Escolar Documentos
Profissional Documentos
Cultura Documentos
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at .
http://www.jstor.org/action/showPublisher?publisherCode=jwiley. .
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
John Wiley & Sons is collaborating with JSTOR to digitize, preserve and extend access to Strategic
Management Journal.
http://www.jstor.org
StrategicManagementJournal
Strat. Mgmt. J., 21: 603-609 (2000)
AND
CORPORATESOCIALRESPONSIBILITY
FINANCIALPERFORMANCE:
CORRELATION OR
MISSPECIFICATION?
ABAGAILMcWILLIAMSl*and DONALDSIEGEL2
1School of Management, Arizona State University West, Phoenix, Arizona, U.S.A.
2The University of Nottingham Business School, Nottingham, U.K.
Researchers have reporteda positive, negative, and neutral impact of corporate social responsi-
bility (CSR) on financial performance. This inconsistency may be due to flawed empirical
analysis. In this paper, we demonstratea particularflaw in existing econometric studies of the
relationship between social and financial performance. These studies estimate the effect of CSR
by regressingfirm performance on corporate social performance, and several control variables.
This model is misspecified because it does not control for investmentin R&D, which has been
shown to be an important determinant of firm performance. This misspecification results in
upwardlybiased estimates of the financial impact of CSR. When the model is properly specified,
we find that CSR has a neutral impact on financial performance. Copyright ? 2000 John
Wiley & Sons, Ltd.
EMPIRICAL STUDIES OF CSR AND PERFi = f(CSPi, SIZEi, RISKi, INDi) (1)
FINANCIAL PERFORMANCE
where
There are basicallytwo types of empiricalstudies
of the relationshipbetweenCSR and financialper- PERFi= long-run economic or financial per-
formance.One set of studiesuses the event study formance of firm i (measures of
methodology to assess the short-run financial accountingprofits)
impact (abnormalreturns)when firms engage in CSPi = a proxy for corporatesocial responsi-
socially responsibleor irresponsibleacts (see, for bility of firm i (based on an index of
example, Clinebell and Clinebell, 1994; Hannon social performance)
andMilkovich,1996;Posnikoff,1997;Teoh,Welch SIZEi proxy for the size of firm i
=a
andWazzan,1999;Worrell,Davidson,andSharma, RISKi=a proxy for the "risk" of firm i
1991; Wright and Ferris, 1997). The results of (debt/assetratio)
thesestudieshavebeen mixed.Forexample,Wright INDi = industryof firm i (4 digit SIC code)
and Ferrisfound a negativerelationship; Posnikoff
reporteda positive relationship;and Teoh et al. The inclusion of the industrydummy (IND) is
found no relationshipbetween CSR and financial to control for some industry-levelfactors that
performance,when examining divestituresfrom have been shown to explain variation in firm
SouthAfricaduringthe Apartheidcontroversy(see performanceacross industries,such as economies
McWilliams,Siegel and Teoh, 1999, for a dis- of scale and competitiveintensity.2We hypothe-
cussionof these studies).Otherstudiesare similarly size thatEquation1 is misspecifieddue to omitted
inconsistenton the relationshipbetweenCSR and variables,becauseit does not controlfor a firm's
shortrun financialreturns(McWilliamsand Siegel, rate of investmentin R&D and the advertising
1997, providesa theoreticaland empiricalcritique intensityof its industry.A more appropriatespec-
of the use of the event study methodologyfor ificationis:
examiningthe impactof CSR).
A second set of studies examines the nature PERFi= f
of the relationship between some measure of
(CSPi, SIZEi,RISKi,INDi, RDINTi, INDADINTi)
corporatesocial performance,CSP (a measureof
CSR), and measures of long term firm perfor- (2)
mance, using accountingor financialmeasuresof
profitability(see, for example,Aupperle,Carroll, where the additionalcovariatesare:
and Hatfield, 1985; McGuire, Sundgren and
Schneeweis, 1988; and Waddock and Graves, RDINTi = R&D intensity of firm i
1997). The results from these studies have also (R&D expenditures/sales)
been mixed. Aupperleet al. foundno relationship INDADINTi= advertising intensity of the
between CSP and profitability,McGuire et al. industryof firm i
found that prior performancewas more closely
relatedto CSP than was subsequentperformance, Excluding R&D in the econometricmodel is
and Waddockand Graves found significantposi- especially problematic,because there is a long
tive relationshipsbetween an index of CSP and standingtheoreticalliteraturelinking investment
performancemeasures such as ROA in the fol- in R&D to improvementsin long-runeconomic
lowing year. performance(Griliches, 1979). In these models,
The inconsistencyof the resultsfromthese stud- R&D is consideredto be a form of investmentin
ies of the relationshipbetween CSR and perfor- "technical"capital.Investmentin technicalcapital
mance is not surprising,given the natureof the resultsin knowledgeenhancement,which leads to
models that form the basis for the empiricalesti- productand process innovation.This innovative
mation.For example,Waddockand Graves(1997) activityenablesfirmsto enhancetheirproductivity.
estimatethe followingeconometricmodel:'
2We will argue that a very specific type of industry effect-
'Note that many studies simply examine correlation coef- industry advertising intensity-must also be (separately) con-
ficients, but with causal implications. trolled for, because it is so closely associated with CSR.
Copyright ? 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 603-609 (2000)
Research Notes and Communications 605
ationX" shoppers.We need only look at the rapid (Powell, 1996; Rumelt,1991; Schmalansee,1975;
growthof such socially responsiblecompaniesas Waring, 1996), the consensus is that industry
Ben & Jerry's,the Body Shop, and HealthValley factors "matter,"in the sense that they explain
to confirmthe importanceof CSR in marketing. a non-negligiblepercentageof the variationin
Supportof CSR may also be used to create a profitability across firms. Thus, INDADINTi
reputationthat a firm is reliable and honest, and should be included in the model, along with
some consumers may assume that the products "size" and "risk,"as a controlvariable.
of a reliable and honest firm will be of high If our conjectures are true (corr (RDINT,
quality.Therefore,advertisingthatprovidesinfor- PERF)> 0, corr (RDINT, CSP) > 0), then the
mation about CSR attributesmay be used to consequencesof omittingR&D from Equation1
create a reputationfor quality or reliability or are clear. As noted in Theil (1971), if an omitted
honesty-all attributes that are important,but regressor,in this case RDINT,is positivelycorre-
may be difficultfor consumersto determine.Such lated with both the dependentvariable (PERF)
advertisingmakes consumers aware of product and the includedregressor(CSP), then the coef-
differentiation(quality) based on CSR attributes. ficient on CSP, in the misspecifiedEquation 1,
For example, New United Motor Manufactur- will be overestimated.
ing, Inc., or NUMMI,the innovativejoint venture Simply put, the positive and significantcoef-
between Toyota and GeneralMotors, was estab- ficient on CSP, as reportedby Waddock and
lished in Fremont, Californiain 1984 to build Graves (1997), could simply reflect the impact
small cars for both companies.The NUMMIplant of R&D on firm performance.It is impossibleto
implementedmany of the latest Japanese"lean isolate the impact of CSP on firm performance
manufacturing"methods (process innovation), unless the model is properlyspecified.A similar
and producedthe Geo Prism, the prototypefor argument could be made for other omitted
GM's new generation of small cars (product regressors,such as advertisingintensity, if they
innovation). Furthermore, through its unique are also positively correlatedwith CSP and firm
partnership with the United Auto Workers performance.
(UAW), NUMMI also implemented a number
of progressive workplace practices, such as a
strong emphasis on teamwork and employee EMPIRICAL ANALYSIS
empowerment. The bottom line is that some
consumersperceived that NUMMI cars, such as To assess the validity of the results reportedin
the Geo Prism, were superior to traditional, studies that employ Equation 1 (Waddock and
American-madecars, in terms of quality and Graves, 1997), we estimate the model outlined
reliability. More germanely, many customers in Equation 2. For this estimation, we linked
also believed that by purchasingthese cars, they Compustatdatato informationon corporatesocial
were demonstratingtheir supportof progressive performanceprovidedto us by the firmof Kinder,
human resource managementpractices and the Lydenberg, and Domini (KLD), which began
UAW. compiling this informationin May 1991. KLD
providesratingsof corporatesocial performance,
The link between advertising and firm or CSP (a measure of corporate social
performance responsibility),for portfolio managersand other
institutionalinvestors who wish to incorporate
The remainingindependentvariablein our pro- social factors into their investment decisions.
posed model-Equation 2- (INDADINTi) is Many of these social investorswant to "screen"
designed to serve as a proxy for the extent of their portfoliosto exclude companiesthat violate
product differentiationat the industrylevel and their social principles. In this context, CSP is
entry barriersthat might serve to enhance firm definedas a (0,1) variable;a firm is either soci-
profitability.Entry barriers are a shared asset ally responsibleor it is not, based on the "screen"
across firmsin an industry,becauseentrybarriers applied.For example, an investmentfirm that is
are an industrylevel construct(McWilliamsand managing a portfolio for evangelical Christians
Smart, 1993). While there is considerabledebate will avoid companiesin the gamblingand alco-
regardingthe magnitudesof industrylevel effects hol industries.
Copyright? 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 603-609 (2000)
Research Notes and Communications 607
KLD uses a combinationof surveys, financial that estimationof Equation1 constitutesa speci-
statements,articles on companiesin the popular ficationerrorthat may resultin an overestimation
press, academicjournals(especiallylaw journals), of the impact of CSP on financialperformance.
and government reports to assess CSP along This overestimationarises because CSP is posi-
eleven dimensions:military contracting,nuclear tively correlated with R&D, which has been
power, gambling, tobacco, alcohol, community foundto be a strongdeterminantof improvements
relations, diversity, employee relations, environ- in economic performance.
ment, and productquality(innovation/R&D),and We argue that firm-levelinvestmentin R&D,
non-U.S. operations (usually environment and and additionalindustryfactors (advertisinginten-
labor relations).5Based on this information,the sity as a proxy for barriersto entry) should also
firm constructedthe Domini 400 Social Index be included in the econometricspecification.To
(DSI 400), the functionalequivalentof the Stan- explicitly test our hypothesis that Equation 1 is
dard and Poors 500 Index for socially respon- misspecified,we examine variantsof Equation2,
sible firms. including the rate of firm level investment in
In orderto be eligible for the DSI 400, a firm R&D and industrydummyvariables(4 digit SIC)
must derive less than 2% of its gross revenue in the model (with advertisingintensity included
from the productionof military weapons, have as a control variable). These findings are
no involvementin nuclearpower, gambling,to- presentedin Table2.
bacco, and alcohol, and have a positive recordin The results confirm our hypothesis regarding
each of the remainingsix categories.For example, the importanceof including R&D and industry
a firm that implementsrecycling and pollution- factors in a model that attempts to "explain"
preventionprograms,provides donationsto con- corporateperformance.As shown in column (1),
servationorganizations,and demonstratesconcern when R&D and industry factors are excluded
for the environmentin its day-to-dayoperations, from the model, the coefficienton CSP is positive
is regardedas having a positive recordalong the and statisticallysignificant.However,when R&D
environmentaldimension. A firm that actively and industryfactors are added to the model, the
promotesminoritiesand women to top managerial magnitudeof the coefficient diminishesdramati-
positions and membershipon the boardof direc- cally and is no longer significant.Additionally,
tors will receive a similar positive score along the "fit"of the model improves,as shown by the
the diversity dimension.Our measureof CSP is increase in the adjustedR2. Thus, our findings
a dummyvariable,with a value of 1 if a firm is underscorethe importanceof using the appropri-
included in the DSI 400 in a given year (for ate specificationwhen estimatingthe "return"on
having passed the "social screen");0 otherwise. CSR investment.6
Our data series, createdfrom a linkage of the
KLD data and Compustat,contains524 firms.To
simplify the econometricanalysis and to ensure DISCUSSION
comparabilitywith existing studies, each of the
variablesin Equation2 is computedas an average Over the last 3 decades, the pressureon firms to
annual value for the years 1991-1996, a time engage in corporatesocial responsibility(CSR)
period that correspondsto the overlap of the has increased.Many managershave respondedto
Compustatand KLD files. Table 1 presentsdefi- these pressures,but many have resisted. Those
nitions, descriptive statistics, and a correlation who resist typically have invoked the trade-off
matrix for the three key variables:PERF, CSP, between socially responsiblebehavior and prof-
and RDINT. itability.Managementresearchershave responded
Several stylized facts are evident from Table 1. to this by attemptingto demonstratethe effect of
The most strikingresultsare thatR&D, CSP, and CSR on profitability.However, the results of
financial performanceall appear to be strongly
positivelycorrelated.This supportsour hypothesis 6A caveat is in order. Our result of no financial impact from
CSR may be a result of the lack of a good measure of CSR.
We use the KLD rating system, which relies heavily on
SAdditional detail on the KLD file and the social "screens" negative screens and includes philanthropic activities. A more
is presented in Waddock and Graves (1997) and Kinder and business-oriented definition of CSR might yield a different
Domini (1997). result. We thank a reviewer for pointing this out.
Copyright ? 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 603-609 (2000)
608 A. McWilliams and D. Siegel
Table 1. Definitions of key variables, descriptive statistics, and correlations (N = 524 firms)
Table 2. Regression results from estimation of variants ation that includes CSP (a measure of CSR) as
of Equation 2 (N = 524 firms, standard errors in a determinant of firm performance, but not R&D
parentheses) will result in upwardly biased estimates of the
CSP variable.
Dependent (1) (2) (3)
Variable: PERF To test our hypothesis, we estimated two mod-
els. The first was the same specification as Wad-
Coefficient on 0.141*** 0.104 -0.062 dock and Graves and the second was one in
CSP (0.052) (0.106) (0.059) which we included R&D intensity. Our results
confirm that CSP and R&D are highly correlated,
Coefficient on - 0.145*** 0.263***
RDINT (0.050) and that, when R&D intensity is included in the
(0.036)
equation, CSP is shown to have a neutral effect
Industry Dummies No No Yes on profitability. This should not be surprising,
(4 digit SIC) because many firms that actively engage in CSR
included are also pursuing a differentiation strategy,
AdjustedR2 0.10 0.19 0.29
involving complementary strategic investments in
*p - 0.10; **p 0.05; ***p 0.01 R&D. This makes it difficult to isolate the impact
Note: All regressions include controls for size, risk, and of CSR on performance without simultaneously
advertisingintensity,which are computedas annualaverages controlling for R&D. Therefore, we caution read-
over the period 1991-1996.
ers to be wary of models that claim to "explain"
firm performance, but do not include important
strategic variables, such as R&D intensity.
empirical studies of the relationship between CSR
and profitability have been inconclusive, reporting
positive, negative, and neutral results. REFERENCES
We hypothesized that this inconsistency could
be due to flaws in empirical analysis. One parti-
A. Carroll and J. Hatfield (1985). 'An
cular flaw is econometric estimation of a mis- Aupperle, K.,examination of the
empirical relationship between
specified model. An example of such a speci- corporate social responsibility and profitability',
fication error is the equation estimated by Academy of Management Journal, 28(2), pp. 446-
Waddock and Graves, 1997, which is misspecified 463.
because it does not include a measure of firm- Austin, D. H. (1993). 'An event study approach to
level investment in R&D. This is unfortunate, measuring innovative output: The case of biotechnol-
ogy', American Economic Review, 83(2), pp. 253-
because there is a large body of empirical evi- 259.
dence showing that investment in R&D has a Ben-Zion, U. (1984). 'The R&D and investment
strong positive impact on profitability. We also decision and its relationship to the firm's market
value: Some preliminary results'. In Z. Griliches
hypothesized that R&D investment and CSR are
(ed.), R&D, Patents, and Productivity. University
likely to be highly correlated, because both are of Chicago Press, Chicago, IL, pp. 134-162.
associated with product and process innovation. Chan,S. H., J. D. Martinand J. W. Kensinger(1990).
If CSR and R&D are highly correlated, an equ- 'Corporateresearchand developmentexpenditures
Copyright ? 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 603-609 (2000)
Research Notes and Communications 609
and share value', Journal of Financial Economics, McGuire, J., A. Sundgren and T. Schneeweis (1988).
26, pp. 255-276. 'Corporate social responsibility and firm financial
Chung, K. H. and P. Wright (1998). 'Corporate policy performance', Academy of Management Journal,
and market value: A q-theory approach', Review of 31(4), pp. 854-872.
Quantitative Finance and Accounting, 11, pp. 293- McWilliams, A. and D. Siegel (1997). 'Event studies
310. in management research: Theoretical and empirical
Chung, K. H., P. Wright and C. Charoenwong (1998). issues', Academy of Management Journal, 40(3),
'Investment opportunities and market reaction to pp. 626-657.
capital expenditure decisions', Journal of Banking McWilliams, A., D. Siegel and S. H. Teoh (1999).
and Finance, 22, pp. 41-60. 'Issues in the use of the event study methodology:
Clark, K. B. and Z. Griliches (1984). 'Productivity A critical analysis of corporate social responsibility
growth and R&D at the business level: Results studies', Organizational Research Methods, 2(4),
from the PIMS database'. In Z. Griliches (ed.), pp. 350-372.
R&D, Patents, and Productivity. University of McWilliams, A. and D. Smart (1993). 'Efficiency v.
Chicago Press, Chicago, IL, pp. 393-416. structure-conduct-performance: Implications for
Clinebell, S. K. and J. M. Clinebell (1994). 'The effect strategy research and practice', Journal of Manage-
of advanced notice of plant closings on firm value', ment, 19(1), pp. 63-78.
Journal of Management, 20, pp. 553-564. Posnikoff, J. F. (1997). 'Disinvestment from South
Griliches, Z. (1979). 'Issues in assessing the contri- Africa: They did well by doing good', Contemporary
bution of R&D to productivity growth', Bell Journal Economic Policy, 15(1), pp. 76-86.
of Economics, 10(1), pp. 92-116. Powell, T. C. (1996). 'How much does industry matter?
Griliches, Z. (ed.) (1998). R&D and Productivity: The An alternative empirical test', Strategic Management
Econometric Evidence. National Bureau of Eco- Journal, 17(4), pp. 323-334.
nomic Research for the University of Chicago Press, Rumelt, R. (1991). 'How much does industry matter?',
University of Chicago Press, Chicago, IL. Strategic Management Journal, 12(3), pp. 167-185.
Guerard, J. B., Jr., A. S. Bean and S. Andrews (1987). Schmalansee, R. (1985). 'Do markets differ much?',
'R&D management and corporate financial policy', American Economic Review, 75, pp. 341-351.
Management Science, 33, pp. 1419-1427. Sundaram,A., T. John and K. John (1996). 'An empiri-
Guerard,J. B., Jr., B. K. Stone and S. Andrews (1988). cal analysis of strategic competition and firm values:
'CorporateR&D expenditures, innovation, and com- The case of R&D competition', Journal of Financial
petition in an international economy'. In R. Shrieves Economics, 40(3), pp. 459-470.
(ed.), Competition in the International Economy. Teoh, S. H., I. Welch and C. P. Wazzan (1999). 'The
Cambridge University Press, Cambridge, UK, effect of socially activist investment policies on the
pp. 215-250. financial markets: Evidence from the South African
Hall, B. H. (1999). 'Innovation and market value', boycott', Journal of Business, 72(1), pp. 35-89.
National Bureau of Economic Research Working Theil, H. (1971). Principles of Econometrics. John
Paper #6984. Wiley & Sons, New York.
Hannon, J. and G. Milkovich (1996). 'The effect of Waddock, S. and S. Graves (1997). 'The corporate
human resource reputation signals on share prices: social performance - financial performance link',
An event study', Human Resource Management, Strategic Management Journal, 18(4), pp. 303-319.
35(3), pp. 405-424. Waring, G. (1996). 'Industry differences in the persis-
Kinder, P. and A. Domini (1997). 'Social screening: tence of firm-specific returns', American Economic
Paradigms old and new', Journal of Investing, 6(4), Review, 86(5), pp. 1253-1265.
pp. 12-19. Worrell, D., W. N. Davidson and V. N. Sharma (1991).
Kinder, P., S. Lydenberg and A. Domini (1993). Mak- 'Layoff announcements and stockholder wealth',
ing Money While Being Socially Responsible. Harper Academy of Management Journal, 34(3), pp. 662-
Business, New York. 678.
Lichtenberg, F. and D. Siegel (1991). 'The impact of Wright, P. and S. Ferris (1997). 'Agency conflict and
R&D investment on productivity: New evidence corporate strategy: The effect of divestment on
using linked R&D-LRD data', Economic Inquiry, corporate value', Strategic Management Journal,
29, pp. 203-228. 18(1), pp. 77-83.
Copyright ? 2000 John Wiley & Sons, Ltd. Strat. Mgmt. J., 21: 603-609 (2000)