Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
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.
Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy of Management
Journal.
http://www.jstor.org
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
Social
Responsibility
Financial
Performance1
Corporate
and
PHILIP L. COCHRAN
ROBERT A. WOOD
Pennsylvania State University
Therelationshipbetweencorporatesocialresponsibility
and financial performance is reexaminedusing a new
methodology,improvedtechnique,and industry-specific
controlgroups.Averageage of corporateassetsisfound
to be highlycorrelatedwithsocial responsibilityranking.
After controllingfor thisfactor, therestill is some correlationbetweencorporatesocialresponsibilityandfinancial performance.
The relationshipbetweena firm'scorporatesocialresponsibility(CSR)or, morerecently,its corporatesocialresponsiveness-andits financialperformancehas beenthe subjectof a livelydebatesincethe 1960s.Researchers
havereachedno realconsensuson the relationshipbetweenthesevariables.
In fact, a recentworkreviewedsevenearlierempiricalstudiesandconcluded
that "economic performanceis not directlylinked, in eithera positive or
negativefashion,to socialresponsiveness"
(Arlow& Gannon,1982,p. 240).
Whetheror not a relationshipexistsclearlyis an importantissue for corporate management.If certainactions (classifiedas socially responsible)
tend to be negativelycorrelatedwith financialperformanceof firms, then
managersmightbe advisedto be cautiousin this area.If, on the otherhand,
a positive relationshipcan be shown to exist, then managementmight be
encouragedto pursuesuch activitieswith increasedvigor or to investigate
the underlyingcauses of this relationship.The focus of this paper is on
the questionof whetherthesetwo factors(CSRand financialperformance)
are related.Only once this questionis answeredcan the questionof causation be addressed.
This paperextendspriorempiricalresearchin threeareas. Certainfactors omittedin previousstudiesare explicitlycontrolledfor by use of improvedfinancialperformancemeasuresandadditionalvariables.A statistical
tool, logit analysis, more suited to the data is employed. Finally, the
'The authors would like to express their appreciation to Anthony Curley, Keith Ord, Robert Pitts,
Steven Wartick, and three anonymous referees for their helpful comments and suggestions.
42
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
1984
Cochranand Wood
43
44
Academyof ManagementJournal
March
this index was used by Moskowitzand a composite of his 1972-1975indexeswas used by Sturdivantand Ginter(1977)in theirstudiesof the relationship between CSR and financial performance.
Anotherpopularreputationindexalso can be tracedbackto Moskowitz.
A surveywas conductedby the National Associationof ConcernedBusiness Students("How businessschool studentsrate corporations," 1972)
in which300 graduatestudentsof businessadministrationwerequestioned
about their views on the social responsibilityof some of the Fortune 500
firms. Indexesgeneratedby this study were used subsequentlyby Vance
(1975), Heinze (1976), and Alexanderand Buchholz (1978).
The second method of measuringCSR is content analysis. Normally,
in contentanalysisthe extentof the reportingof CSR activitiesin various
firmpublicationsand especiallyin the annualreportis measured.This can
consist of simply noting whetheror not a particularitem (such as pollution control)is discussedeitherqualitativelyor numerically,or it can mean
actuallycountinga numberof items. A commonlyused sourcefor content
analysisis a series of studies conductedby Beresford(1973, 1975, 1976).
Contentanalysishas two significantadvantages.First, once the particular variableshave been chosen (a subjectiveprocess),the procedureis reasonablyobjective.Thereforethe resultsare independentof the particular
research.Second, becausethis techniqueis more mechanical,largersample sizes are possible.
However,contentanalysisalso has some drawbacks.The choiceof variables to measureis subjective.Further,content analysisis only an indication of what firmssay they are doing, and this may be verydifferentfrom
what they actuallyare doing. At best, one certainlycould postulatethat
firmsthat are awareof these issuesare those that will discussthem as well
as act on them. On the other hand, one could imaginethat firms that are
doingpoorlyon this frontwouldfeel an extraincentiveto makethemselves
look good by touting their achievementsin their annual reports.
The first study to use content analysiswas Bowmanand Haire (1975).
In this study the authorsexaminedonly the food processingindustryand
developedtheirown indexbasedon the numberof linesof the annualreport
devoted to CSR. Subsequently,severalother studies (Abbott & Monsen,
1979;Anderson& Frankle,1980;Ingram,1978;Preston, 1978)used content analysisindexes based on Beresford'swork.
Neithercontentanalysisnor reputationindexescan be consideredwholly
adequatemeasuresof CSR. The problemof measuringsocial responsibility or responsivenessof firmsneedsconsiderablymoreattentionin this literature. Yet, at the moment there obviously are not better measures
available.
Measurement of Financial Performance
1984
45
performanceeither. In fact, thereis a wide rangeof such measures.However,mostmeasuresof financialperformancefall into two broadcategories:
investorreturnsand accountingreturns.Both haveenjoyedperiodsof popularity, and both have evolved considerablyover the course of the past
decade.
InvestorReturns.Thebasicideaunderlyinginvestorreturnsis thatreturns
should be measuredfrom the perspectiveof the shareholders.The first
studiesto employ investorreturnsas a measureof financialperformance
werethose of Moskowitz(1972)and Vance(1975). In both of thesestudies
changesin priceper sharewas used as the investorreturnsindex. As most
subsequentstudieshave noted, this measureis clearlyflawed. The change
in priceper shareis only one elementof investorreturns.Dividendincome
is the other, and it must be includedin any measureof investorreturns.
Abbott and Monsen(1979)used the changein sharepriceplus dividends
as their measureof investor returns.However, this, too, is insufficient.
Simple returns(changein price per share plus dividends)fail to capture
another dimensionof vital importanceto investors-namely, risk.
In acceptedfinancetheory, the risk of holdingassetsis measuredby the
covarianceof the expectedreturnon the assetwiththatof the overallmarket.
This measure,which is commonlyreferredto as "beta," typicallyis obtained for a stock by regressingits realizedreturnson those of a broad
basedmarketindex. The regressionslope coefficientprovidesthe beta estimate. An averagebeta is 1. A stock with a beta above 1 is consideredan
aggressivestock because it will tend to move faster, either up or down,
thanthe market.Correspondingly,
a stockwitha betabelow 1 is considered
a defensive stock (Curley& Bear, 1979).
It was preciselythis failureto adjustfor riskthat led to the (apparently)
resultsof Moskowitz(1972)andof Vance(1975).Moskowitz's
contradictory
studyindicatedthat firmswithhighCSRratingsoutperformedthe market.
Vance, two years later, concludedjust the opposite. An examinationof
Moskowitz'shigh CSR firms over the 1970-1979periodindicatesthat the
portfolio of these firms had a beta of 1.56. The period that Moskowitz
examined,the first half of 1972, was a bull market(i.e., rising), and the
periodthat Vancelooked at, 1972through 1974, was a bear market(i.e.,
falling). Therefore,the apparentcontradictionbetweentheir results can
be explainedby the riskinessof the returnsof the firmsthat they had classifiedas sociallyresponsibleand not by the variabletheybelievedthey were
examining.
Two studies did use risk adjustedmeasuresof investorreturns.These
were Alexanderand Buchholz(1978) and Andersonand Frankle(1980).
However, there is a problemwith the use of even a "clean" measureof
investorreturnsfor this type of study. This problemis summarizedby one
of the tenets of modernfinancetheory, the efficient marketshypothesis.
Simplystated,this tenetpositsthat as informationthat mightaffect future
cash flows of a firm becomesavailable,it immediatelywill be reflectedin
its currentshareprice.The implicationof this is that evenif CSRdoes lead
to improvedfinancialperformance,as soon as the marketbecomesaware
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
46
Academyof ManagementJournal
March
groupingcriterionbeingmaximizationof the differenceof a functionalrelationship betweeneach of the two subgroupsat each iteration. The functional relationshipthat Ingramuses is excess marketreturnfor each firm
as the dependentvariable;he uses fiscal year, excessaccountingearnings,
and industryas explanatoryvariables.
Ingram'sproceduredivideshis sampleof 116 firms into 10 subgroups
whereineach subgrouphas two sets of firms-one having higher excess
marketreturnsthan the other. In seven of these subgroupings,firms in
the higherexcessmarketreturncategoryhave betterCSR ratingsthan do
those in the lower excess marketreturncategory. The reverseis true for
the remainingthree subgroups.
Although this techniqueavoids many of the problemsencounteredby
earlierstudies,one mustbe cautiousin interpretingIngram'sresultsas support for a correlationbetweenCSR and financialperformance.If, in fact,
Ingram'snullhypothesis,thatis, thatthereis no relationshipbetweensocial
responsibilitydisclosuresand financialperformance,were true, then 7 or
more of the 10 groupswould have disclosureratingsin the higherfinancial performancecategorieswith a frequencyof 11.32 percent.This clearly is not a strong rejectionof the null hypothesis.
AccountingReturns.Accountingreturnsare the other primarymethod
of measuringfinancialperformance.The basicidea behindusingaccounting returnsas a measureof financialperformanceis to focus on how firm
earningsrespondto differentmanagerialpolicies.The most commonmeasuresof accountingreturnsusedin studiesof this questionare simplyearnings per share(EPS) or price/earnings(P/E) ratios. Bragdonand Marlin
(1972),Bowmanand Haire(1975),Folgerand Nutt (1975),Heinze(1976),
Preston(1978),andSpicer(1978)all usedEPS, P/E ratios,or somealgebraic
variationof them as at least one of the measuresof financialperformance
in their studies.
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
1984
47
There are several problems, however, associated with using EPS or P/E
ratios as such a measure. Both are strongly influenced by the rate of growth
and accounting practicesof firms (Beaver& Morse, 1978). In addition, these
financial performance measures cannot be accuratelycompared across firms
without considering financial leverage influences and risk differences. This
does not mean that one cannot use accounting returns, quite the oppositeaccounting returns may be the best proxy for financial performance. However, the particular measures used in previous studies have serious defects.
Samples in Previous Studies
Most of the previous work in this area employed samples that were too
small to result in any safely generalizableresults. The Folger and Nutt (1975)
study, for example, examined only nine firms. In fact, 6 of the 14 studies
examined here had samples of less than 30 firms.
Second, the control groups in a number of these studies were too small.
In several of the studies the small sample of CSR firms was segmented and
subgroups were compared. In others the sample was matched with an external control group of similar size. All but four of the studies employed
control groups smaller than 100 firms.
In order to overcome this problem, several researchers(Abbott & Monsen,
1979; Alexander & Buchholz, 1978; Moskowitz, 1972; Preston, 1978; Vance,
1975) compared their samples to broad market averages such as the Standard and Poors 500. This step represents an improvement, but comparison to industry control groups is superior. Accounting practices, operating leverage, and other variables that may influence test results will be more
homogeneous within industries.
Sturdivant and Ginter (1977) recognized this problem and grouped their
sample into four somewhat homogeneous subgroups. Unfortunately, by
doing so they reduced their overall sample size and did not, necessarily,
create subgroups with sufficient homogeneity.
Finally, the time period(s) employed in a number of the previous studies
was too short. Seven of these studies employed time periods equal to or
less than two years. Only five of the studies used time periods greater than
or equal to five years.
Results of Previous Studies
As noted earlier, the previous empirical work in this area reached no real
consensus on the nature of the relationship between CSR and financial performance. Of the 14 studies examined, fully 9 (Anderson & Frankle, 1980;
Bowman & Haire, 1975; Bragdon & Marlin, 1972; Heinze, 1976; Ingram,
1978; Moskowitz, 1972; Preston, 1978; Spicer, 1978; Sturdivant & Ginter,
1977) found some positive relationship between CSR and financial performance. However, as noted earlier, because of incorrect specification of the
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
48
Academyof ManagementJournal
March
financialperformanceproxy, small sample size, and questionablemethodology, these resultsmust be viewed with suspicion.
Three studies (Abbott & Monsen, 1979;Alexander& Buchholz, 1978;
Folger& Nutt, 1975)found no reallinkbetweenthesevariables.Alexander
and Buchholz,as wellas AbbottandMonsen,usedformsof investorreturns
as proxiesfor financialperformance.Abbott and Monsen,however,failed
to accountproperlyfor risk.Alexanderand Buchholzdid properlyaccount
for risk, but they did not employan eventstudyand thus could have failed
to discovera relationshipeven if one did exist.
Onestudy(Vance,1975)founda negativerelationshipbetweenthesevariables. However, as discussedearlier,his proxy for investorreturns(only
change in share price) clearlywas insufficient.
Method
Measuresof Social Responsibility
49
1984
Table 1
Firms in Sample
CSR
#a
2
3
3
3
2
2
3
2
3
2
1
3
2
3
2
3
2
I
I
3
2
3
2
1
2
3
I
2
2
1
2
3
I
I
3
3
1
I
1
Firm
SIC
#
Industry
differencesacrossfirmsas well as differentapplicationsof accountingprinciples. These distortionswere controlled for insofar as possible.
Threeaccountingreturnsmeasureswereemployedinitially:(1) the ratio
of operatingearningsto assets, (2) the ratio of operatingearningsto sales,
and (3) excessmarketvaluation.Eachhas certainstrengthsand weaknesses.
The ratio of operatingearnings(as measuredby operatingearningsbefore depreciation,Compustatdata item 13, less depreciationand amortization, Compustatdata item 14) to assets measuresthe relativeefficiency
of asset utilization. A major strengthof this ratio is that it is free from
the effects of bias that can resultfrom differencesin capital structurebetween firms. However,it can be distortedby the effect of inflation on the
book valueof the assets.For example,two firmsmayhaveidenticalphysical
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
Academyof ManagementJournal
50
March
1984
51
Dependent Variable/
Interval
Best
t Statistics for
CSR Dummy Variables__
Honorable
Worst
Mention
Coeficient of
Multiple Determination
(Adjusted for d.f.)
.91
1.34*
.09
-1.34
.47
.59
.57
.58
-.21
-.89
.29
.15
1.92*
1.05
-.04
1.13
.50
.38
1.07
1.22
.24
.59
-2.07**
-2.24**
.29
.34
-1.24
-1.78*
-.68
-1.04
.08
-.02
.76
.69
1
"*p<.
**p< .05
***p<.01
52
March
Table 3
FinancialVariableCorrelation
1970-1974, 1975-1979a
Operating
Earnings/
Assets
Operating earn1.000 (1.000)
ings/assets
Operating earnings/sales
Asset age
Asset turnover
Excess value
Operating
Earnings/
Sales
.587
Asset
Age
(.823)
.055
(.136)
1.000 (1.000)
.241
1.000
(.102)
(1.000)
Asset
Turnover
.115
Excess
Value
(.249)
.570
(.561)
-.469 (-.226)
-.036 (-.233)
1.000 (1.000)
.567
.207
-.178
1.000
(.656)
(.062)
(.005)
(1.000)
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
1984
53
multicollinearity
problembetweenEV and operatingearnings/sales.Therefore the test was repeatedwith the inclusion of each and both of these
variables.
Logit analysisprovidesa chi-squaretest for the significanceof each explanatoryvariablein predictingthe CSR categoryin whichan observation
(firm) will fall. The logit model estimatedis:
CSR = a0+ a1OES + a2AGE + a3TURN + a4EV+ E,
Pairwise
CSR Comparisons
Coefficient
Estimate
Chi-Square
Statistic
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
-.408
.237
.645
-6.219
-2.121
-4.098
19.616
15.346
4.271
-.957
-.533
-.425
1.611
.974
.637
.53
.25
1.61
.08
.01
.08
5.05*
4.08*
.38
.36
.15
.10
1.80
.70
.92
1975-1979
Effect
Intercept
Operating earnings/sales
Asset age
Asset turnover
Excess value
Pairwise
CSR Comparisons
Coefficient
Estimate
Chi-Square
Statistic
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
Best/worst
Honorable mention/worst
Best/honorable mention
.416
.994
.578
38.745
48.046
-9.301
45.035
41.604
3.431
-2.242
-3.768
-1.374
-1.606
-.868
2.162
.29
2.00
1.02
2.93
4.46*
.33
5.63*
5.02*
.22
.94
2.40
.71
.50
.16
1.32
*p< .05
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
54
Academyof ManagementJournal
March
Conclusionsand Implications
Thispaperextendsthe studyof financialperformanceandcorporatesocial
controlgroups,
responsibility
by usingan enhancedsample,industry-specific
and statisticaltools new to this area of research.The major conclusionis
that withinindustrygroupsthe financialvariablemost stronglycorrelated
with CSRis assetage and that omissionof this variableresultsin a spurious
correlationof CSRand financialperformance.Specifically,firmswitholder
assets have lower CSR ratings.
The source of the correlationbetweenasset age and CSR is of interest.
One possibleexplanationis that firmswith older assetsconstructedplants
in a periodwhenregulatoryconstraintswereless severethanthey aretoday.
For example,if a firm built a plant priorto the mid-sixtiesthereis an excellentchancethat its facilitypollutesmorethanone builtin the recentpast.
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
1984
Cochranand Wood
55
Future ResearchDirections
Futureresearchin this area could proceed in a numberof directions.
First,bettermeasuresof CSR are desperatelyneeded.It may neverbe possibleto measureCSRobjectively.Thereforeresearchin this areacouldfocus
onperceptionsof CSR.Regularsurveysof businesspeople,businesswriters,
businessschool faculty, and the publicat largecould give researchersreliable reputationindexesfor time seriesstudies. In addition, the Beresford
studiescould be replicatedfor recentyears. Such content analysisstudies
in some sensesmay reflectthe firm'sperceptionsof its socialresponsibility.
Second, more extensivemeasuresof CSR also are needed. It would be
very useful to have CSR rankingson at least the Fortune 500 firms over
severalyears.Theremightbe separaterankingsfor eachof severaldifferent
constituencies.This could give researcherssome indicationsof the effects
of certainpolicy changes on perceptionsof CSR and subsequenteffects
on financial performance.
Third, the possibilityand implicationsof additionalexplanatoryvariablesmustbe explored.The resultsof this study(i.e., that assetage is highly
correlatedwith CSR)throwdoubt on the resultsof most previousstudies.
The implicationsof this new factor should be exploredand additionalexplanatoryfactors sought.
Finally,causalityshouldbe investigated.No workto datehas statistically
demonstratedthe directionof causationbetweenthese two variables.One
promisingtechniquewouldbe an eventstudy.However,in orderto conduct
a valideventstudybetterand moreextensiveCSRrankingswillbe required.
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions
56
March
References
Abbott, W. F., & Monsen, R. J. On the measurement of corporate social responsibility: Self-report
disclosure as a method of measuring social involvement. Academy of Management Journal, 1979,
22, 501-515.
Alexander, G. J., & Buchholz, R. A. Corporate social responsibility and stock market performance.
Academy of Management Journal, 1978, 21, 479-486.
Anderson, J. C., & Frankle, A. W. Voluntary social reporting: An iso-beta portfolio analysis. The
Accounting Review, 1980, 55(3), 467-479.
Arlow, P., & Gannon, M. J. Social responsiveness, corporate structure and economic performance.
Academy of Management Review, 1982, 7, 235-241.
Beaver, W., & Morse, D. What determines price-earnings ratios? Financial Analysts Journal, 1978,
34(4), 65-86.
Beresford, D. R. Compilation of social measurement disclosures in Fortune 500 annual reports-1973.
Ernst and Ernst, 1973.
Beresford, D. R. Social responsibility disclosure in 1974 Fortune 500 annual reports. Ernst and Ernst,
1975.
Beresford, D. R. Social responsibility disclosure-1975 survey of Fortune 500 annual reports. Ernst
and Ernst, 1976.
Bowman, E. H., & Haire, M. A strategic posture toward corporate social responsibility. California
Management Review, 1975, 18(2), 49-58.
Bragdon, J. H., & Marlin, J. A. T. Is pollution profitable? Risk Management, 1972, 19(4), 9-18.
Council of Economic Priorities. Economic priorities report, 1971, 1(6), 13-23.
Curley, A. J., & Bear, R. M. Investment analysis and management. New York: Harper, 1979.
Errunza, V. R., & Senbet, L. W. The effects of international operations on the market value of the
firm: Theory and evidence. Journal of Finance, 1981, 36(2), 401-417.
Folger, H. R., & Nutt, F. A note on social responsibility and stock valuation. Academy of Management Journal, 1975, 18, 155-159.
Heinze, D. C. Financial correlates of a social involvement measure. Akron Business and Economic
Review, 1976, 7(1), 48-51.
How business school students rate corporations. Business and Society Review, 1972, 2, 20-21.
Ingram, R. W. An investigation of the information content of (certain) social responsibility disclosures.
Journal of Accounting Research, 1978, 16, 270-285.
Moskowitz, M. Choosing socially responsible stocks. Business and Society Review, 1972, 1, 71-75.
Moskowitz, M. Profiles in corporate responsibility. Business and Society Review, 1975, 13, 29-42.
Preston, L. E. Analyzing corporate social performance: Methods and results. Journal of Contemporary
Business, 1978, 7(1), 135-149.
Spicer, B. H. Investors, corporate social performance and information disclosure: An empirical study.
The Accounting Review, 1978, 53(1), 94-111.
Sturdivant, F. D., & Ginter, J. L. Corporate social responsiveness: Management attitudes and economic
performance. California Management Review, 1977, 19(3), 30-39.
Thomadakis, S. A value-based test of profitability and the market structure. The Review of Economics
and Statistics, 1977, 59(2), 179-185.
Vance, S. Are socially responsible corporations good investment risks? Management Review, 1975,
64(8), 18-24.
Philip L. Cochran is Assistant Professor of Business Administration at Pennsylvania State University.
Robert A. Wood is Assistant Professor of Finance at Pennsylvania State University.
This content downloaded from 121.52.153.182 on Mon, 14 Sep 2015 16:56:46 UTC
All use subject to JSTOR Terms and Conditions