03 $6,23.30 0.52 0.54 0.49

2005 103 -0.63 336.98 2006 0.87 $2,322.37 $1,577.73 $7,016.17 $5,012.96 $8,35.92 0.56 0.53 0.50

2006 101 -0.68 370.82 2007 0.86 $2,363.95 $1,744.99 $7,605.70 $5,335.71 $9,42.81 0.57 0.58 0.48

2007 100 -0.46 355.66 2008 0.85 $2,637.01 $1,870.99 $7,886.62 $5,859.16 $1,039.77 0.58 0.57 0.48

2008 97 -0.54 403.08 2009 0.84 $2,152.23 $1,586.89 $8,436.42 $4,938.80 $941.03 0.56 0.52 0.45

2009 78 -0.33 419.08 2010 0.87 $2,852.04 $2,289.34 $11,573.68 $7,027.71 $1,278.72 0.73 0.51 0.43

902 -0.49 346.19 0.87 $2,443.26 $1,797.72 $7,842.74 $5,548.72 $957.09 0.69 0.55 0.49

Panel B: sample distribution by year [KLDt ? DEA(t?2)]

1998 30 0.00 289.12 2000 0.97 $3,986.93 $2,755.84 $9,848.66 $8,877.13 $1.432.00 1.40 0.60 0.51

1999 33 0.03 302.32 2001 0.97 $3,747.41 $2,571.02 $10,043.09 $8,256.88 $1,457.20 1.29 0.60 0.51

2000 35 0.09 313.84 2002 0.97 $3,377.91 $2,698.17 $10,422.93 $8,256.16 $1,338.32 1.23 0.61 0.50

2001 50 -0.12 313.45 2003 0.93 $2,966.28 $2,189.94 $9,710.46 $6,686.31 $1,188.11 0.95 0.57 0.53

2002 50 0.06 328.86 2004 0.93 $3,410.21 $2,635.95 $11,430.44 $7,933.18 $1,343.75 1.04 0.52 0.53

2003 113 -0.46 318.14 2005 0.84 $1,910.65 $1,345.92 $5,525.30 $4,205.51 $626.61 0.52 0.55 0.49

2004 109 -0.88 336.98 2006 0.85 $2.169.64 $1,437.81 $6,590.32 $4,702.77 $784.49 0.53 0.54 0.48

2005 96 -0.58 370.82 2007 0.87 $2,476.04 $1,834.91 $7,984.39 $5,601.68 $992.06 0.60 0.57 0.47

2006 97 -0.60 355.66 2008 0.85 $2,712.60 $1,925.95 $8,121.81 $6,031.89 $1,072.92 0.60 0.56 0.48

2007 91 -0.56 403.38 2009 0.85 $2,303.64 $1,690.94 $9,007.05 $5,272.15 $1,003.44 0.60 0.53 0.44

2008 79 -0.35 419.79 2010 0.88 $2,764.29 $1,999.15 $1,096.23 $6,369.32 $1,119.72 0.66 0.53 0.43

783 -0.43 350.37 0.88 $2,641.83 $1,905.36 $3,381.36 $5,941.70 $1,021.56 0.72 0.56 0.48

Panel C: sample distribution by year [KLDt ? DEA(t ? 3)]

1998 31 0.03 302.32 2001 0.97 $3,959.51 $2.724.91 $10,582.12 $8,730.07 $1,541.57 1.37 0.61 0.50

1999 33 0.03 313.84 2002 0.97 $3,714,32 $2,587.25 $10.427.19 $8,306.81 $1,370.44 1.24 0.62 0.49

2000 35 0.11 313.44 2003 0.96 $4,062.26 $2828.86 $12,704.47 $8,980.22 $l,618.96 1.28 0.58 0.49

2001 50 -0.12 328.86 2004 0.93 $3,315.63 $2,466.09 $10,800.65 $7,562.45 $1,281.48 0.99 0.55 0.53

2002 43 0.23 318.14 2005 0.93 $4,012.53 $2,935.05 $11,928.79 $9,078.68 $1,413.69 1.12 0.55 0.51

2003 104 -0.47 336.98 2006 0.85 $2,237.05 $1,551.52 $6,870.14 $4,878.69 $818.45 0.55 0.54 0.49

2004 101 -0.83 370.82 2007 0.86 $2,329.95 $1,742.61 $7,555.93 $5,294.06 $940.66 0.56 0.53 0.47

2005 92 -0.50 355.66 2008 0.86 $2,848.36 $2,029.74 $8,547.57 $6,347.35 $1,131.36 0.63 0.55 0.47

2006 87 -0.62 403.38 2009 0.87 $2,406.34 $1,765.05 $9,412.29 $5,509.28 $1,051.97 0.62 0.52 0.45

2007 79 -0.48 419.08 2010 0.85 $2,780.66 $2,001.29 $10,128.99 $6,389.75 $1,121.27 0.66 0.52 0.43

660 -0.39 357.32 0.89 $2,886.26 $2,073.95 $9,261.35 $6,489.55 $1,130.18 0.77 0.55 0.48

KLDi,t corporate social responsibility (CSR) score of firm i in year t

COMPINTt market competition intensity (measured by the Herfindahl industry concentration index) in year t. This index is the sum of squared market shares of the firms in the

industry. The lower the Herfindahl industry concentration index, the higher the competition intensity

DEAi,t relative productivity score calculated by the BCC model of Data Envelopment Analysis (DEA) of firm i in year t

COGSi,t total cost of goods sold (Compustat Item #41) of firm i in year t

XSGAi,t total selling, general and administrative expenses (Compustat Item #189) of firm i in year t

ASSETSi,t total assets (Compustat Item #6) of firm i in year t

SALESi,t total net sales (Compustat Item #12) of firm i in year t

OCFi,t total net cash flows from operating activities (Compustat Item #308) of firm i in year t

MKSAHREi,t total net sales (Compustat Item #12) of firm i in year t/total market sales (chemical industry) in year t

LEVi,t leverage ratio [total liabilities (Compustat Item #9 ? Compustat Item #34)/total assets (Compustat Item #6)] of firm i in year t

ASSETAGEi,t net property, plant and equipment (Compustat Item #8)/gross property, plant and equipment (Compustat Item #7) of firm i in year t

258 L. Sun, M. Stuebs

123

Page 9: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

Table 2 Correlations among Selected Variables Chemical Industry (SIC = 2800–2899)

Panel A: Correlations [KLDt ? DEA(t?1)] Observations = 902

KLDt DEA(t?1) COGS(t?1) XSGA(t?1) ASSETS(t?1) SALES(t?1) OCF(t?1) MKSHARE(t?1) LEV(t?1) ASSETAGE(t?1)

0.1916 0.1520 0.3756 0.2802 0.3136 0.3669 0.2911 -0.1545 0.1295

KLDt (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001)

0.2250 0.1806 0.2044 0.1782 0.2271 0.2185 0.2298 0.0030 0.0139

DEA(?1) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.9291) (0.6769)

0.0073 0.2662 0.5282 0.6626 0.8415 0.5916 0.8235 0.1103 -0.0734

COGS(t?1) (0.8268) (\.00011 (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.0009) (0.0275)

0.1346 0.1766 0.7699 0.8977 0.8996 0.9571 0.8869 0.0027 0.1828

XSGA(t?1) (\.0001) (\.0001) (\.0001 (\.0001) (\.0001) (\.0001) (\.0001) (0.9345) (\.0001)

0.0713 0.2006 0.8743 0.9232 0.9196 0.9194 0.8850 0.0166 0.1293

ASSETS(t?1) (\0.0322) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.6185) (\.0001)

0.0624 0.3073 0.9497 0.9004 0.9561 0.9173 0.9817 0.0531 0.0786

SALES(t?1) (0.0612) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.111) (0.0182)

0.1686 0.3671 0.7925 0.8342 0.8812 0.8915 0.9005 -0.0090 0.1556

OCF(t?1) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.7884) (\.0001)

0.0759 0.3214 0.9447 0.8987 0.9561 0.9949 0.8892 0.0567 0.0911

MKSHARE(t?1) (0.0226) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.0889) (0.0062)

-0.2069 0.0054 0.4146 0.2660 0.2944 0.3328 0.1475 0.3297 -0.1964

LEV(t?1) (\.0001) (0.8706) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001)

0.1512 0.0127 -0.0165 0.1995 0.2102 0.1104 0.2265 0.1299 -0.2095

ASSETAGE(t?1) (\.0001) (0.7031) (0.6199) (\.0001) (\.0001) (0.0009) (\.0001) (\.0001) (\.0001)

Panel B: Correlations [KLDt ? DEA(t?2)] Observations \783

KLDt DEA(t?2) COGS(t?2) XSGA(t?2) ASSETS(t?2) SALES(t?2) OCF(t?2) MKSHARE(t?2) LEV(t?2) ASSETAGE(t?2)

0.1593 0.1488 0.3694 0.2780 0.3089 0.3664 0.2963 -0.1444 0.1304

KLD (\.00011 (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.0003)

0.2450 0.2368 0.2251 0.2128 0.2699 0.2518 0.2795 0.0361 0.0534

DEA(t?2) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.3135) (0.1355)

0.0180 0.3065 0.5297 0.6637 0.8444 0.5854 0.8293 0.1018 -0.0529

COGS(t?2) (0.61531 (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.0043) (0.1393)

0.1364 0.2261 0.7699 0.8979 0.8982 0.9621 0.8890 0.0042 0.2101

XSGA(t?2) (0.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.9057) (\.0001)

0.0756 0.2522 0.8721 0.9258 0.9185 0.9189 0.8908 0.0095 0.1551

ASSETS(t?2) (0.03441 (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.7909) (\.0001)

0.0716 0.3473 0.9472 0.9037 0.9581 0.9150 0.9859 0.0449 0.1060

SALES(t?2) (0.0452) (\.0001) (\.0001) (\.0001) (\.00011 (\.0001) (\.0001) (0.2094) (0.0030)

0.1727 0.3935 0.7839 0.8411 0.8826 0.8888 0.8999 -0.0175 0.1876

OCT(t?2) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.6244) (\.0001)

0.0881 0.3634 0.9430 0.9028 0.9585 0.9956 0.8868 0.0477 0.1167

MKSHARE(t?2) (0.0136) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.1822) (0.0011)

-0.2008 0.0436 0.4135 0.2555 0.2831 0.3248 0.1327 0.3231 -0.2078

LEV(t?2) (\.0001) (0.2231) (\.0001) (\.0001) (\.0001) (\.0001) (0.0002) (\.0001) (\.0001)

0.1502 0.0128 0.0235 0.2559 0.2547 0.1624 0.2878 0.1782 -0.2147

ASSETAGE(t?2) (\.0001) (0.7010) (0.5116) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001)

Panel C: Correlations [KLDt ? DEA(t?3)] Observations \ 660

KLDt DEA(t?3) COGS(t?3) XSGA(t?3) ASSETS(t?3) SALES(t?3) OCF(t?3) MKSHARE(t?3) LEV(t?3) ASSETAGE(t?3)

0.1824 0.1467 0.3527 0.2742 0.2988 0.3515 0.2875 -0.1407 0.1249

KLDt (\.0001) (0.0002) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.0003) (0.0013)

0.2699 0.2538 0.2270 0.2216 0.2817 0.2597 0.2913 0.0343 0.0468

D EA(t?3) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.1056) (0.2299)

0.0263 0.3626 0.5236 0.6570 0.8417 0.5792 0.8285 0.1157 -0.0489

C0GS(t?3) (0.4993) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.0029) (0.2100)

0.1414 0.2404 0.7604 0.9012 0.8973 0.9643 0.8892 -0.0001 0.2246

Corporate Social Responsibility and Firm Productivity 259

123

Page 10: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

performance have higher productivity in the subsequent

year, supporting the main hypothesis (H1). For the mod-

erating hypothesis (H2), the coefficient (a3) on the inter-

action of KLD and COMPINT is negative (-0.0001) and

significant (p = 0.0089). Thus, H2 is also supported in

Model 1.

Model 2 tests the relationship between CSR in year t

and DEA in year (t ? 2). As shown in Table 3, the coeffi-

cient on KLD (a1) is positive (0.0360) and statistically

significant (p = 0.0366), supporting a positive relationship

between CSR in year t and firm productivity in year (t ? 2).

This suggests that chemical firms with better CSR perfor-

mance have higher productivity in year (t ? 2), supporting

the main hypothesis (H1). For the moderating hypothesis

(H2), the coefficient (a3) on the interaction of KLD and

COMPINT is negative (-0.0001) and marginally signifi-

cant (p = 0.0967). Thus, H2 is also supported in Model 2.

Model 3 tests the relationship between CSR in year

t and DEA in year (t ? 3). As shown in Table 3, the

coefficient on KLD (a1) is positive (0.0119) and insig-

nificant (p = 0.5037) in the regression model. For the

moderating hypo-thesis (H2), the coefficient (a3) on

the interaction of KLD and COMPINT is negative and

insignificant (p = 0.8015). Thus, results do not support a

significant relationship between CSR in year t and produc-

tivity (DEA) in year (t ? 3)

Additional results in Table 3 indicate a significantly

positive relation between firm productivity (DEA) and firm

performance (ROA) and risk (LEV). The above positive

associations suggest that (1) chemical firms with higher

ROA also demonstrate higher productivity than firms with

lower ROA. This is consistent with Bowlin (1999); and (2)

firms with higher risk (LEV) have higher productivity than

firms with lower risk. Tobit regression does not compute an

R2 or pseudo-R2. Table 3 reports the adjusted R2 from

ordinary least squares (OLS) regression for each of the

three models.

Conclusion

In this study, we examine the relationship between CSR and

firm productivity in the U.S. chemical industry. Specifically,

we examine the relationship between CSR in year t and firm

productivity in year (t ? 1), (t ? 2), and (t ? 3), respec-

tively. We use DEA, a non-parametric method, to measure

firm productivity. Regression analysis reveals that CSR in

year t is positively related to firm productivity in year (t ? 1)

Table 2 continued

Panel C: Correlations [KLDt ? DEA(t?3)] Observations \ 660

KLDt DEA(t?3) COGS(t?3) XSGA(t?3) ASSETS(t?3) SALES(t?3) OCF(t?3) MKSHARE(t?3) LEV(t?3) ASSETAGE(t?3)

XSGA(t?3) (0.0003) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.9985) (\.0001)

0.0EO9 0.2616 0.8670 0.9229 0.9188 0.9192 0.8967 0.0139 0.1698

ASSETS(t?3) (0.0377) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.7214) (\.0001)

0.0830 0.3555 0.9420 0.9036 0.9595 0.9148 0.9378 0.0542 0.1180

SALES(t?3) (0.0330) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.1644) (0.0024)

0.1802 0.3861 0.7803 0.8503 0.8908 0.8933 0.9036 -0.0133 0.2030

OCF(t?3) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.7341) (\.0001)

0.1007 0.3724 0.9381 0.90233 0.9599 0.9955 0.8917 0.0573 0.1266

MKSHARE(t?3) (0.0096) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (0.1417) (0.001)

-0.1864 0.0349 0.4501 0.2737 0.3023 0.3503 0.1568 0.3482 -0.2101

LEV(t?3) (\.0001) (0.3708) (\.0001) (\.0001) (\.0001) (\.0001) (0.0002) (\.0001) (\.0011)

0.1772 0.0128 0.0105 0.2721 0.2562 0.1668 0.2960 0.1804 -0.2430

ASSETAGE(t?3) (\.0001) (0.7420) (0.7878) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001) (\.0001)

Pearson correlation is above and Spearman correlation is below the diagonal

p values based on two-tailed tests are in parentheses

DEA is run for each year

KLDi,t corporate social responsibility (CSR) score of firm i in year t

DEAi,t relative productivity score calculated by the BCC model of Data Envelopment Analysis (DEA) of firm i in year t

COGSi,t total cost of goods sold (Compustat Item #41) of firm i in year t

XSGAi,t total selling, general and administrative expenses (Compustat Item #189) of firm i in year t

ASSETSi,t total assets (Compustat Item #6) of firm i in year t

SALESi,t total net sales (Compustat Item #12) of firm i in year t

OCFi,t total net cash flows from operating activities (Compustat Item #308) of firm i in year t

MKSAHREi,t total net sales (Compustat Item #12) of firm i in year t/total market sales (chemical industry) in year t

LEVi,t leverage ratio [total liabilities (Compustat Item #9 ? Compustat Item #34)/total assets (Compustat Item #6)] of firm i in year t

ASSETAGEi,t net property, plant and equipment (Compustat Item #8)/gross property, plant and equipment (Compustat Item #7) of firm i in year t

260 L. Sun, M. Stuebs

123

Page 11: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

and year (t ? 2) at a significant level. Overall results support

a positive relationship between CSR performance and firm

productivity, suggesting that stronger CSR performance

leads to higher productivity among U.S. chemical firms.

Additional results also suggest that the impact of CSR

on firm’s productivity is greater for chemical firms facing

intense market competition.

The primary contribution of this study is to provide

initial empirical evidence to support the validity of the

hypotheses raised in prior research (e.g., Vilanova et al.

2009). That is, engaging in CSR activities can lead to

higher firm productivity. The findings of this study also

have significant managerial implications. On one hand, this

study can improve managers’ understanding of why CSR

matters in the U.S. chemical industry. That is, actively

engaging in CSR activities can make chemical firms more

productive in future years, consistent with the notion that

doing good (CSR) can bring future benefits to chemical

firms. On the other hand, the competition in the chemical

industry is fierce, because profits are often linked to the

protection of patents. Such protection is very limited on

time. After the protection of patent is expired, other

chemical firms can manufacture the same products. It is

critical for chemical firms, especially the original manu-

facturers, to main a high level of productivity, to compete

with their competitors after the protection of patents expire.

Thus, the findings from this study may explain why

chemical firms in the U.S. have actively engaged in CSR

activities since the last decade, because CSR can lead to

higher productivity.

Table 3 Tobit Regression Analysis

Model 1: DEAi;ðtþ1Þ ¼ a0 þ a1 � KLDi;t þ a2 � COMPINTi;t þ a3 � KLDi;t � COMPINTi;t þ a4 � LATi;ðtþ1Þ

þ a5 � ROAi;ðtþ1Þ þ a6 � LEVi;ðtþ1Þ þ ei

Model 2: DEAi;ðtþ2Þ ¼ a0 þ a1 � KLDi;t þ a2 � COMPINTi;t þ a3 � KLDi;t � COMPINTi;t þ a4 � LATi;ðtþ2Þ

þ a5 � ROAi;ðtþ2Þ þ a6 � LEVi;ðtþ2Þ þ li

Model 3: DEAi;ðtþ3Þ ¼ a0 þ a1 � KLDi;t þ a2 � COMPINTi;t þ a3 � KLDi;t � COMPINTi;t þ a4 � LATi;ðtþ3Þ

þ a5 � ROAi;ðtþ3Þ þ a6 � LEVi;ðtþ3Þ þ £i

Model 1 Model 2 Model 3

[KLDt ? DEA(t?1)] [KLDt ? DEA(t?2)] [KLDt ? DEA(t?3)]

Variable DEA(t?1) DEA(t?2) DEA(t?3)

Intercept 0.9809 0.9143 0.9842

p value \0.0001 \0.0001 \0.0001

KLDt 0.0509 0.0360 0.0119

p value 0.0013*** 0.0366** 0.5037

COMPINTt -0.0004 -0.0004 -0.0005

p value 0.0004*** 0.0013*** 0.0002***

KLDt 9 COMPINTt -0.0001 -0.0001 -0.0001

p value 0.0089*** 0.0967* 0.8015

LAT 0.0044 0.0092 0.0052

p value 0.1450 0.0034*** 0.1131

ROA 0.4019 0.1766 0.2564

p value \0.0001*** \0.0001*** \0.0001***

LEV 0.0550 0.0459 0.0331

p value 0.0003*** 0.004*** 0.0474**

Obs. 902 783 660

Adj. R2 (from OLS) 0.2266 0.1472 0.1925

DEA is run for each year

KLDi,t corporate social responsibility (CSR) score of firm i in year t

COMPINTt market competition intensity (measured by the Herfindahl industry concentration index) in year t. This index is the sum of squared

market shares of the firms in the industry. The lower the Herfindahl industry concentration index, the higher the competition intensity

DEAi,t relative productivity score calculated by the BCC model of Data Envelopment Analysis (DEA) of firm i in year t

LATi,t natural log of total assets (Compustat Item #6) of firm i in year t

ROAi,t return on asset ratio [operating income (Compustat Item #13)/total assets (Compustat Item #6)] of firm i in year t

LEVi,t leverage ratio [total liabilities (Compustat Item #9 ? Compustat Item #34)/total assets (Compustat Item #6)] of firm i in year t

*** p \ 0.01, ** p \ 0.05, * p \ 0.1

Corporate Social Responsibility and Firm Productivity 261

123

Page 12: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

This study has several limitations. First, chemical firms

often particulate in CSR activities gradually, and the stages

of such participation can be difficult to determine. Second,

this study, like other prior studies, may be subject to selection

bias. That is, firms selected and rated by the KLD database

are already good performers. Third, the study focuses on one

industry in the United States. Caution needs be exercised

when generalizing the conclusions of this study. Future

studies can explore the link between CSR and firm produc-

tivity in other industries or the link between CSR and other

competitiveness dimensions in Vilanova et al. (2009).

References

Aupperle, K., Carroll, A., & Hartfield, J. (1985). An empirical

examination of the relationship between corporate social respon-

sibility and profitability. Academy of Management Journal, 28(2),

446–463.

Banker, R. D. (1984). Estimating most productive scale size using

Data Envelopment Analysis. European Journal of Operational

Research, 17, 35–44.

Banker, R. D., Charnes, A., & Cooper, W. W. (1984). Some models

for estimating technical and inefficiencies in Data Envelopment

Analysis. Management Science, 30(9), 1078–1092.

Banker, R. D., Cooper, W. W., Seiford, L. M., Thrall, R. M., & Zhu,

J. (2004). Returns to scale in different DEA models. European

Journal of Operational Research, 154, 345–362.

Barney, J. (1991). Firm resources and sustained competitive advan-

tage. Journal of Management, 17(1), 99–120.

Beurden, P., & Gossling, T. (2008). The worth of values—A literature

review on the relation between corporate social and financial

performance. Journal of Business Ethics, 82, 407–424.

Bowlin, W. F. (1999). An analysis of the financial performance of

defense business segments using Data Envelopment Analysis.

Journal of Accounting and Public Policy, 18(4), 287–310.

Chandler, A. (2005). Shaping the industrial century: The remarkable

story of the evolution of the modern chemical and pharmaceu-

tical industries. Boston, MA: Harvard Business Press.

Charnes, A., Cooper, W. W., & Rhode, E. (1978). Measuring the

efficiency of decision making units. European Journal of

Operational Research, 2, 429–444.

Chen, C., Pattern, D., & Roberts, R. (2008). Corporate charitable

contributions: A corporate social performance or legitimacy

strategy? Journal of Business Ethics, 82(1), 131–144.

Cho, C., Pattern, D., & Roberts, R. (2006). Corporate political

strategy: An examination of the relation between political

expenditures, environmental performance, and environmental

disclosure. Journal of Business Ethics, 67(2), 139–154.

Cochran, R., & Wood, R. (1984). Corporate social responsibility and

financial performance. Academy of Management Journal, 27(1),

42–56.

Cooper, W. W., Seiford, L. M., & Tone, K. (2000). Data envelopment

analysis: A comprehensive text with models, applications.

References and DEA-Solver Software: Kulwer Academic Pub-

lishers, Boston, MA.

Deckop, J. R., Merriman, K. K., & Gupta, S. (2006). The effect of

CEO pay structure on corporate social performance. Journal of

Management, 32(3), 329–342.

Dhaliwal, D., Li, O., Tsang, A., & Yang, Y. (2011). Voluntary nonfinancial

disclosure and the cost of equity capital: The initiation of corporate

social responsibility reporting. The Accounting Review, 86(1), 59–100.

Dopuch, N., Gupta, M., Simunic, D., & Stein, M. (2010). Production

efficiency and the pricing of audit services. Contemporary

Accounting Research, 20(1), 47–77.

Dyson, R. G., Allen, R., Camanho, A. S., Podinovski, V. V., Sarrico,

C. S., & Shale, E. A. (2001). Pitfall and protocols in DEA.

European Journal of Operational Research, 132, 245–259.

European Commission (EC). (2002). Green Book: promoting a

European framework for corporate social responsibility. http://

europa.eu.int/comm/employment_social/soc-dial/csr/greenpaper.

htm.

Fare, R., Grosskopf, S., & Lovell, C. A. K. (1985). The measurement

of efficiency of production. Boston, MA: Kluwer Nijhoff.

Feroz, E., Goel, H., & Raab, R. L. (2008). Performance measure for

accountability in corporate governance: A data envelopment

analysis approach. Review of Accounting and Finance, 7, 121–130.

Feroz, E., Kim, S., & Raab, Rl. (2003). Financial statement analysis:

A data envelopment analysis approach. Journal of Operational

Research Society, 54, 48–58.

Graves, S., & Waddock, S. A. (1994). Institutional owners andcorporate social performance. Academy of Management Journal,

18, 303–317.

Griffin, J. J., & Mahon, J. F. (1997). The corporate social performance

and corporate financial performance debate: Twenty-five years

of incomparable research. Business and Society, 36(1), 5–31.

Hamel, G., & Prahalad, C. K. (1989). Strategic intent. Harvard

Business Review, 3, 63–76.

Hurston, P. (2011). Economic outlook for U.S. Chemical industry

mixed. http://www.americanchemistry.com/Media/PressReleases

Transcripts/RelatedPDF/Economic-Outlook-for-US-Chemistry-

Industry-Mixed-Shale-Gas-Offers-Bright-Spot.pdf.

Johnson, R. A., & Greening, D. W. (1999). The effects of corporate

governance and institutional ownership types on corporate social

performance. Academy of Management Journal, 42(5), 564–576.

Karnani, A. (2010). The case against corporate social responsibility.

MIT Sloan Management Review. http://sloanreview.mit.edu/

executive-adviser/2010-3/5231/the-case-against-corporate-social-

responsibility/.

Kay, J. (1993). Foundations of corporate success. Oxford: Oxford

University Press.

Luo, X., & Bhattacharya, C. B. (2006). Corporate social responsibil-

ity, customer satisfaction, and market value. Journal of Market-

ing, 70, 1–18.

Luo, X., & Bhattacharya, C. B. (2009). The debate over doing good:

Corporate social performance, strategic marketing levers, and

firm-idiosyncratic risk. Journal of Marketing, 73, 198–213.

Luo, X., & Donthu, N. (2006). Marketing’s credibility: A longitudinal

investigation of marketing communication productivity and

shareholder value. Journal of Marketing, 70, 70–91.

McGuire, J., Sundgren, A., & Schneeweis, T. (1988). Corporate social

responsibility and firm financial performance. Academy of

Management Journal, 31(4), 854–872.

Mintzber, H. (1993). The rise and fall of strategic planning. New

York: Free Press.

Moore, G. (2001). Corporate social and financial performance: An

investigation in the U.K. supermarket industry. Journal of

Business Ethics, 34, 299–315.

Nelling, E., & Webb, E. (2009). Corporate social responsibility and

financial performance: The ‘‘virtuous circle’’ revisited. Review of

Quantitative Finance and Accounting, 32, 197–209.

Porter, M., & Kramer, M. (2006). The link between competitive

advantage and corporate social responsibility. Harvard Business

Review, 84(12), 78–92.

Ruf, B. M., Muralidhar, K., Brown, R. M., Janney, J. J., & Paul, K. (2001).

An empirical investigation of the relationship between change in

corporate social performance and financial performance: A stake-

holder theory perspective. Journal of Business Ethics, 32, 148–156.

262 L. Sun, M. Stuebs

123

Page 13: Corporate Social Responsibility and Firm Productivity: Evidence from the Chemical Industry in the United States

Shropshire, C., & Hillman, A. (2007). A longitudinal study of

significant change in stakeholder management. Business and

Society, 46(1), 63–87.

Tsoutsoura, M. (2004). Corporate social responsibility and financial

performance: The ‘‘virtuous circle’’ revisited. Working Paper,

University of California at Berkeley.

Vilanova, M., Lozano, J., & Arenas, D. (2009). Exploring the nature

of the relationship between CSR and competitiveness. Journal of

Business Ethics, 87, 57–69.

Waddock, S. (2003). Myths and realities of social investing.

Organization and Environment, 16, 369–380.

Waddock, S., & Graves, S. (1997). The corporate social perfor-

mance—Financial performance link. Strategic Management

Journal, 18(4), 303–319.

Corporate Social Responsibility and Firm Productivity 263

123


Recommended