Contemporary research has emphasised the importance of the monitoring role played by independent boards in the settlement of agency conflicts, which usually arise because of the long-term nature of CSR investments. Generally, CSR activities demand considerable long-term investments without instant financial rewards. Hence, self-serving managements might abstain from taking CSR-related projects because of its short tenures (Liao, Lin, & Zhang, 2018). However, CSR investments are anticipated to provide long-term benefits to stockholders by virtue of easy access to finance, improved reputation and reduced risk (Saeidi et al., 2015). This long-term nature of CSR investments leads to divergence of interest between the principal and the agents (Albuquerque, Koskinen, & Zhang, 2018; Tauringana & Chithambo, 2015). As self-serving activities of managements cannot be countered by legislative action, the monitoring role of corporate boards becomes increasingly important in such situations (Roe, 1991).

Research demonstrates that besides economic resources, a fundamental element required by companies to initiate sustainability-related activities is relational and human capital (Tate & Bals, 2018). Projects related to CSR require people-intensive structures and cross-functional coordination for successful designing and execution (Pohl & Tolhurst, 2010). Generally, independent directors contribute to a firm’s relational and human capital by providing professional expertise, competencies, external links and unique skills (Chen, Hsu, & Chang, 2016). This valuable human capital helps firms in resolving environmental uncertainties, acquiring vital resources and managing external dependencies, resulting in enhanced sustainability performance (Ramón-Llorens, García-Meca, & Pucheta-Martínez, 2018). In the same vein, Liao, Luo and Tang (2015) maintained that independent directors create environmental opportunities by utilising their specific expertise. Furthermore, literature also cites that diverse backgrounds and good stakeholder orientation edify independent directors’ capabilities in maintaining a trade-off between environmental and financial accountability (Al-Dah, 2019). This strengthens corporate boards towards attaining a fine balance between short- and long-term goals, leading to varying interests of multiple stakeholders being accommodated.

In addition, the independent directors remain cautious for stakeholders’ demands; hence, they develop CSR campaigns based on innovative vision, which ultimately enhances firm’s competitiveness and prestige in the society (Garcia-Sanchez, 2014) and also proves to be helpful for the continuation of their directorship (De Villiers, Naiker, & Van Staden, 2011). Independent directors have a good stakeholder orientation because of their diverse backgrounds and are generally more capable of managing external contingencies (Liao et al., 2015). Moreover, the lack of financial stake of these directors makes them more capable of balancing between financial and social accountability (Wang & Dewhirst, 1992). This shows that they would focus on a broader view of firm performance as opposed to the narrow view, which is all about financial gains. The above-mentioned arguments lead to the first hypothesis of the study:

Our next research objective was to examine if family control moderates the relation of board independence and CSR. The two contrasting theoretical opinions regarding the association between family businesses and corporate social performance are found in the literature.

One line of research suggests that family firms are less socially responsible. In this context, Kellermanns, Eddleston and Zellweger (2012) pointed out that family firms are more concerned with financial returns, owing to their huge financial investments in the firms. Singal (2014) showed that family firms with ample resources considered sustainability activities merely as an expense and not as a management strategy or an opportunity to improve firm performance. In addition, family firms are found to behave responsibly as well as irresponsibly at the same time. They show responsible behaviour for external stakeholders in a bid to save reputation and compromise when dealing with internal stakeholders for ensuring control (Cruz, Larraza-Kintana, Garcés-Galdeano, & Berrone, 2014).

Another line of research with a socio-emotional wealth (SEW) perspective argues that family businesses generally focus on non-economic goals, for instance, longevity, preservation of reputation in public and identity (Blodgett, Dumas, & Zanzi, 2011). Family firms are guided by the desire of passing down a viable heritage; hence, family businesses indulge in sustainability activities so that future generations can inherit a firm with a positive image (Miller & Le Breton-Miller, 2006).

The diverse results based on the views of agency and SEW perspectives can be understood in the light of contingency theory, which suggests that the relationship between two variables can vary with context or culture (Mokhtar, Jusoh, & Zulkifli, 2016). In this background, Bing and Li (2019) showed that indulging in CSR activities may adversely affect the financial performance (FP) of firms, which explains that family businesses would opt for CSR engagement only in case it adds to firms’ reputation and welfare (Zellweger, Nason, Nordqvist, & Brush, 2013). In this context, literature has found that family involvement in firms is negatively linked with sustainability practices (Labelle, Hafsi, Francoeur, & Ben Amar, 2018; Nekhili, Nagati, Chtioui, & Rebolledo, 2017). Moreover, the long-term nature of CSR pay-off might clash with the risk-averse nature of family ownership (Chrisman & Patel, 2012), which explains why family firms avoid investing in CSR activities.

These discussions consequently point towards a strong conflict of interests between independent directors and family directors. Generally, family firms are majorly owned and managed by family members (Chu, 2011). This distinguishing feature of family firms may affect the behaviour of board members, including independent directors. Furthermore, a great degree of nepotism is observed when it comes to the selection of directors in family firms and especially in the appointment of outside directors (Gabrielsson & Huse, 2005). In brief, family businesses generally resort to informal control mechanisms, for example, hiring and promotion of candidates who carry the same values as that of the family members (Collin & Ahlberg, 2012). Consequently, principal–agent conflict is resolved but at the same time principal–principal conflict is raised, which increases the influencing ability of family members in calling the shots and ultimately weakening the true spirit of the CG system. Accordant with this logic are the findings of García-Ramos & García-Olalla (2011), which further support the idea that an independent director’s role in family-led firms is ceremonial and is limited to just providing resources and not performing a monitoring task. Therefore, board meetings in family firms are conducted just to give a formal approval to what the family directors have already decided (Gabrielsson & Huse, 2005). Against this background, Leung, Richardson and Jaggi (2014) provided evidence that the positive relation between board independence and firm performance is weaker in family-led firms. This indicates that the effectiveness of independent board is jeopardised under family leadership. Along the same lines, Cho and Kim (2007) showed a weak positive impact of independent directors on firm performance, which is mitigated by the concentrated ownership structure in a negative fashion. Although Gomez-Mejia, Cruz and Imperatore (2014) reported a positive impact of founding family corporations on financial disclosure, Darmadi (2013) showed that family presence curtails the effectiveness of an independent board.

Recent studies (García-Sánchez, Gómez-Miranda, David, & Rodríguez-Ariza, 2019; Jizi, 2017; Muttakin, Khan, & Mihret, 2018) have investigated the relationship between board independence without including or controlling for the ownership pattern in their research models. This consequently leads to biased results because CG literature shows that ownership is one of the major determinants of CSRP (Hu, Zhu, Tucker, & Hu, 2018). Some studies that considered ownership structures either did not control for endogeneity effectively (Bansal, Lopez-Perez, & Rodriguez-Ariza, 2018; Chang, Oh, Park, & Jang, 2017) or employed a sample from developed countries (Cuadrado-Ballesteros, Rodríguez-Ariza, & García-Sánchez, 2015). Du, Weng, Zeng, Chang and Pei (2017) proposed that CSR is country-specific and recommended to specially study the emerging markets for this reason. Our study contributes to the CSR literature with evidence from one of the largest developing markets in South Asia, namely, China.

In line with the above arguments, we hypothesise that independent directors discourage participation in CSR activities owing to family members’ desire. Based on the above discussion and empirical evidence, the following hypothesis is proposed about the moderating role played by family control:

China is an ideal setting for this study because family-led business structures are comparatively more prominent in emerging economies (Yu, Lumpkin, Sorenson, & Brigham, 2011). In China, corporate ownership is centralised in the hands of a few families (Yuan, Hua, & Junxi, 2008). Moreover, through the mechanism of pyramidal ownership, controlling families usually enjoy extra power and consequently have more potential as well as incentive to expropriate the stakes of minority shareholders (García-Ramos & García-Olalla, 2011). These family-led businesses routinely appoint friends and family on governance boards and in management positions to maintain control and protect their personal interests. Given the structure of ownership, the nature of agency problem in China is very different from the Western world, which is intensified by the weak market control mechanism. Such a situation turns into a special case for independent directors, where they are expected to safeguard the rights of minority shareholders of family-led firms.

Rankins (RKS) database is employed to obtain CSR ratings of our sample firms, which has been widely used in CSR research in the Chinese market (Liao et al., 2018). Corporate social responsibility performance variable is represented by log of CSR ratings released by RKS.

China Securities Regulatory Commission (CSRC) advocates board independence and clearly states that one-third of the listed companies’ board shall comprise independent members. A dummy variable is created on the basis of this one-third criterion, where it is coded as 1 if more than one-third of the directors are independent and 0 otherwise. The CSRC (2001) Guidance Opinion describing an independence test for determining board independence is followed in counting a director as independent or not.

Majority of the businesses worldwide are run by a dominant shareholder, typically families (Meglio, 2019). Following Biswas, Roberts and Whiting (2019), proportion of family members on board is used as a proxy for family presence (Fam_P), which is calculated as the number of family members on the board divided by board size. Secondly, proxy used for family firms is the family ownership because ownership stake reflects the power an owner holds to influence management decisions. Following Jaggi, Leung and Gul (2009), we classify a firm as family-led at a cut-off of 20% ownership. A dummy variable (Fam_O) equals 1 if more than 20% ownership belongs to family members and 0 otherwise.

Following the literature on CSR, various control variables are included in our regression model to separate the impact of board independence on CSR engagement of a firm. Governance controls include multiple directorships (MD), which is natural log of the average number of other directorships held by board members. Chief Executive Officer (CEO) duality (CEO_Dual) is measured by a dummy variable that takes the value of 1 if the CEO is also the board’s chairman and 0 if a single person does not hold both positions. Presence of CSR committee in a firm is captured by a dummy variable (CSR_Com) having a value of 1 in case of a firm having a separate CSR committee and 0 otherwise. Firm controls include firm size (Size) calculated as log of total sales to standardise absolute values of total sales and to evade the non-normality of the distribution. As a proxy for Slack (SLK), current ratio is used, which is measured as current asset by current liabilities. We control for firm’s FP by including Return on Asset as its proxy, which is calculated as a quotient between operating profit and total assets. Firm’s indebtedness (INDEBT) is measured as the quotient between total liabilities and total assets. Finally, to control for industry effects (IND), the dummy variables are used because it is possible for CSRP to fluctuate between industries. If the company is active in one of the industries, the dummy variable is assigned a value of 1 and 0 otherwise. Year fixed effect (YEAR) is also controlled for any time trend present in the data.

Table 1 presents the descriptive statistics and correlations for regression variables. To control for the effect of outliers, all continuous variables are winsorised at 1% and 99% of the distribution. The mean score for CSRP is 2.99, which shows that CSRP for companies averaged about 2.99 over the study period. It is evident from the standard deviation that there exists a high degree of heterogeneity amongst the sample firms with regard to their sustainability performance. Moreover, the fraction of firms having an independent board is very low in our sample; the mean value of Bind is 0.15, which means that 15% of the firms in the sample have independent boards meeting the one-third criterion of CSRC in its true spirit.

Finally, we note that there is a reasonable degree of family presence in Chinese firms. Thus, we conclude that on average, the non-financial firms in China are headed by family boards. The correlation matrix indicates that Bind positively correlates with CSRP (0.21, p < 0.01), whereas Fam_P and Fam_O are negatively correlated with CSRP. All of these findings are aligned with our hypotheses. In addition, MD, CSR_Com, FS, SLK and FP positively correlate with CSRP, whereas CEO_Dual and INDEBT negatively correlate, which is consistent with the results of previous studies (Cruz et al., 2014; Haque, 2017).

There is a possibility that CSRP is endogenous, that is, there may be a reverse causality between a firm’s sustainability performance and its board independence. To address the concerns over the selection of firms maintaining an independent board, a common approach is to find an instrumental variable (Heckman, 1974). In search of an instrumental variable, we estimated alternative Heckman models but could not find an appropriate one. Thus, to account for the issue of endogeneity, we adopted PSM. Accordingly, a matched sample is selected for firms with independent boards (treated group) and for firms without independent boards (control group), which are in other respects sufficiently similar. This control group has firms in which the board is not independent but otherwise matches with the attributes of the treated group. In the first step, to create these groups, we follow prior literature (Rosenbaum & Rubin, 1983) and employ a logit model on the full sample for determining the probability of the firm in having an independent board: Bindi,t=β0+β1Sizei.t+β2Growthi.t+β3INDEBTi.t+εit)[Eqn 3]

Following Heckman and Navarro-Lozano (2004), we include variables in the logit model presented below to estimate the propensity score that influences both the board independence and the CSRP.

The results reported in Table 2 show that on average firms with independent boards are larger and have more growth opportunities. In addition, the statistically significant likelihood ratio (X² = 50.08, p < 0.001) shows the combined significance of the explanatory variables and therefore indicates a good model fit.

In the second step, we used the nearest neighbour method with replacement and the calculated propensity scores to compare firms with independent boards (treatment group) with a sample of control firms without independent boards (control group) matched with the propensity to have an independent board. The propensity-score-matched control sample is generated by matching each firm from the treatment group with up to three firms from the control group having the nearest propensity score. We bound the calliper distance between propensity scores with 5% so that the absolute difference between the propensity scores for firms with independent boards and its matched peers did not go over 0.05.

In the third step, the matched and unmatched samples are compared. We expect that firms with independent boards (treated group) and without independent boards (control group) are, on average, different before matching, whereas no significant difference is expected to be present after the procedure of PSM. The matched sample generates 1108 firm-year observations of which 554 observations have independent boards and rest of the observations do not have independent boards. Results of covariate balance analysis are presented in Table 3 to demonstrate the similarity between the treated and control groups, pre- and post-matching. The results indicate that before matching, there was a significant difference between firms in the following characteristics: size and growth. These results point towards the problem of potential self-selection bias in the original sample. Hence, PSM removes the observable difference between each of the characteristics of treatment and control group, whereas the difference between CSRP of these groups remains significant.

Finally, Equations 1 and 2 are estimated using matched samples with weighted standard errors to control for potential endogeneity in data. These weights are calculated by considering the frequency of firm-year observations in control group used as matches. Table 4 presents the regression results of PSM sample and shows that Bind significantly increases the CSRP of firms (coefficient [coef.] 0.231, p < 0.01). The positive association found between Bind and CSRP allows us to support our first hypothesis, that is, an independent board leads to increased commitment of firms to engage in CSR activities. This result is similar to Haque (2017), who also documented a positive association between the presence of independent boards and CSR engagement of firms. Interaction effects in Table 4 present the result of examining Equation 2, that is, the possible moderating impact of family control on this relationship by introducing Fam_P and Fam_O and their interactions with Bind.

Results show that the interaction term ‘BindFam’ has a significantly negative impact on CSRP, whilst Bind has still a significantly positive impact, implying that family presence can diminish the positive influence of independent boards on CSRP of firms. This result indicates that independent directors are reluctant about firm’s participation in sustainability activities in family firms (coef. Bind + coef. BindFam = 0.290−0.460 = −0.170) rather than in non-family firms (coef. 0.290). Results are similar for the additional ‘Fam_O’ proxy of family-led firms. Overall, the obtained results provide an empirical support to the second hypothesis of this study regarding the moderating effect of family firms on the relationship between the presence of independent boards and sustainability performance. These results imply that the presence of family in business subdues the independence of independent directors. It indicates that independent directors are generally unable to make any significant contribution for CSR promotion, contrary to what is shown in Western literature. Therefore, independent directors need to be empowered so that they may effectively contribute in corporate decisions. Our results are in accordance with Yusuf, Yousaf and Saeed (2018) that board’s independence is generally weakened by the dominant presence of strong social ties and networking within the board. Hence, independent directors also become partners with the management of the firm and thus may not get any forum for exercising their independence.

Regarding control variables, we find them significantly associated with firm’s CSRP. Specifically, the regression coefficient for Size is significantly positive, suggesting that larger firms have a positive implication for CSR activities. With the growing size of companies and the consequent increase in their resources, they generally develop CSR plans and aggressively participate in CSR activities in order to satisfy the demands of various stakeholders (McWilliams & Siegel, 2001). The estimated coefficient for CEO_Dual is negative because CEO duality results in agency conflicts. It also negatively impacts the effectiveness of the board’s monitoring role (Goyal & Park, 2002), which may reduce the chances of embarking on long-term investment for sustainability projects. Similar to the argument of Peters and Romi (2013), the regression coefficient of CSR_Com is significantly positive. This coefficient shows that the presence of CSR committee symbolises the firm’s commitment towards sustainability. The SLK coefficient reported is positively significant, which indicates that the firm’s decision to show good citizenship behaviour depends on the resources available. Hence, firms with ample SLK resources devote more funds to sustainability-related activities. According to resource-based view (RBV) and similar to the results of Qiu, Shaukat and Tharyan (2016), the coefficient of FP implies that companies with sufficient financial resources are more likely to be engaged in sustainability-related activities. Significantly, negative coefficient of INDEBT suggests that the payment of interest by highly leveraged firms impedes their spending on social activities because creditors exert pressure on their spending decisions (Brammer & Pavelin, 2008). Moreover, positive coefficient of MD shows that holding chairs in multiple boards certifies a director’s knowledge and diverse experience which help them implement various sustainability practices in their firms.

The graph in Figure 1 shows a buffering interaction effect where family weakens the effect of board independence on CSRP of firms.