Professional Documents
Culture Documents
D’ARCIMOLES Charles-Henri
IAE de Tours, CERMAT
76 Rue Roger Salengro
37000 Tours
Tél. : +(33) 2.47.05.68.93
E-mail : darcimoles@droit.univ-tours.fr
TREBUCQ Stéphane
Université Montesquieu-Bordeaux IV, CRECCI
29 rue de la Cape, Rés. Biarritz F28
33200 Bordeaux
Tél. : +(33) 5.56.02.64.61
E-mail : trebucq@montesquieu.u-bordeaux.fr
SUMMARY :
Recent scholarship in financial and strategic management suggests that there is a positive, neutral,
or negative relationship between corporate social performance and financial performance. Using a
new source of data on corporate social performance, this paper reports first results for France.
Corporate social performance is found to have no relationship with financial performance. This
study also confirms the necessity to control models for investment in research and development.
*
Partial support for this research was provided by the CER2D (Centre de Recherche sur le Développement
Durable) and AReSE (Agence de Rating Social et Environnemental sur les entreprises).
1
INTRODUCTION
Over the last three decades, there has been increasing pressure on firms in the United States to
improve their corporate social performance (CSP). In France, this movement is only three years
old. French firms are now regularly surveyed by new rating agencies in order to score their
empirical researchers are now interested in knowing how stakeholder management can bring
better financial performance. Ethical funds are said to outperform market indices, and French
managers now wonder if they should create value for their shareholders or for all their
stakeholders.
Previous literature has studied the relationship between financial performance and firms' social
responsibility or social performance, but results remain inconclusive. (Roman et al., 1999). We
also propose to test this relationship on the French market for the first time. Using a new French
source of data, the French equivalent of KLD (Kinder, Lydenberg, Domini & Co.) ratings, this
paper reports results based on the model used by Waddock and Graves (1997) and the criticism
2
THE PROBLEM OF MEASUREMENT
CSP, as a multidimensional construct, needs to consider the main stakeholders of the firm and, for
this reason, remains difficult to measure. Firstly, CSP attributes and major stakeholders can
change from one country to another, due to cultural or methodological differences. Secondly, CSP
ratings are rather sensitive not only to the quality of information, but also to the nature and the
The equivalent of the 'Fortune' reputation and social responsibility index has been available in
France since 1996. We decided not to use it because of the same halo effects that have been
observed in the United States (Baucus, 1995). Since 1998, French firms have had similar social
and environmental ratings as KLD's. Ethical fund managers now use this information in France.
After two years of trials, the 1999 ratings appear to be reliable. This is the first opportunity we
have had to reproduce the study of Waddock and Graves (1997) in France.
HYPOTHESES
Our hypotheses address the existence of a relationship between financial performance and CSP,
and if any, its sign and direction. As previous literature has shown, the sign of the relationship may
be negative, neutral or positive. Whatever the sign of the relationship, it is also possible that CSP
3
Negative association
When costs engaged to improve CSP exceed advantages, financial performance can be impaired.
Firms which go far beyond their legal duties can then be handicapped in terms of international
competitivity. Conversely, firms which neglect certain stakeholders such as employees or the
Positive association
Proponents of this line of reasoning predict better financial results for companies which avoid
claims of primary stakeholders and are able to anticipate their satisfaction. This argument makes
sense in the long-term because good relations with employees, suppliers and customers are
essential for the survival of the firm. Actions which support the community can improve the
company's reputation and have a positive impact on sales. An environmental protection research
plan can also help to anticipate future norms and avoid sanctions and taxes.
Neutral association
Some strategy researchers assert that there are too many intervening variables to observe any
direct relationship between financial performance and CSP. Thus, there is no reason to observe
any relationship, except possibly by chance. As McWilliams and Siegel (2000) have proved, the
relationship between financial performance and CSP disappears when more accurate variables are
introduced into econometric models, such as research and development intensity. Others suggest
that the stakeholder orientation can also remain a pure marketing strategy with null effects on
4
financial performance. Corporate social performance is also too complex to be grasped in one or
several measures. This measurement problem contributes to the masking of any observable
We mentioned earlier that there are two conflicting views concerning the direction of causality
between CSP and financial performance. Like Waddock and Graves (1997), we call the positive
impact of financial performance on CSP the 'slack resources theory', and the positive impact of
CSP on financial performance 'good management theory'. Waddock and Graves (1997) have
used a time lag method and concluded that there is a positive interaction between CSP and
financial performance through time. But the model used by McWilliams and Siegel (2000) found
that the good management theory was no longer supported when a research and development
variable was introduced. However, the research note of these authors did not examine the slack
5
We propose to retain the two hypotheses of Waddock and Graves (1997), and add a third
METHOD
Measuring CSP
In order to measure CSP, we constructed an average based on five corporate social performance
attributes rated across the French SBF120 by the firm AReSE (Social and Environmental Rating
Agency)1. AReSE is the French partner of the American KLD (Kinder, Lydenberg, Domini).
AReSE is also a subsidiary of two French financial institutions : the State-owned 'CDC' (Caisse
des Dépôts et Consignations) and the 'Caisse d'Epargne' bank. AreSE's analysts focus exclusively
management.
AReSE's rating scheme is relatively different to that of KLD. Firstly, social issue participation and
exclusionary screens are not used by AReSE. Secondly, AReSE does not use areas of concern
6
and areas of strength, as in KLD's methodology, but a list of non- financial criteria within each
dimension. AReSE rates companies on five attributes of CSP (ER for employee relations, ENV
for environment, SHA for shareholder relations, PRD for product quality and relations with
providers and customers, COM for community), providing a multidimensional assessment. The
AReSE categories are rated on a scale ranging from 1 (backward), 2 (on the way), 3 (going well),
4 (advanced) to 5 (pioneer). These ratings also take into account the specificities of each industry.
As in the case of KLD, AreSE's team of analysts works with data gathered from management
The French social and environmental ratings are quite recent. They were first introduced in 1998,
and we believe it is too early to find a French panel of experts able to weight AReSE rating
categories. Using the simple average of the five ratings to approximate Corporate social
performance (CSP), we have implicitly given an equal weighting to each major category of
Control variables
Previous literature has indicated a need to control not only for industry, risk and size (Ullman,
1985; Waddock and Graves, 1997), but also for research and development intensity
(McWilliams and Siegel, 2000). Size is a relevant variable because social and environmental
expenses and the ability to communicate may differ between small and large firms. A firm's size is
measured by three alternative variables : total sales, total assets and the number of employees.
1
Data source : AReSE, 38 avenue Franklin Roosevelt, 77120 Fontainebleau-Avon, France.
7
Financial efforts towards employees or environmental protection also depend upon the
management's room for manoeuvre. As a proxy for management's risk tolerance, we use the debt
to total capital ratio. In order to test the possible misspecification of the model used by Waddock
and Graves (1997) with American data on the French market, we have used the research and
development expenditures on net sales ratio, following McWilliams and Siegel (2000).
Financial performance
Financial performance (profitability) was measured using three accounting variables : return on
assets, return on equity, cash flow on sales. These ratios provide a range of usual measures used
to assess corporate financial performance. All financial data was derived from the 'Worldscope'
database.
Analysis
Table 1 gives a listing of the average sector CSP ratings for the 12 main sectors. As confirmed by
the Kruskal Wallis test (chi-square = 4.29 and p = 0.96), CSP has been computed by AReSE in
order to give the 12 sectors similar CSP averages. As we have already observed, this is a
different methodology than that of KLD. Table 2 provides descriptive statistics for the variables
used in this study. Table 3 gives the correlation matrice for the five CSP attributes. With a sample
of 99 French firms, regression analysis was used to test hypotheses 1 and 2, similar to the
methodology used by Waddock and Graves (1997). We tested hypothesis 1 using CSP (1999)
8
as the dependent variable and profitability (1998) as the independent variable, while controlling for
industry, size and risk for the year 1998. We then tested hypothesis 2 using profitability (2000) as
the dependent variable and CSP (1999) as the independent variable, and employing the same
control variables for 1999. We finally added research and development intensity as an
independent variable in order to test a possible false relationship observed between CSP and
RESULTS
As can be seen in Table 4(a), CSP for 1999 does not correlate with 1998 financial performance
and other 1998 control variables. These data are used in a regression model to test hypothesis 1,
with CSP as a dependent variable and financial performance as an independent variable. Table
4(b) provides the correlation matrice of variables used to test hypothesis 2, with financial
correlation observed between financial performance in 2000 and CSP in 1999. Furthermore, one
can note the usefulness of control variables. Hence, we can see the negative correlation between
debt and financial performance (return on assets or return on equity), and the negative correlation
9
Table 5 presents the results corresponding to the following econometric model with t = year 1999
where :
CSPt,i = a proxy for corporate social performance (based on the average of the five
AReSE ratings)
PERFt-1,i = financial performance in the previous year (return on assets, return on equity, or
RISK t-1,i = a proxy for the "risk" of the firm in the previous year (debt/total capital)
SIZE t-1,i = a proxy for the size of the firm in the previous year (total sales, total assets or
number of employees)
Industry dummy variables are not statistically significant and are not reproduced in Table 5. The
exclusion of R&D from this econometric model is not problematic at this point, because models 1
to 9 of Table 5 do not show an effect of financial performance on CSP, where Waddock and
Graves (1997) had found a positive relation with 1990 data and a larger sample of 470 American
firms. These results do not support hypothesis 1 for French firms, which posits that higher
performance leads to better CSP. Where Waddock and Graves found no relationship between
CSP and size, our results show that CSP is related to the number of employees. This is only
10
significant (p<0.10) when return on equity or cash flow on sales is used. One can also wonder if
larger firms really have a better CSP or merely have a better ability to communicate than smaller
firms.
The second hypothesis contended that better CSP could lead to higher performance. Like
Waddock and Graves (1997), we used a similar econometric model with t = year 2000 and :
As can be seen in Table 6, results show that return on equity (2000 data) depends significantly on
1999 CSP (models 4-6). Other financial performance measures do not depend on 1999 CSP.
Waddock and Graves (1997) had found a positive relationship, but French data show a significant
negative CSP effect when financial performance is measured by the return on equity. Contrary to
the assumption of hypothesis 2, we could then conclude that a better CSP leads to lower return
on equity. However, according to McWilliams and Siegel (2000), the omission of research and
development makes the model 2 misspecified. We then tested a third model, as follows :
where :
RDINTt-1,i = R&D intensity of firm i in the previous year (R&D expenditures / net sales)
11
Results of model 3 are provided in Table 7. As can be seen, R&D intensity appears to have a
positive and significant effect on all financial performance variables. The negative and significant
effect of CSP on the return on equity also disappears in models 4-6 in the presence of the R&D
intensity variable. The results confirm, as indicated by hypothesis 3, the neutral relationship
between financial performance and CSP. This tends to prove the misspecification of model 2, but
unlike McWilliams and Siegel (2000), any significant correlation between CSP and R&D intensity
This study has attempted to address what is becoming a crucial question in France : whether there
is a relationship between corporate social performance and financial performance, and if so, in
what direction the causation runs. To test our hypotheses on French data, we estimated three
models. Models 1 and 2 had the same specifications as Waddock and Graves (1997) and model
3 was the one proposed by McWilliams and Siegel (2000). Results finally indicate the neutral
effect of CSP on financial performance, and the importance of R&D expenditures in remaining
profitable.
12
Implications for future research
We caution readers, as McWilliams and Siegel (2000) did, to be wary of models that claim to
confirm the effect of corporate social performance without simultaneously controlling for important
Much also remains to be done to improve empirical studies on the relationship between CSP and
financial performance. Other measures can be used. For example, methodologies such as data
envelopment analysis (Epstein and Henderson, 1989; Bendheim et al., 1998) could be useful in
separate CSP attributes could also be used in econometric models (Hillman and Keim, 2001).
Measures such as MVA, EVA and Tobin's Q could also be used to assess firms' financial
performance.
On the one hand, our findings do not suggest that a corporate social performance engagement is
not useful, or is unable to bring competitive advantage in the long run. It would be useful to
determine whether or not our results indicating a neutral relationship will hold over time. On the
other hand, it appears that better CSP does not reduce financial performance. It will be all the
more difficult for French managers not to improve their stakeholder relations as the use of social
and environmental ratings will be generalized. Hence, the French government has recently imposed
13
more transparency on French firms, and companies are now obliged to disclose an annual
environmental and social report. The use of social and environmental ratings will certainly
contribute to reassure French citizens in a context in which it will soon be time to reform the
French retirement system and settle on pension funds. The best stakeholder managed firms could
then benefit from a stronger stock demand. For these reasons, the relationship between CSP and
14
REFERENCES
Baucus MS. 1995. Halo-adjusted Residuals - Prolonging the Life of a Terminally Ill Measure of
Corporate Social Performance. Business & Society 34:227-235.
Bendheim CL, Waddock SA, Graves SB. 1998. Determining Best Practice in Corporate-
Stakeholder Relations Using Data Envelopment Analysis. Business & Society 37:305-338.
Epstein MK, Henderson JC. 1989. Data Envelopment Analysis for Managerial Control and
Diagnosis, Decision Sciences 20:90-119.
Hillman AJ, Keim GD. 2001. Shareholder value, stakeholder management, and social issues :
What's the bottom line ? Strategic Management Journal 22:125-139.
McWilliams A, Siegel D. 2000. Corporate social responsibility and financial performance :
correlation or misspecification ? Strategic Management Journal 21:603-609.
Roman RM, Hayibor S, Agle BR. 1999. The relationship between social and financial
performance - repainting a portrait. Business & Society 38:109-125.
Ullman AH. 1985. Data in search of a theory : A critical examination of the relationships among
social performance, social disclosure, and economic performance of US firms. Academy of
Management Review 10:540-557.
Waddock SA, Graves SB. 1997. The corporate social performance - financial performance link.
Strategic Management Journal 18:303-319.
15
Table 1. Sectors in the sample
Categories Code Sector N CSP Min. Max.
INDUSTRIALS (01-07) 01 Energy 3 3.00 2.80 3.20
02 Intermediate goods 6 3.23 2.80 4.00
03 Construction 9 3.02 2.40 3.60
04 Capital goods 12 3.08 2.20 3.60
05 Automobile 8 2.95 2.20 3.60
06 Other consumer goods 9 3.07 2.20 3.80
07 Food 5 3.16 2.60 4.00
SERVICES (08-09) 08 Distribution 8 2.80 2.00 3.60
09 Other services 23 2.98 2.40 3.60
FINANCIALS (10-12) 10 Property 5 3.08 2.60 3.60
11 Financial services 10 3.02 2.60 3.60
12 Investment companies 1 2.60 2.60 2.60
16
Table 4(a). Correlation matrices : Correlations with 1998 financial data and 1999 CSP
CSP ROA ROE CFS D/C Sales Assets No. empl
CSP 1.000 - 0.031 - 0.046 0.143 0.011 0.084 - 0.052 0.168
ROA 1.000 *** 0.739 0.162 ** - 0.220 *** - 0.274 *** - 0.293 * - 0.199
ROE 1.000 0.027 - 0.105 - 0.071 - 0.096 - 0.098
CFS 1.000 0.152 - 0.017 0.080 - 0.162
D/C 1.000 0.161 *** 0.264 0.148
Sales 1.000 *** 0.689 *** 0.650
Assets 1.000 ** 0.205
* p ≤ 0.10 ; ** p ≤ 0.05 ; *** p ≤ 0.01
Table 4(b). Correlations with 2000 profitability. 1999 CSP. and 1999 financial controls
CSP ROA ROE CFS D/C Sales Assets No. empl
CSP 1.000 - 0.029 - 0.156 0.080 0.016 0.075 -0.016 0.159
ROA 1.000 *** 0.650 *** 0.293 *** - 0.408 ** - 0.209 ** - 0.243 -0.135
ROE 1.000 * 0.178 ** - 0.204 0.114 0.088 0.055
CFS 1.000 0.077 - 0.061 0.045 -0.150
D/C 1.000 0.166 ** 0.219 ** 0.220
Sales 1.000 *** 0.633 *** 0.658
Assets 1.000 * 0.189
* p ≤ 0.10 ; ** p ≤ 0.05 ; *** p ≤ 0.01
17
Table 5. Regression analysis using 1999 CSP as the dependent variable and 1998 financial data as independent
variables
Control variables
Debt / Total capital -1.82 E-4 4.09 E-4 -2.45 E-4
Total sales 2.07 E-6
Total assets -5.47 E-7
Number of employees 1.40 E-6
4 5 6
Independent variable : ROE -5.21 E-3 -5.76 E-3 -4.72 E-3
Control variables
Debt / Total capital -2.42 E-4 6.03 E-4 -4.38 E-4
Total sales 2.12 E-6
Total assets -5.01 E-7
Number of employees 1.39 E-6 *
7 8 9
Independent variable : CFS 4.72 E-3 4.122 E-3 5.77 E-3
Control variables
Debt / Total capital -3.82 E-4 4.033 E-4 -6.07 E-4
Total sales 2.87 E-6
Total assets -4.10 E-7
Number of employees 1.59 E-6 *
18
Table 6. Regression analysis with 2000 financial performance (profitability) as dependent variable and 1999
CSP as the key independent variable with 1999 control variables
Control variables
Debt / Total capital - 0.087 *** - 0.084 *** - 0.091 ***
Total sales - 4.73 E-5
Total assets - 7.48 E-6
Number of employees - 1.75 E-6
Control variables
Debt / Total capital - 0.178 ** - 0.179 ** - 0.181 **
Total sales 2.161 E-4 *
Total assets 2.258 E-5
Number of employees 3.63 E-5
Control variables
Debt / Total capital 0.050 0.038 0.065
Total sales - 6.49 E-5
Total assets 3.70 E-6
Number of employees - 3.08 E-5 *
19
Table 7. Same regression analysis as Table 6, introducing the R&D independent variable
Control variables
Debt / Total capital - 0.149 ** - 0.152 * - 0.147 **
Total sales 5.70 E-5
Total assets 4.45 E-5
Number of employees 4.369 E-6
Control variables
Debt / Total capital - 0.451 ** - 0.455 ** - 0.463 **
Total sales 3.99 E-4
Total assets 2.49 E-4
Number of employees 7.34 E-5
Control variables
Debt / Total capital - 0.192 ** - 0.197 ** -0.196 **
Total sales 1.63 E-4
Total assets 1.16 E-4
Number of employees 2.87 E-5
20