Corporate social responsibility, business groups and financial performance: a study of listed Indian firms

This study explores the relationship between corporate social responsibility (C.S.R.) and financial performance of Indian firms. We also examine the relationship between C.S.R. and financial performance in context of Indian business group firms, which are known to have unique characteristics which differ from those of Indian stand-alone firms. Using a sample of Indian listed firms between 2010 and 2015, we find that C.S.R., as measured by E.S.G. disclosure score, has a U-shaped relationship with Tobin’s Q, supporting the slack resource theory at lower level of CSR and supporting the stakeholder theory at higher level of C.S.R. The empirical results imply that an improvement in CSR actions does not always result in higher firm value but should exceed a certain level of C.S.R. to have a positive effect on firm value. In addition, we find that at lower level, a negative relationship between C.S.R. and Tobin’s Q weakens in group affiliate firms. However, this complement effect of business group disappears at higher level, weakening the positive relationship between C.S.R. and Tobin’s Q. This study offers new insights for the different influence of business groups on C.S.R. performance. ARTICLE HISTORY Received 5 January 2018 Accepted 17 December 2018


Introduction
Of late, corporate social responsibility (C.S.R.) has been considered a very important factor in management. In particular, from the viewpoint of shareholder, the issue is whether C.S.R. may serve to enhance corporate financial performance. In this context, the relationship between C.S.R. and firm performance has been explored by many authors (Waddock & Graves, 1997;Griffin & Mahon, 1997;Lin, Yang, & Liou, 2009;Saeidi, Sofian, Saeidi, Saeidi, & Saaeidi, 2015). Numerous studies have attempted to empirically examine these relationships. For instance, while one strand of literatures suggests that C.S.R. has a positive impact on firm performance (Oeyono, Samy, & Bampton, 2011), another strand shows that C.S.R. has a negative impact on firm performance (Servaes & Tamayo, 2013). Moreover, these mixed results have led to at the low level of C.S.R., Tobin's Q decreases with C.S.R. while at the high level of C.S.R., Tobin's Q increases with C.S.R. More importantly, even after controlling for firm specific variables, these empirical results still show significantly non-linear relationship. However, the effect of C.S.R. on R.O.A. is insignificant in Indian firms, implying that the impact of C.S.R. is related to future firm value but not related to accounting performance. Second, this study finds that the C.S.R. initiatives of groupaffiliated firms are higher than those of stand-alone firms but the relation between C.S.R. and firm value is less pronounced in group affiliated firms than in stand-alone firms. The results suggest that group affiliated firms are inclined to actively engage in social activities to enhance group's reputation. This study contributes towards the extant C.S.R. literature by empirically examining the relationship between C.S.R. and financial performance in India.
The remainder of the article is organised as follows. Section 2 presents literature review. Sections 3 and 4 present hypothesis and test methodology. Section 5 presents our empirical results on CSR and financial performance. Section 6 provides summary and conclusion.
For instance, Tsoutsoura (2004) documents that CSR has a significant positive effect on firm performance in the S&P 500 firms during 1996-2000. He supports the view that socially responsible corporate performance may be associated with a series of bottom-line benefits. Oeyono et al. (2011) find that C.S.R. has a positive impact on Indonesia's top 50 firms. Lin et al. (2009) document that C.S.R. has a positive impact on financial performance by analysing Taiwan's 1,000 firms that consider their R&D expenditures as one of their business strategies for sustainable development during 2002-2004. They suggest that while C.S.R. does not have a significant positive impact on short-term financial performance, it offers a remarkable long-term fiscal advantage. However, Smith, Yahya, & Marzuki Amiruddin (2007) document a negative impact of C.S.R. on firm performance. Recently, literature examines the nonlinear relationship between C.S.R. and firm performance. For example, Nollet, Filis, & Mitrokostas (2016) examine the nonlinear relationship between corporate social performance and return on capital of the S&P firms during 2007-2011. They find the evidence of a U-shaped relationship between corporate social performance and accounting based measures of financial performance.
Recent studies on C.S.R. and financial performance link in India remain inconclusive. Mishra & Suar (2010) investigate whether C.S.R. influences firm performance in India. They find that managers' initiatives regarding C.S.R. have a positive impact on firm performance. Further, they argue that Indian firms should implement C.S.R. for the benefit of their stakeholders. Rajput et al. Rajput, Batra, & Pathak (2012) document that C.S.R. has a positive impact on financial performance. Padhi (2013) observes that after C.S.R. has been categorised on the basis of government, firm, and individual, it has a significant impact on firm performance. Kumar (2014) documents that C.S.R. in India has a long-term view from late 1980s. He suggests that the philanthropic ideas of Indian firms based on religious beliefs are similar to those of the West. In particular, C.S.R. has been performed in various forms such as philanthropic contribution to charity, service to local community, and an increase in employment. C.S.R. activities can decrease firm's earning in the short-term, but in fact contribute to firm's earnings in the long-term. More recently, Garg (2016)  He finds that C.S.R. impacts the value of Indian corporate sector highly and significantly using panel regression. He suggests that C.S.R. performance of companies impacts corporate performance not only for current year, but also for the following years. Bihari and Pradhan (2011) suggest that C.S.R. activities in Indian banks have positive impact on performance. However, some authors find mixed results on the relationship between C.S.R. and financial performance. For instance, Saxena and Kohli (2012) find that CSR in India has an insignificant impact on firm performance in banking industry. Aggarwal (2013) empirically examines the relationship between C.S.R. and financial performance, suggesting mixed and inconclusive results.
Several studies have empirically examined the C.S.R. performance from business group standpoint (Choi et al., 2018;Guo, He & Zhong, 2018). For example, Choi et al. (2018) find that group affiliated firms in Korea is associated with higher C.S.R., while ownership disparity between cash flow and control by controlling insider shareholders is associated with lower C.S.R., supporting opportunistic rent expropriation theory. On the other hand, business group and a state-owned enterprise (S.O.E.) at the same time have weaker C.S.R. performance, indicating that C.S.R. engagement is a strategy for companies to pursue political justification in China. However, until now it is not clear how C.S.R. affects financial performance in the business group.

Theoretical background and hypotheses development
3.1. Relationship between C.S.R. and financial performance The stakeholder theory documents that treating and managing for stakeholder helps a firm enhance value, that is, good for firm performance (Freeman et al., 2004;Waddock and Graves, 1997). According to the instrumental stakeholder theory, C.S.R. appears to be helpful to firms. That is, the satisfaction of various stakeholder groups is instrumental for financial performance. Moreover, the stakeholder-agency theory suggests that the implicit and explicit contract entailed by bilateral stakeholder-manager relationships serve as monitoring and enforcement mechanisms that curb opportunistic behaviour by managers at the expense of other stakeholders (Hill & Jones, 1992). Further, by addressing and balancing the claims of multiple stakeholders (Donaldson and Preston, 1995), managers can increase the efficiency of their organisation's adaptation to external demands (Orlitzky et al., 2003). As a strategy, C.S.R. promotes a firm's competitive advantage by weighing and addressing the claims of various constituents in a fair and rational manner. For instance, firms may improve their image, reputation and brand name through C.S.R. activities (Benito-Hern andez, Platero-Jaime, & Esteban-S anchez, 2016).
According to the slack resource theory, however, a company is able to carry out its activities by the resources owned by the firm (Fauzi & Idris, 2009). Slack resources are defined as the pool of resources in an organisation that is in excess of the minimum necessary to produce a given organisation output (Nohria & Gulati, 1996). Patten (2002) suggests that C.S.R. may lead to further consumption of firm resources such as time, labour, and capital. Hence, firms may spend slack resources in order to enhance C.S.R. activity. Specifically, a firm should hold a good financial position to contribute to the corporate social performance. Further, some authors argue that social activities involve financial costs and siphon off capital and other resources from firm (Aupperle et al., 1985;Preston & O'Bannon, 1997).
According to the above arguments, the relationship between C.S.R. and firm performance may have U-shape relation. In other words, an increase in C.S.R. may make management resource more aggravated, thereby reducing financial performance. Specifically, investments in C.S.R. may decrease financial slack.
On the contrary, however, an increase in C.S.R. may align with the shareholder as well as stakeholders, thereby increasing financial performance. As indicated by Kumar (2014), C.S.R. in India can contribute towards firm's earnings enhancing the firm's reputation, brand value and interest convergence of stakeholders.
Taken together, as C.S.R. scores are at a lower level, C.S.R. investment costs may be reflected immediately in firm performance but its benefits may not materialise. Thus, we expect a negative effect of C.S.R. scores on financial performance at a lower level, indicating that slack resource theory dominates at a lower level of C.S.R. scores. However, as C.S.R. scores are above a certain level, C.S.R. investment costs may increase marginally but its benefits may start being reflected in firm performance at last, because C.S.R. investments have accumulated enough to start reaping the benefits by improving firm's image, reputation, and/or brand name. Thus, at a higher level of C.S.R. scores, the relation between C.S.R. scores and firm performance may change from a negative one to positive one, when C.S.R. investment benefits become greater than its costs. Based on the above rationale, we hypothesise that: Hypothesis 1: The relationship between C.S.R. and financial performance is negative at a low level of C.S.R. scores and becomes positive at a high level of C.S.R. scores.

Business groups and C.S.R.
Indian business groups are known to have unique characteristics. Business groups play an important role in the Indian economy, and have been doing so during a large part of the twentieth century (Kali & Sarkar, 2005). Indian business groups strive to serve their communities and the society at large, historically considering social recognition. Indian business groups have persisted more with C.S.R. activities than have stand-alone firms. Group-affiliated firms intend to share the vision and objectives of the group. Hence, their shared C.S.R. activities may result in the enlargement of group's C.S.R. Khanna & Yafeh (2007) indicate that group-affiliated firms are interested in group reputation for risk sharing. Furthermore, Gopalan et al. (2007) argue that group-affiliated firms support financially distressed member firms in order to maintain the group's reputation. In addition, Choi et al. (2018) argue that C.S.R. can be used as a means of improving reputation capital to buffer the bad events.
According to these views, Indian group-affiliated firms share their C.S.R. activities not only to increase group's reputation but also to share risks and costs. On the other hand, compared with stand-alone firms, group-affiliated firms may need more C.S.R. investment costs and more time until they can reap the group's C.S.R. benefits. Cuervo-Cazurra (2018) argue that business groups may need higher costs as C.S.R. activities increase, due to the rapid growth and diversification of business groups.
In sum, at a lower level of C.S.R. scores, C.S.R. investment costs may be shared among group-affiliated firms, resulting in less negative effect of C.S.R. on financial performance, compared with stand-alone firms. On the other hand, for group-affiliated firms, C.S.R. investment costs may still occur considerably until their C.S.R. scores reach a certain level, resulting in less positive effect of C.S.R. on financial performance than for stand-alone firms at a higher level of C.S.R. scores. Based on this rationale, we hypothesise that: Hypothesis 2: Business group weakens the non-linear relationship between C.S.R. and firm performance.

Data collection
To test the hypotheses, we use annual accounting data for all Indian firms listed on the National Stock Exchange (N.S.E.) and Bombay Stock Exchange (B.S.E.) during 2010-2015. This study excludes the data offered by financial companies from the sample, since the financial policy of such firms is often driven by regulatory aspects. All data used are obtained from Bloomberg. The sample starts from 2010 owing to limited environmental, social, and governance (E.S.G.) data. It excludes firms that have either incomplete financial data or negative asset values. The sample includes unbalanced panel 214 firms and 1,191 firm-year observations.

Research model
This article tests the hypotheses through t-tests and panel regressions and explores the relationship between C.S.R. and financial performance in Indian firms. First, the study uses a t-test to analyze the difference between the group affiliated firms and stand-alone firms. Second, this study uses the panel regression model given in (1) to test our hypotheses. We use this model to control for the time-invariant unobserved firm features that might be correlated with the explanatory variables in the model. Third, as shown in Table 4, there is an endogeneity problem indicating that companies with a high financial performance can have a high C.S.R. performance. It is well known that all of the pooled O.L.S., fixed effect and random effect models cannot solve the endogeneity problem in the presence of a lagged dependent variable. Thus, we use Arellano & Bond (1991)'s system generalised method of moments (G.M.M.) to solve this problem. The G.M.M. estimation greatly reduces the bias problem. More specifically, this study develops a model implied by Nollet et al. (2016): FP it is a firm financial performance indicator. This article uses two proxies for firm financial performance (Yu-Shu, Chyi- Lin, & Altan-Uya, 2015). First, Tobin's Q is a proxy for firm value (Jo & Harjoto, 2011;Luo & Bhattacharya, 2006). Tobin's Q is defined as the book value of total assets minus the book value of equity plus the market value of equity divided by the book value of total assets. Second, ROA is a proxy for firm performance (McGuire et al., 1988). ROA is defined as net income divided by the book value of total assets. In particular, Tobin's Q is a variable that measures firm value based on market value, and ROA is a variable that measures firm performance based on accounting data.
A formal definition of C.S.R. is the commitment of a business to contribute to sustainable economic development, working with employees, their families, the local community and society at large to improve their quality of life, which has been proposed by the World Business Council for Sustainable Development. According to this definition, C.S.R. has been considered as concept including environment, employee, community, and shareholder. ESG is the key variable to measure C.S.R., which is approximated by the Bloomberg's E.S.G. disclosure (Nollet et al., 2016). E.S.G. breaks it into three components: Environmental, Social and Governance. Environ, Social and Gov are variables to measure Environmental, Social and Governance of companies, respectively. In particular, E.S.G. is a proprietary Bloomberg score based on the extent of a company's E.S.G. disclosure. The score ranges from 0.1 for companies that disclose a minimum amount of E.S.G. data to 100 for those that disclose every data point collected by Bloomberg. Each data point is weighted in terms of importance, with data such as greenhouse gas emissions carrying greater weightage than other disclosures. The score is also tailored to different industry sectors. In this way, each company is only evaluated in terms of the data that is relevant to its industry sector. Specifically, based on the previous studies (Jo & Harjoto, 2011), we assume E.S.G. affects Tobin's Q.
In addition, square term of ESG is a proxy to examine nonlinear relationship between C.S.R. and financial performance. Financial risk (Leverage), Sales growth rate (SaleGr), Research and Development (R&D), Firm size (Lnasset), Profit (Profitability) serve as control variables. Leverage is a proxy variable that analyzes the impact of financial distress costs, and is measured as total debt divided by total assets. SaleGr is a proxy variable for growth rate. R&D is measured as R&D divided by assets. Firm size (Lnasset) is measured as the natural logarithm of total assets. Profit (Profitability) is measured as the earnings before interest and tax divided by total assets. Furthermore, since we expect that new law on C.S.R. policy may influence firm performance from 2013, we include a Policydummy variable, which takes the value of 1 during the new law enforcement period (i.e. 2013-2015) and zero otherwise. The parameter u i denotes the unobservable heterogeneity or the firm's unobservable individual effects, controlling for the particular characteristics of each firm. Finally, e it denotes the random disturbance. Table 1 Table 2 shows E.S.G. statistics for Indian listed companies. The second and the fourth columns in Table 2 show that E.S.G. and Social scores increased from 21.836 to 22.358 and from 13.819 to 21.676, respectively from 2010 to 2015. However, Environ and Gov scores decreased from 14.691 to 12.542 and from 46.328 to 46.019, respectively, from 2010 to 2015. Overall, for Indian firms, E.S.G. increases, but responsibility on environment and governance decreases. Namely, increase of C.S.R. in Indian firms seems to be mainly caused by a rapid increase of their social responsibility. More specifically, in 2013, E.S.G. increased sharply, but in 2015, E.S.G. decreased rapidly. These results suggest that government's attempts to enhance CSR are not effective.  Table 3 reports Pearson correlation coefficients for dependent and independent variables. Most of the control variables are significantly correlated with Tobin's Q or ROA at 1% significant level. In particular, Tobin's Q and ESG show a significantly positive correlation (0.117). And Tobin's Q shows a significantly positive correlation (0.079, 0.142, 0.138) with Environ, Social, and Gov, respectively. Tobin's Q and Leverage show a significantly negative correlation (-0.214), which means firm value decreases when it increases financial leverage. R&D shows a significantly positive correlation (0.152) with Tobin's Q while Salegr shows an insignificantly negative correlation (-0.021) with Tobin's Q. ROA and Tobin's Q show a significantly positive correlation (0.367). ROA and ESG show a significantly a positive correlation (0.152). In addition, the correlation matrix shows that there are no high correlations between independent variables, indicating the absence of a multicollinearity problem.

Comparison between high C.S.R. firms and low C.S.R. firms
This section examines whether the means of variables differ significantly between high C.S.R. and low C.S.R. firms. A parametric t-test is used for the comparison of the means. High C.S.R. and low C.S.R. firms are classified by the average of ESG. In Table 4, the average Tobin's Q for high C.S.R. firms is 1.749, significantly higher than 1.100 for low C.S.R. firms. In addition, the average ROA for high C.S.R. firms is 0.082, significantly higher than 0.048 for low C.S.R. firms. It indicates that high C.S.R. firms have higher performance and value. Furthermore, high C.S.R. firms have insignificantly higher Salegr and R&D (0.006) compared to low C.S.R. firms.

Comparison between group-affiliated firms and stand-alone firms
In this section, we investigate whether the means of variables differ between groupaffiliated and stand-alone firms through a parametric t-test. In Table 5, we divide the sample into two groups, namely, group-affiliated and stand-alone firms. The average ESG disclosure scores for group-affiliated firms is 25.687, significantly higher than 21.926 for stand-alone firms. In addition, this study examines sub-component (Environ, Social and Gov) of ESG. The average disclosure scores of Environ, Social and Gov for group-affiliated firms are 17.768, 21.682 and 48.494, significantly higher than 13.898, 16.803 and 46.105 for stand-alone firms, respectively. More specifically, disclosure score of Social shows the highest difference among sub-component, suggesting that group-affiliated firms hold higher financial resources than those of standalone firms. These are similar to the results of group firms and non-group firms in Korea (Choi et al.,2018), being different from the results of group firms and nongroup firms in China (Guo et al., 2018). Interestingly, the average Tobin's Q for group-affiliated firms is 1.072, significantly lower than 1.384 for stand-alone firms, indicating that group-affiliated firms have lower growth opportunity than that of stand-alone firms.

Panel unit root test
Before a panel regression analysis, we perform a panel unit root test to identify whether all variables in the model are stationary. If one of the variables used in the estimation is nonstationary, not only the result of estimation has a spurious regression problem, but also the estimated parameters might be biased. In order to avoid the spurious regression problem, all the variables need to be stationary in the estimation. The null hypothesis of non-stationary versus stationary in each variable is tested Table 5. Comparison between Group-affiliated firms and Stand-alone firms.

Group-affiliated firms
Stand-alone firms  using the group mean panel unit root test (Cheng, Liu, & Chien, 2010). We employ two different panel-based unit root tests, the Levin-Lin-Chu ADF (L.L.C.) (Levin, Lin, & Chu, 2002) and the Im-Pesaran-Shin PS ADF (I.P.S.) (Im, Pesaran, & Shin, 2003), to investigate the null hypotheses of unit roots of all the variables chosen in the models. Table 6 shows the results of panel unit root tests. The nulls of unit roots are all rejected, which indicates that all the variables are stationary. Thus, we perform a panel regress analysis.

Hausman specification test
Hausman specification test is used to determine which one of the alternative panel analysis methods such as fixed effects model and random effects model is more adequate in the panel regression model. Concerning this, H0 hypothesis suggests that "random effects exist" and H1 hypothesis suggests that 'random effects do not exist'. Table 7 shows that H0 hypothesis is rejected for the model, at the significance level of 1%, thus all of the individual effects in the models are not random, but fixed. In other words, fixed effects model is more effective than random effects model. Consequently, the panel data regression is analysed by the fixed effects model in this study.

CSR and firm performance
This section examines how CSR affects firm performance through a panel regression model and G.M.M. estimation. Columns (1) and (3) of Table 8 show the fixed effects panel regression results for the effects of ESG and ESG 2 on firm performance. Column (1) indicates the relationship between CSR and Tobin'Q is U-shaped, supporting Hypothesis 1. In particular, the coefficients of ESG and ESG 2 show -0.023 (t¼-2.159) and þ0.0003 (t ¼ 1.698), respectively. When ROA is used as the variable for firm performance, column (3) does not show the U-shaped relationship with no statistical significance. The coefficients of ESG and ESG 2 are -0.00002 (t¼-0.008) and -0.00001(t¼-0.027), respectively. Columns (2) and (4) of Table 8 show the G.M.M. estimation results for the effects of ESG and ESG 2 on firm performance. They also indicate the relationship between C.S.R. and firm performance is U-shaped. The coefficients of ESG and ESG 2 on Tobin's Q in column (2) are -0.029 (t¼-3.296) and þ0.001 (t ¼ 3.694), respectively, both being significant at 1% level. However, the coefficients of ESG and ESG 2 on ROA in column (4) are -0.0006 (t¼-0.353) and 0.00001 (t ¼ 0.331), respectively, both being insignificant. Therefore, we conclude that C.S.R. has a U-shaped relation with Tobin's Q in Indian firms, supporting Hypothesis 1. However, C.S.R. is not related to such accounting measure as ROA. Furthermore, for the average level of E.S.G. at 19.126 (from Table 1), the total effect of ESG on Tobin's Q is -0.023 Â 19.126 þ 0.0003Â(19.126) 2 ¼ -0.330. Even for the maximum level of E.S.G. at 50 (from Table 1), the total effect is also negative, indicating that the overall effect of ESG on Tobin's Q is negative. Thus, we conclude that although Hypothesis 1 is supported, the overall effect of C.S.R. on firm performance is still negative even at high level of C.S.R.
In addition, Leverage has a significantly negative impact on Tobin's Q and ROA, suggesting that financial risk has a negative effect on firm performance. Policydummy has a significantly negative impact on Tobin's Q and ROA in G.M.M. estimation. All models are statistically significant and the adjusted R-squared is typically in the range 0.365-0.804.

Group effect on the relationship between C.S.R. and financial performance
In this section, we investigate the relationship between C.S.R. and firm performance of business group affiliated firms compared to that of stand-alone firms. Table 9  reports the results of a panel regression model and G.M.M. estimation with respect to the joint effect between C.S.R. and business group on firm performance. In columns (1) and (2) of Table 9, C.S.R. and C.S.R. squared show significantly negative and positive relationships with Tobin's Q, respectively, confirming our previous results in Table 8. Furthermore, we use an interaction term (E.S.G. Â Group dummy) to examine the role of business group on the relationship between C.S.R. and firm performance. Column (1) of Table 9 shows an insignificantly positive effect on the relationship between E.S.G. Â Group dummy and Tobin's Q, which is inconsistent with hypothesis 2. However, consistent with hypothesis 2, Column (2) of Table 9 shows the interaction terms with business group (E.S.G. Â Group dummy and E.S.G. 2 Â Group dummy) have significantly positive and negative coefficients, respectively. On the other hand, in columns (3) and (4) of Table 9, all the coefficients of C.S.R., C.S.R. squared and the interaction terms between business group and C.S.R. are statistically insignificant, except C.S.R. coefficient of column (3) that is significant at 10% level.

Conclusion
Prior empirical studies have examined the relationship between C.S.R. and firm performance, but the evidence presented is mixed. This paper explores the relationship between C.S.R. and financial performance in Indian firms listed on the N.S.E. and