Corps de l’article

The debate on the commitment to Corporate Social Responsibility (CSR, hereafter) has attracted the interest of several researchers around the world. The literature has mainly focused on the drivers of the CSR (Galbreath, 2010). Specifically, the social behavior of family firms is less studied in literature. Debicki et al. (2009) argue that it is important to pay attention to the aspects of social responsibility in family firms. Indeed, family firms have specific characteristics. They have mostly concentrated ownership structure, undiversified portfolios, long-term horizons, and are involved within management (Morck et al. 1988). Their motivations for CSR practices may then differ from other types of ownership structures (Craig and Dibrell, 2006). The existing literature finds conflicting results regarding the effect of family control on CSR performance (Dyer and Whetten, 2006; Berrone et al. 2010; Block and Wagner, 2014; Ducassy and Montandrau, 2015; El Ghoul et al. 2016; Cui et al. 2018; Labelle et al. 2018). According to this literature, there are two competing theoretical frameworks for the family control-CSR relationship. The first one is the socio-emotional perspective suggesting that family members are inclined to preserve their socio-emotional wealth (SEW, hereafter) apart from economic considerations (Berrone et al. 2012). Along with this perspective, families are committed to CSR activities to protect their emotional goals (Dyer and Whetten, 2006; Berrone et al. 2010; Déniz and Suárez, 2005). The second perspective is the agency theory which highlights the opportunistic behaviour of families. Indeed, in case of excess control, families are less likely to engage in CSR activities and will divert company resources for their own benefits for expropriation purposes (El Ghoul et al. 2016).

Most research on the relationship between family firms and CSR performance focuses on a single perspective without dealing with the degree of family involvement within the company. However, family members can have significant control over the firm and perpetuate their influence over the company in several ways, such as holding voting rights beyond their cash flow rights, the involvement of family members in the management etc. Bingham et al. (2011) show that the extent of family involvement leads to a different emphasis on CSR initiatives. Our study falls within this research trend as it investigates the extent to which family control affects CSR performance. Specifically, our study aims to examine the relationship between family control and CSR performance not only through family involvement in capital and management but also through family identification with the company and excess control using both socio-emotional wealth and agency theory perspectives. Our study is particularly interesting in the French context where companies are mostly held by families (Bouzgarrou and Navatte, 2014).

We also examine the effect of board characteristics on CSR performance in family firms. The extant literature emphasizes that the board of directors is a corporate governance device likely to reduce agency conflicts (Anderson and Reeb, 2003). We choose to focus on board features given the relative weakness of the institutional and legal environment in France[1]. Thus, the role of the board of directors is more valuable and is able to influence decision-making regarding CSR activities.

The contribution of this paper is then threefold. First, this study contributes to the literature on the CSR performance of family firms by distinguishing between the agency and socio-emotional perspectives in the context of family-owned companies. Previous studies generally focus on a single theory to examine the relationship between family firms and CSR performance (Bartkus et al. 2002; Barnea and Rubin, 2010; Berrone et al. 2010; Block and Wagner, 2014; Dyer and Whetten, 2006; El Ghoul et al. 2016). We draw on Ducassy and Montandrau (2015) who find that French family firms have a neutral effect on social performance. We consider family firms as a heterogeneous group as their behaviour can vary according to family involvement in the company (Sharma et al. 2012; Marques et al. 2014; Labelle et al. 2018).

Second, we provide new evidence on the moderating role of board features on the relationship between CSR performance and family control. We highlight the influence of gender diversity and small boards to mitigate opportunistic family behavior and protect minority interests in French firms. Third, most of the studies explore the relationship between family control and CSR performance through conditional mean regression estimations. This study is based on a novel technique to test the level of CSR engagement of family firms i.e. the quantile regression approach. This technique assumes that the effect of family control on CSR performance could vary among different levels of the CSR distribution. Indeed, family behavior toward CSR could be heterogeneous considering high and low levels of CSR.

Based on a sample of 97 French companies between 2005 and 2016, we find that the effect of family control on CSR performance is not homogeneous. Family identity and ownership enhance CSR performance for firms with low-CSR levels. As for family involvement in management, we find that CEO family members are positively associated with all CSR levels. More specifically, we find that the strongest effect of CEO family member is recorded for low-CSR firms. However, when the family possesses control rights beyond their cash flow rights, the relationship between family control and CSR performance turns negative, suggesting that controlling families may have expropriation purposes and are likely to privilege their personal interests. These findings suggest that the effect of family control on CSR performance varies according to the family’s involvement in the company and to the level of CSR engagement. Regarding board features, the negative effect of excess family control on CSR performance is less prevalent in presence of board gender diversity and for small board firms.

The remainder of the paper is organized as follows. Section 2 presents the theoretical framework. Section 3 presents the existing literature and develops our hypotheses. Section 4 describes the sample and presents the research design. This is followed by results and discussion in Section 5. The last section concludes the paper.

Theoretical Framework

Agency Theory Versus Socio-Emotional Wealth Perspective

Most studies on family businesses are based on the agency theory framework (Schulze et al. 2003; Anderson and Reeb, 2003; Depoers et al. 2020). According to this framework, families are opportunistic and privilege their private benefits to satisfy their economic goals. For instance, Anderson and Reeb (2003) suggest that families engage in opportunistic activities at the expense of minority shareholders’ interests. Burkart et al. (2003) argue that family firms may expropriate private benefits of control through excessive salaries and perquisites, transactions with related parties, and special dividends.

In line with these arguments, families will pay less attention to CSR activities. Indeed, CSR activities are considered by families as an additional cost and a source of wealth dissipation. According to Labelle et al. (2018), families “feel the brunt” of overinvestment in CSR activities. In this sense, El Ghoul et al. (2016) find a negative impact of family control on CSR performance in the East Asian context and thus support for the expropriation view. Similarly, in an international sample, Labelle et al. (2018) show that family firms have low levels of CSR performance compared to non-family firms. The authors also find that in weakly protected investors’ rights settings, family firms seek financial performance at the expense of social performance.

The preceding discussion shows that the agency theory perspective is linked to opportunistic behavior of family firms and to expropriation purposes. Family members will be therefore less likely to engage in CSR activities.

However, the agency theory remains insufficient to tackle the specificities of family businesses. Gómez-Mejía et al. (2007) develop a model to better deal with family firms’ specificities. They point out that this model makes it possible to distinguish family businesses from their non-family counterparts. Their model, called socio-emotional wealth (SEW, hereafter), is an extension of the behavioral theory and claims that “families are emotionally linked to their businesses”.

As a result, family firms do not only seek economic gains from their participation in the business but are motivated by non-financial, emotion-related goals, such as reputation, longevity, and firm succession to further generations (Berrone et al. 2010; Gómez-Mejía et al. 2011; Gómez-Mejía et al. 2007). In family firms, “the preservation of SEW is anchored at a deep psychological level among family owners whose identity is inextricably tied to the organization” (Berrone et al. 2010, p. 87). Berrone et al. (2012) suggest a set of five dimensions of SEW grouped under the acronym FIBER: Family control and influence, Identification of family members with the firm, Binding social ties, Emotional attachment of family members, and Renewal of family bonds to the firm through dynastic succession. Numerous empirical studies have used the SEW approach, considered as potentially a dominant paradigm in the field of family business research (Cennamo et al. 2012; Block and Wagner, 2014; Kalm and Gómez-Mejía, 2016; Poletti-Hughes and Williams, 2019).

Most of the research in SEW approach suggests that family firms are more likely to engage in CSR activities than their non-family counterparts to protect the emotional benefits they gain from controlling the company. From this perspective, family members engage in CSR activities to enhance family identity, image, and reputation (Berrone et al. 2010; Déniz and Suárez, 2005; Gallo, 2004). Along with this perspective, Cruz et al. (2014) show that family firms are more likely to adopt social practices and to be socially responsible to protect their reputation and image, and hence, increase their SEW. Dyer and Whetten (2006) suggest that family firms invest in CSR activities to build corporate image, improve employee loyalty, influence customer perceptions, support long-term community growth, encourage innovation, and invest in the future. The authors show that family firms have higher levels of CSR performance than their non-family counterparts. Barnea and Rubin (2010) also argue that family firms are more motivated to protect their reputation than non-family firms. They engage in CSR activities to create a family brand identity among different stakeholders. Studies by Børsting and Thomsen (2017) show that family firms enjoy a good reputation and are socially responsible companies. According to Cennamo et al. (2012), family members worry about the reputation of the company and are likely to preserve their own SEW.

These emotional perspective-based arguments predict a positive association between CSR performance and family control as family firms have to be proactive with various stakeholders in order to preserve their socio-emotional wealth.

Hypothesis Development

Family firms are considered as a heterogeneous group because the behavior of family members is likely to vary according to the level of family involvement (Berrone et al. 2012; Marques et al. 2014; Labelle et al. 2018). In this study, we focus on four major dimensions of family involvement: ownership, management, identity, and excess control. Consistent with previous theoretical developments, we draw on the perspectives of the SWE and agency theory to construct our set of hypotheses. Indeed, there is a trade-off between incurring economic costs of CSR activities and preserving the family’s socio-emotional wealth (Cennamo et al. 2009).

Following Berrone et al. (2012), we first consider that the SWE perspective can best capture the uniqueness of family businesses. The objective of preserving the SWE could thus have an impact on the relationship between family firms and CSR performance. Our first three hypotheses are along these lines: we believe that family involvement in capital and management and the self-identification of family members with the firm increase CSR performance. Our fourth hypothesis focuses on a particular configuration of family firms that is, excess control, in which family members behave opportunistically in the pursuit of financial goals. Following this agency perspective, we suggest that family firms with excess control (when voting rights exceed financial rights), under-invest in CSR activities and then exhibit weaker CSR performance. Finally, we suggest the existence of a moderating effect of corporate governance on the relationship between excess control and CSR performance.

Family Involvement in Capital and Management

A key feature of family business is the control and influence family members have over strategic decisions (Anderson and Reeb, 2003). The power to control can be exerted by holding large capital shares in the hands of family members and their presence in the top management positions. Several studies show that the implication of family members in the ownership and management enhances the control dimension of SEW (Berrone et al. 2010). In other words, the degree of family members’ involvement in the capital and management may be considered as a proxy for the intensity of their SWE.

In order to enhance their socio-emotional wealth, family firms develop proactive relationships with their stakeholders (Cennamo et al. 2012). In this sense, several studies show that the concentration of ownership and management in hands of families is associated with more CSR engagement. Berrone et al. (2010) suggest that family owners are aware of their firm’s reputation through irresponsible actions. Huang et al. (2009), Gómez-Mejía et al. (2011), and Sharma and Sharma (2011) show that family ownership is likely to have a positive impact on environmental performance. Cui et al. (2018) provide evidence that family-owned firms with a family member as CEO are likely to engage in CSR activities to preserve the socio-emotional wealth.

We then assume that the involvement in ownership and management leads the family to engage in CSR activities to preserve their emotional wealth.

The preceding discussion leads to the following hypotheses:

H1. Family ownership has a positive effect on CSR performance.

H2. A CEO family member has a positive effect on CSR performance.

Family Identity

Previous literature points out that joining the family name with the business creates a unique identity in family firms (Berrone et al. 2010). Indeed, when the company has the same name as the family (eponymous company), the reputation of the company becomes a very sensitive issue to all stakeholders. For instance, the family name can appear on a company’s products or services. In this case, the family is encouraged to engage in social activities. Berrone et al. (2010) suggest that family businesses exhibit high levels of corporate social responsibility because of their strong identification with the firm’s name. Dyer and Whetten (2006) and Barnea and Rubin (2010) argue that family firms are more inclined to protect their reputation than non-family firms since the company is often named as the family. As a result, the latter seeks to engage in CSR activities to create a family brand identity among different stakeholders. This leads to the following hypothesis:

H3. Family identification with the company has a positive effect on CSR performance.

Family Excess Control

The goal of improving SEW motivates the family to get support from stakeholders (Berrone et al. 2012). This idea is at odds with agency concerns that CSR activities are a source of wealth dissipation. Family firms have generally mostly concentrated ownership structure and undiversified portfolios (Morck et al. 1988). As a result, family shareholders will “feel the brunt” of overinvestment in CSR activities and will bear a large portion of the firm’s social responsibility burden (Labelle et al. 2018). The tension between SEW motives and the cost of CSR activities suggests a trade-off likely to affect the behavior of family firms towards CSR activities. Particularly, when families have control rights over their cash flow rights, the tendency to engage in opportunistic behaviors prevails the need to preserve socio-emotional wealth. Family-owned companies hold excess voting rights compared to their cash flow rights through the pyramidal structure or double-voting rights. This is likely to increase their power within the company (La Porta et al. 1999). In addition, excess control gives the family the ability to control and influence decisions without bearing the consequences of their decisions (Depoers et al. 2020). This may encourage such firms to reap private benefits and to seek their own interests (Faccio and Lang, 2002).

Therefore, family firms with an excess of control place greater emphasis on activities that benefit themselves even though they have strong SEW (Claessens et al. 2002). In this sense, Kellermanns et al. (2012) suggest that the SEW perspective has a “dark side”, as high levels of SEW can be associated with low levels of proactive stakeholder engagement. In fact, families with excess control are less pressured by their stakeholders and thus, tend to fall back on their opportunistic behavior by engaging less in CSR activities. Such families can feel secure and therefore seek to decrease the resources to satisfy stakeholders’ needs (Morck and Yeung, 2004). Therefore, when families hold excessive control, the achievement of financial goals prevails over SEW preservation motives, leading to a decrease in CSR performance.

The preceding arguments suggest that excess family control exacerbates the opportunistic behavior of the family leading to less engagement in social and environmental issues. Our fourth hypothesis is then as follows:

H4. Excess family control has a negative effect on CSR performance.

The Moderating Effect of Board of Directors’ Features

As the risk of expropriation by family firms is exacerbated through their excess control, we now investigate the moderating effect of the board of directors on the relationship between excess family control and CSR performance. The monitoring role of the board of directors depends on some board features, e.g., size, independence, CEO duality, and gender diversity, which could influence CSR engagement by family-controlled firms. We focus on these four board attributes for two reasons. On the one hand, these attributes are the more tested in previous empirical studies, allowing comparisons to be made. On the other hand, most of these attributes are regulated in the French context, which demonstrates their importance.

Board Size

Board size is an important device able to influence the effectiveness of board monitoring (Jensen, 1993). The empirical literature is not unanimous regarding the effect of board size on board decisions. Some literature is in favour of large boards because they have diversified experience and qualifications likely to enhance the quality of board monitoring (De Villiers et al. 2011). However, other researchers argue that large boards can be less effective in controlling managerial discretion and in making decisions. Goodstein et al. (1994) document that a large board is more likely to cause difficulty in reaching consensus in the decision-making process. According to Hermalin and Weisbach (2003), a smaller board is easier to coordinate and can then be more effective in monitoring management. Studies on the effects of board size on CSR performance shows that boards with a small number of directors are likely to better coordinate and make decisions in the interests of all stakeholders by investing in CSR (Kassinis and Vafeas, 2002). Moreover, Prado-Lorenzo and García-Sanchez (2010) and Bai (2013) find a smaller board is associated with higher CSR performance. Given the effectiveness of smaller boards toward CSR performance, we formulate our hypothesis as follows:

H5. Board size amplifies the negative relationship between excess family control and CSR performance.

Board Independence

Board independence is likely to improve the board of directors’ efficiency. Numerous studies point out that the effectiveness of the board when it comes to CSR depends on the independence of its members (Jo and Harjoto, 2011). Hence, companies with a high proportion of independent directors are more concerned about CSR engagement. Others point out that external directors seem less committed to economic performance and more concerned about reputation, sustainability, and improvement of CSR (Ibrahim and Angelidis, 1995 and Post et al. 2011). Indeed, the reputation of independent directors is strongly linked to the reputation of companies. They are, therefore, interested in enhancing the company’s responsible behavior since their professional reputation is at stake (García-Sanchez et al. 2011).

In addition, independent directors are hired to protect stakeholders’ interests (Pfeffer, 1973; Fama and Jensen, 1983; Johnson and Greening, 1999; Ajina et al. 2019). Recently, Fuente et al. (2017) argued that independent directors bring their expertise, connections, external know-how, and contacts not only to satisfy the interests of stakeholders but also to ensure the long-term survival of the company.

We then assume that the negative relationship between excessive family control and CSR performance is less prevalent in firms with high board independence.

H6. Board independence weakens the negative relationship between excessive family control and CSR performance.

CEO Duality

Duality refers to one person standing as both the CEO and the chairman of the board. This increases the power of the CEO and consequently the likelihood of opportunistic managerial behavior (Surroca and Tribo, 2008). In relation to CSR engagement, Webb (2004) shows that CEO duality is less prevalent in socially responsible firms than non-socially responsible ones.

Based on the agency theory, CEO duality is likely to empower the CEO and to reduce the efficiency of the board’s monitoring role. This will result in neglecting stakeholders’ interests and in lessening the engagement in CSR activities as the CEO will prefer to advance his or her own interests (Michelon and Parbonetti, 2012). Along with this perspective, Shahzad et al. (2016) prove a negative association between CEO duality and corporate social performance. This is due to agency costs associated with the increasing power of the CEO over the board. In family firms, family members become more entrenched and are inclined to invest in CSR activities. We then assume that in the case of CEO duality, the negative relationship between CSR performance and excessive family control is exacerbated.

H7. CEO duality amplifies the negative relationship between excessive family control and CSR performance.

Gender Diversity

Gender diversity refers to the appointment of women in the boardroom. Board gender diversity is one important feature of board effectiveness (Liao et al. 2015). Adams and Ferreira (2009) show a positive association between the presence of women and the quality of board oversight. Nekhili et al. (2017) point out that the presence of women in the boardroom is a substitute for board independence in French companies.

Numerous studies highlight the positive qualities of women. According to Wood and Eagly (2009), women on the board have a set of psychological characteristics that could enable them to satisfy the interests of stakeholders. Francoeur et al. (2008) emphasize that diversity within the board facilitates the resolution of complex issues. In addition, women directors are more aware of the social engagement of the company and thus more likely to preserve the interests of stakeholders in the meetings (Nielsen and Huse, 2010).

Borghesi et al. (2014) and Galbreath (2016) show that gender diversity enhances the board’s ability to effectively manage CSR activities. Hillman et al. (2002) and Bear et al. (2010) show that women are more likely to meet the needs of the community, while Williams (2003) shows that female directors are more charitable than men. Hence, women may encourage family members to make better social decisions. Moreover, Peake et al. (2017) find that women play an important role in family firms’ participation in community social responsibility. We then assume that the negative relationship between CSR performance and excessive family control is less prevalent in firms with high board gender diversity.

H8. Board gender weakens the negative relationship between excessive family control and CSR performance.

Research Design

Sample and Data

Our sample includes all firms listed on the CAC all tradable index. The CSR data were extracted from Thomson Reuters DataStream using the Assets4 model that provides environmental, social, and governance information. Board data and ownership structure were hand-collected from annual reports of listed firms. These were located on the AMF website. Finally, financial and accounting data were retrieved from the Compustat database. After matching Compustat and Datastream databases, we are left with a sample of 97 French-listed companies covering the period 2005-2016, i.e. 1164 firm-year observations.

The industry distribution of the selected companies is presented in Table 1. According to the classification of Campbell (1996), this table shows that the consumer durables industry is the most represented industry in our sample with 55.68% of the companies. It is also the most socially responsible sector with an average CSR score of 82.605, followed by the basic industry sector with 78.810. The least responsible companies are those belonging to the capitals goods industry with an average of 56.860.

Variables’ Measurements

Corporate Social Responsibility

Our main variable is the CSR performance measured by a score extracted from Assets4 model. This score includes three pillars: social, environmental, and governance. Each pillar includes subcategories. The environmental pillar includes emission reduction, resource reduction, and product innovation. In the social score, there are indicators of workforce, human rights, community, and product responsibility. As for the governance pillar, it includes board structure, board function, compensation, shareholder rights, and vision and strategy. Each data on these pillars were converted into a score from 0 to 100. In this study, we choose to remove the governance score pillar because it is likely to interfere with governance variables (board characteristics) included in our models. The CSR performance score is then the mean of the environmental and social scores. We normalized the CSR score by a logarithmic transformation to have a mean equal to 0 and a standard deviation of 1.

Table 1

Distribution of the sample by sector

Distribution of the sample by sector

-> Voir la liste des tableaux

Family control

To measure “family control”, we construct the four following measures:

Family ownership (FAM-OWN) is the percentage of shares held by the family (Chen et al. 2008 and Burkart et al. 2003). This variable captures the family influence and control dimension of the socio-emotional wealth model. We expect a positive link between this variable and CSR performance.

CEO family member (FAM-MAN) is a binary variable that equals to 1 if the CEO is a member of the controlling family, and 0 otherwise (Chen et al. 2008). This variable also captures the family influence and control dimension. We expect a positive effect of CEO family member on CSR performance.

Family identification (FAM-IDENT) is measured as a dummy variable coded as 1 if the name of the firm is part of the family name, and 0 otherwise. This variable captures family identification with the company i.e. the second dimension of socio-emotional wealth model. We expect a positive link between this variable and CSR performance.

Excess family control (EXCESS) captures the expropriation risk of family members. We expect this variable to be negatively related to CSR performance. Following Masulis et al. (2011), we measure excess control by the ratio of voting rights on cash flows rights held by family members.

Control Variables

We include company-level variables to control various factors that may affect CSR performance.

Firm size (SIZE) captures the existence of developments and economy of scales. As they grow, large firms face more pressures from their stakeholders, constraining them to respond to their demands. Large companies are likely, therefore, to engage in socially responsible activities than their smaller counterparts (El Ghoul et al. 2016 and Labelle et al. 2018). We expect a positive link between this variable and CSR performance. Firm size is measured by the log of total assets.

Debt ratio (LEV) measures the firms’ attitude toward risk. Barnea and Rubin (2010) report a negative link between debt ratio and CSR performance suggesting that firms with higher financial risk may pay less attention to CSR activities. We expect a negative effect of the debt ratio on CSR performance. Debt ratio is measured by the ratio of total debt to total assets.

Growth opportunity(MTB) is the ratio between equity market value and equity book value (El Ghoul et al. 2016 and Cui et al. 2018). This ratio is a measure of a firm’s growth opportunities. Firms with high growth opportunities are more likely to engage in CSR activities to attract more investors and, therefore, reduce the cost of capital (Dhaliwal et al. 2011). We expect a positive relationship between growth opportunities and CSR performance.

Return on assets (ROA) is firm performance. Dyer and Whetten (2006) show that firm’s financial performance is highly correlated with its socially responsible decisions. Return on assets is measured by the ratio of net income before extraordinary items to total assets. It is expected that firm performance is positively associated with CSR performance.

In addition to these firm-level variables, we also consider firm, industry, and year effects in all our regressions. Appendix 1 reports all variables used in this study.

Model Specification

We first use the conditional mean regression in panel data.

Where:

FAMILY-CONTROLit: is either FAM-OWNit= percentage of shares held by family members; FAM-MANit= a binary variable that equals to 1 if the CEO is a member of the controlling family, and 0 otherwise; FAM-IDENTit= a dummy variable that equals to 1 if the name of the firm is part of the family name, and 0 otherwise; EXCESSit= the ratio of voting rights on cash flow rights held by family members.

BOARDit: these are board features, namely: BSIZEit= the total number of directors; INDEPit= the proportion of independent directors on the board relative to the total number of directors; DUALITYit= a dummy variable that takes the value of 1 if the CEO is also the chairman of the board, and 0 otherwise; DIVERSITYit= the proportion of women on the board of directors.

CONTROLS: are control variables including FSIZEit= the log of total assets in fiscal year t; LEVit= the ratio of total debt to total assets; MTBit= the market-to-book ratio; ROAit= the return on assets.

Secondly, we estimate our models using the quantile regression approach. This approach helps overcoming the limitations within the linear regression (OLS estimations). Indeed, simple linear regressions can be inefficient if the errors are highly non-normal or if the sample is slightly asymmetric where quantile regressions are robust to non-normal errors and outliers (Gallego-Álvarez and Ortas, 2016). Furthermore, this approach displays a complete picture of the effect of independent variables on the dependent one (Conyon and He, 2017). The classic OLS model examines the average relationship because it assumes that the conditional distribution of the CSR variable is homogeneous. However, the slopes of the CSR score variable are different through different parts (quantiles) of the distribution. We estimate three different quantiles of the CSR variable: the 25th quantile expresses low levels of CSR performance, 50th is the median level, where the 75th expresses high levels of CSR performance. Indeed, we assume that the relationship between family control and CSR performance may vary along with different levels of CSR performance. Standard deviations of the estimated parameters are obtained using the bootstrap method (Cameron et al. 2008). Bootstrap procedures can also handle the joint distribution of various quantile regression estimators, which can simultaneously test the slope parameters on different quantiles (Li et al. 2015).

Results and Discussion

Descriptive Statistics

Table 2 shows that the CSR score varies between 5.24 and 98.49, with an average of 69.567 and a standard deviation of 24.035. These results suggest that there is considerable variation in the CSR performance among French companies.

As for family control, Table 2 shows that on average, 43.94% of companies’ shares are held by family members. Family participation in capital reaches a maximum of 80.97% suggesting a concentration of family ownership in the French context. We also notice that 42.5% of family companies are managed by a CEO who is a family member. Regarding the identity of the company, 34.95% of the family-controlled firms in our sample hold the same name as the family. Besides, the average of the EXCESS variable is 1.13, which suggests that families have on average about 13% of voting rights above their cash flow rights.

Descriptive statistics in Table 2 also show that boards have an average of 13 members. Board independence is on average 43.42%. In addition, 43% of our sampled firms have a dual structure on the board where there is no separation between the CEO and the chairman positions. Finally, the proportion of women appointed on the boardroom is on average 20.28%. This proportion is likely to increase with the Copé-Zimmermann law that constrains companies to appoint a quota of 40% of women in the boardroom from 2017.

Table 2

Descriptive statistics

Descriptive statistics

Note: This table reports the descriptive statistics. See appendix 1 for variable definitions.

-> Voir la liste des tableaux

Before testing our hypotheses, we check for the lack of the multicollinearity problem between the independent variables. Table 3 shows the Pearson correlation matrix between our independent variables. The correlations do not exceed the threshold of 0.8 (Gujarati, 2004). For a deeper analysis, we calculate the VIF values. Table 3 shows that the average VIF is 1.48, which is far below the limit of 10 (Neter, 1986). We then confirm the lack of the multicollinearity problem.

Multivariate Analysis

Family Control and CSR Performance

Tables 4 and 5 report the results of the family control effect on CSR performance. Columns 1 and 2 report OLS and OLS firm-level fixed effects, respectively. We next present our quantile regression results in column 3.

Table 3

Pearson correlation matrix (N = 1164)

Pearson correlation matrix (N = 1164)

Note: Here, * indicates statistical significance at the 10% level. See appendix 1 for variable definitions.

-> Voir la liste des tableaux

The OLS regression in Table 4 shows a positive and significant relationship between family ownership and CSR performance at the 1% level. However, the OLS regression with fixed effects shows a non-significant relationship. These mixed results are based on the conditional mean of the CSR distribution. Using quantile regressions, we find that the effect of family ownership on CSR performance is positive and significant at the 25th and 50th quantiles of CSR and becomes insignificant at a higher CSR engagement level (75th quantile). The largest magnitude is recorded at the 25th quantile, with a coefficient of 0.141, compared to the median (0.106). These results suggest that there is a heterogeneous effect of family ownership on the CSR distribution. Specifically, family ownership has a stronger positive impact on low levels of CSR relatively to high-CSR engagement. This means that the effect of family ownership is stronger in low-CSR firms. These findings suggest that family firms are likely to enhance their low-CSR levels to preserve their socio-emotional wealth. However, when the firm is already highly engaged in CSR activities, families do not seek to enhance their CSR commitments. These results are consistent with those of Dyer and Whetten (2006) and support Hypothesis 1. Table 4 also shows the effect of family ownership on both CSR dimensions, namely the environmental dimension (ENVIRONMENT) and the social dimension (SOCIAL), respectively. The results show that family ownership is positively associated with low levels of both environmental and social dimensions.

Table 4

Effect of family ownership on CSR performance

Effect of family ownership on CSR performance

Note: This table presents a regression result of family ownership on CSR performance indicator variables and controls for the full sample, 2005-2016, inclusive. Coefficients and standard deviations are estimated at three different quantiles (25th, 50th, and 75th). Here, *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. See appendix 1 for variable definitions.

-> Voir la liste des tableaux

Table 5 reports the results on the effect of CEO family members and CSR scores. The quantile regressions show that the effect of the presence of a CEO family member on CSR performance differs across quantiles of the CSR variable. At the 25th, median and 75th quantiles, we find a positive and significant relation between CEO family members and CSR performance. However, the strongest effect is recorded for the 25th quantile (0.418), compared to the median (0.137) and the 75th quantile (0.061). These findings suggest that family members with both financial and human capital involved within the company are more committed to increase low-CSR performance compared to companies with already high-CSR levels. Our empirical results are consistent with those of Cui et al. (2018) suggesting that family firms with a CEO family member are likely to invest in CSR activities in order to preserve the socio-emotional wealth of their family, as predicted in Hypothesis 2. Table 5 also shows that the positive relationship between CSR performance and CEO family members holds for both CSR dimensions.

Regarding the identity of the family, Table 5 shows that when the company name coincides with the family name, families are likely to improve their low CSR performance. Indeed, the FAM-IDENT coefficient is positive and significant at the 25th quantile and at the median, and insignificant at the high-CSR level (75th quantile). This finding also supports the emotional perspective as the reputation of the family is important for all stakeholders. As a result, family firms are more inclined to enhance their CSR engagement; particularly when the firm commits less to CSR activities. This is consistent with the findings of Barnea and Rubin (2010) and confirms our Hypothesis 3. Similar results are reported for environmental and social scores. Particularly, the family identity is more sensitive to environmental improvements, which supports Berrone et al. (2010), who argue that the family’s desire to protect its socio-emotional wealth can lead to a higher environmental engagement.

We now consider excess family control. Table 5 shows that when the family holds control rights beyond their cash flow rights, there is a negative and significant effect across all the CSR distribution. Family-controlled firms have then less incentive to engage in CSR activities to privilege personal interests. This result confirms our Hypothesis 4. This finding suggests that when families hold excessive control, the achievement of financial goals prevails over SEW preservation motives, leading to a decrease in CSR performance. This finding supports the idea that in countries where minority interests are weakly protected, family firms with voting rights over their cash flow rights are likely to behave opportunistically (La Porta et al. 1999). Thus, excess family control is associated with high agency costs and with an expropriation risk of minority interests.

Overall, our quantile regression results show that the effect of family control on CSR performance depends on the family’s involvement in the business and also on the CSR engagement level.

For the control variables, CSR has a positive relationship with firm size for the 25th, 50th, and 75th quantiles. Nevertheless, the positive effect (Waddock and Graves, 1997 and Labelle et al. 2018) is stronger at the lowest level of CSR performance. This result shows that large companies increase their weak CSR performance. As for the effect of debt, the results show a negative and significant effect on all quantiles. This finding suggests that highly indebted companies adopt a less socially responsible behavior (Barnea and Rubin, 2010 and Labelle et al. 2018). The ROA variable positively affects social performance as found by Labelle et al. (2018). However, the strongest magnitude is recorded for the lowest quantile, suggesting that companies with better financial performance (ROA) and low-CSR scores are more likely to improve their CSR engagement.

The Moderating Effect of Board Attributes

We now test the moderating effect of board attributes on one feature of family control i.e. excess control. Table 6 reveals that using the quantile regression approach, the impact of board size is negative. For small-size boards, family excess control increases firm’s engagement in CSR activities, particularly, low-CSR levels. Hence, in family firms, a small board is able to control the opportunistic behavior of family members toward stakeholder interests. Indeed, family firms would be able to further improve their low-CSR levels.

Table 6 shows also that the interaction terms between board independence and excess control (EXCESS*INDEP) and duality functions (EXCESS*DUALITY) are insignificant using both conditional mean and quantile regression techniques.

Table 6 shows that the interaction term between excess control and diversity is positive and significant at the 25th quantile and at the median, but insignificant at the upper CSR quantile. These results suggest that the effect of excess control on CSR performance turns positive when women are appointed on the boardroom. This finding suggests that board gender diversity helps monitoring family actions and encourages the family to enhance at least their weak engagement in CSR activities. This finding confirms that women on boards constrain the expropriation behavior of the controlling family regarding CSR commitment, as their presence is considered an effective control device. It also supports the Copé-Zimmermann law adopted in 2011, which has required that French companies appoint women to a minimum quota of 40% of board directorships from 2017. This result supports our hypothesis 8.

Table 5

Effect of excess family control, a CEO family member, and identity on CSR performance

Effect of excess family control, a CEO family member, and identity on CSR performance

Note: This table presents regression results of excessive family control, a CEO family member, and family members’ identification with the company on CSR performance indicator variables and controls for the full sample, 2005-2016, inclusive. Coefficients and standard deviations are estimated at three different quantiles (25th, 50th, and 75th). Here, *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. See appendix 1 for variable definitions.

-> Voir la liste des tableaux

Table 6

The moderating effect of the board of directors

The moderating effect of the board of directors

Note: This table presents a regression of the moderating effect of the board of directors’ indicator variable and controls for the full sample, 2005-2016, inclusive. Coefficients and standard deviations are estimated at three different quantiles (25th, 50th, and 75th). Here, *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. See appendix 1 for variable definitions.

-> Voir la liste des tableaux

Robustness Checks

We perform an additional analysis to check the robustness of our results. We use an alternative estimation technique i.e. the binomial negative regression. Indeed, our dependent variable, the CSR score, is a count data variable that varies between 0 and 99. The negative binomial estimation is then more appropriate for count data because it helps overcoming the problem of over-dispersion in data (Allison, 2009). The results reported in Table 7 remain qualitatively unchanged and show a positive relationship between family identity and involvement in capital and management and CSR performance. However, a negative relationship is reported between excessive family control and CSR performance.

Discussion and Conclusion

The purpose of this paper is to investigate the effect of family identity and involvement in capital and management on CSR performance for French-listed companies. It also sheds light on the moderating effect board features have on the relationship between excess family control and CSR performance.

Using the quantile regression approach, we find that the effect of families on CSR performance is heterogeneous across the CSR distribution. More specifically, we show that family ownership and identity have a positive impact on CSR performance, particularly in firms with low-CSR performance. In addition, the results show that family involvement in management improves all CSR levels supporting the socio-emotional perspective. However, our findings show that families that have excess control engage less in CSR activities suggesting that controlling families may have expropriation purposes and are likely to privilege their personal interests.

Our findings are important twofold: first, unlike Ducassy and Montandrau (2015), we underline the heterogeneity of French family businesses. We prove that family behavior toward CSR varies with family involvement within the company by focusing on two dimensions of the SEW perspective: family control and influence and identification with the firm. Second, by extending previous studies on family firms and CSR performance (Dyer and Whetten, 2006; Berrone et al. 2010; Déniz and Suárez, 2005; El Ghoul et al. 2016), we show that the effect of family control on CSR performance depends on the level of engagement in CSR activities, using the quantile regression approach.

Table 7

Binomial Negative regression

Binomial Negative regression

Note: This table presents a negative binomial analysis. See Table 1 for the variable definitions. Here, *, **, and *** indicate statistical significance at the 10%, 5%, and 1% levels, respectively. See appendix 1 for variable definitions.

-> Voir la liste des tableaux

We also investigate the moderating effect of board features on the relationship between family control and CSR performance in a situation where the family has an excess of control. The findings show new evidence on the importance of gender diversity and small boards as effective devices able to mitigate the negative effect of excess family control on CSR performance.

In terms of theoretical implications, our findings support that both socio-emotional and economic considerations are important in family businesses. By using both perspectives, we contribute to the literature on CSR for family firms. We argue that, the effect of family control with regard to CSR performance is dynamic as it is affected by both owners’ socio-emotional wealth (SEW) and economic goals. Indeed, our findings show that the SEW motive does not necessarily lead to proactive stakeholder management suggesting a trade-off between incurring economic costs of CSR activities and preserving the family’s socio-emotional wealth. Our results suggest that family firms can engage in harmful stakeholder behavior when they control decisions through excess control. Accordingly, the family behavior regarding CSR activities is related to the degree of family involvement within the firm.

These results have practical implications as they help managers, shareholders, and stakeholders analyzing family behavior regarding CSR engagement. Market participants must be aware that CSR decisions in family firms vary depending on the family’s involvement in the business. Our results provide useful insights on the benefits in CSR performance of family identity and involvement in capital and management. However, when families have excess control through double or multiple shareholdings or through pyramiding, the family behavior could harm stakeholders’ interests as controlling families engage less in CSR activities to privilege their own interests. Policymakers may enforce then controlling mechanisms to encourage companies in general and family firms in particular enhancing their CSR practices.

However, this study has some limitations. First, we only focus on two SEW dimensions (i.e., family control and influence and identification of family name). Thus, significant work remains to be done to deepen our understanding on the heterogeneity of family businesses. In particular, future research could focus on different SEW dimensions such as intergenerational succession, the age of family business, values, and culture. Second, to test the moderating effect of the board, we focus on four features traditionally tested in literature. Other attributes such as director experience and education may also be tested.