Disclosure of Corporate Social Responsibility and Firm Performance: Evidence from India

Disclosure of CSR

Stakeholders’ perception about the company’s social responsible conduct determines CSR business returns (Greening & Turban, 2000; Peterson, 2004). Since corporate social initiatives are strategic approach to differentiate a company from its competitors (Vogel, 2005), it is utmost important for every company to disclose effectively their performances related to CSR (Du, Bhattacharya & Sen, 2010). An effective and transparent communication of such activities improves credibility of the company and reduces its reputational risk (Forehand & Grier, 2003). Thus, the extant literature reveals that large number of studies relating to social responsibility disclosure practices have been undertaken both in developed and developing economies. For instance, Guthrie and Parker (1990) studied the corporate social disclosure practice in Australia, the United Kingdom and the United States of America. Their findings revealed that the overall level of disclosure was 98 per cent in the USA, 85 per cent in the UK followed by 56 per cent in Australia. Similarly, Thom and Decoutere (2009) reported that such level of disclosure was 81 per cent in Belgium. In the case of leading companies from Fortune Global 500, Lungu, Carauabu and Dascelu (2011) found such extent of disclosure to be around 88 per cent followed by 78 per cent (Abbott & Monsen, 1979). On the contrary, the literature review from emerging economies revealed lower level of disclosure. For instance, in Bangladesh the level of CSR disclosure was found to be 41 per cent only (Belal, 2001; Imam, 2000) followed by 26 per cent in Malaysia and Singapore (Ahmed, Sulaiman & Siswantoro, 2003; Tsang, 1998) and around 46 per cent in India (Aggarwal, 2001; Murthy, 2008). Similarly, a very recent study by Bhatia and Chander (2014) reported that the overall level of social responsibility disclosure ranges from 8 per cent to 52 per cent in case of 30 Sensex companies in India. Belal (2001) evaluated social reporting practices in Bangladesh in order to explain why social disclosure is more prevalent. The author explained that the reason for high disclosures relating to employee category was probably due to the emphasis on workers’ welfare and strong labour policy of Bangladesh. Employees from developing countries are given more importance as perhaps they contribute more revenue for the company. Thus, in the case of developing nations, ‘human resource’ related information is disclosed more (Aggarwal, 2001; Andrew et al., 1989; Murthy, 2008).

As per the above literature on CSR reporting, apart from the other developing countries, Indian companies have lower level of disclosure as compared to the companies from developed countries. One plausible reason for low disclosure may be that the earlier researchers have used annual reports for computing disclosure score along with different methodologies. The present study uses only sustainability reports published by the Indian firms employing GRI framework as the base for computing disclosure score of CSR (including its components). Hence, the present study is diferent from the earlier studies on CSR in India.

CSR and Firm Performance

Apart from studies relating to the social responsibility disclosure practices, researchers have also focused on its association with firm performance. Association is broad term, which is simultaneously used for both the influence of independent variable on dependent variables as well as their relationships (Chen et al., 2015; Hackston & Milne, 1996; Mishra & Suar, 2010). In the present context, the term ‘association’ indicates the impact of explanatory variables on the outcome variable in the model.

While investigating the association between CSR and firm performance, a plethora of empirical studies used the Kinder, Lydenberg, Domini (KLD) index for measuring CSR. For instance, employing KLD index as a proxy to CSR, Waddock and Graves (1997), Berman et al. (1999), Akpinar et al., (2008), and Hillman and Keim (2001) found CSR to be positively associated with firm performance in developed countries like the USA. On the other hand, Garcia-Castro, Canela and Arino (2010) investigated the influence of social performance on the financial performance for 658 US firms from 1991 to 2005. By using ordinary least square (OLS) model in the pooled data, the study found that the association was positive and significant. Since the pooled OLS model fails to capture the intercorrelation between the variables over the years within the firm, the results of other panel data regression model are more suitable. However, the results of other panel data regression models showed an insignificant association.

The extant literature also indicates that social performance enables better firm performance due to the imperative that it attracts socially responsible consumer, alleviates the possible threat of regulation and improves firm’s reputation with customers (Baron, 2001; Jones, 1995; Lev, Petrovits & Radhakrishnan, 2008). Some proponents (Griffin & Mahon, 1997; Preston & O’Bannon, 1997; Soloman & Hansen, 1985) found that when firm makes investment in CSR, it results in higher financial return, which eventually leads to better firm image. Arguing in a similar vein, Clarkson (1995) concluded that a firm can enhance its performance by disclosing more information relating to CSR for satisfying stakeholders needs. Likewise, Porter and Van Der Linde (1995) and Roberts and Dowling (2002) observed that CSR helps in developing reputational advantage, which eventually enhances financial performance of the firm. Again, by employing meta-analysis, Orlitzky et al. (2003) reported simultaneous relationship between CSR and financial performance. Similarly, Choi, Kwak and Choe (2010) empirically investigated the impact of CSR on firm’s financial performance in case of 1,222 Korean firms. Their results indicated a significant positive influence of CSR on firm’s financial performance. The findings are consistent with the study of Uadiale and Fagbemi (2012) for Nigerian firms. On the contrary, Fombrun and Shanley (1990) reported an insignificant association between CSR and financial performance and concluded that it is not possible to control the large number of intervening variables. Similarly, there are some studies where the researchers failed to extricate any significant influence of CSR on firm performance (Becchetti et al., 2005; McWilliams & Siegel, 2000; Ullmann, 1985). Such association can also be negative when firm’s social responsibility involves costs, which are higher than the benefits derived there from. Empirically, Brammer et al. (2006) and Ho and Taylor (2007) found negative association between CSR and firm performance.

The impact of social responsibility on firm performance in case of Indian manufacturing firms was examined by Mishra and Suar (2010). Their findings indicated that CSR activities towards key stakeholders may improve the profitability of the Indian firms. Regarding CSR practices in Indian public sector companies, Singh and Ahuja (1983) analysed the content of 40 annual reports and found mixed results relating to the association between CSR and profitability. While analysing CSR practices during 1984–1985 to 1990–1991, in case of Steel Authority of India Limited, Batra (1996) observed inconsistencies in CSR practices.

The extant literature, thus, indicate that the researchers observed mixed results relating to the association between CSR and firm performance, although such association is found to be positive in a majority of the studies conducted in developed economies. However, only a few researchers found CSR to be positive and significantly associating with firm performance, but a large number of studies revealed an insignificant association, in developing economies. This difference may be resulting from differences in methodology adopted, study period examined and the wide variation in variables used to measure CSR.

According to Carrots and Sticks (2013), there exists a positive theoretical relationship between CSR and companies performance. Based on this relationship, the following alternative hypothesis is formulated for an empirical testing in the Indian context.

Regarding the influence of the components of CSR on firm performance, the empirical observation is scanty. Chen et al. (2015) investigated this issue and found positive relationship between the components of CSR and firm performance, which supports the theoretical association. We also expect that the components of CSR will have a positive impact on firm performance and hence, the following hypothesis is developed.

Source of Data, Study Period and Sample

For the purpose of this study, secondary data are collected from corporate sustainability reports and annual reports of non-financial Indian firms listed in National Stock Exchange (NSE) and Bombay Stock Exchange (BSE) for six years from 2008–2009 to 2013–2014. The sustainability report is used to measure the disclosure score of CSR and its components. On the other hand, annual report is used for measuring firm performance. The study period is limited to only six years because most of the Indian companies have started reporting on responsibility issues from 2008–2009 onwards (Carrots & Sticks, 2013).

In the present study, non-financial companies publishing sustainability reports are considered because financial companies do not comply with all the indicators specified in the GRI sustainability reporting framework. It is also worth noting that company’s non-financial activities can also be judged on the basis of their continuous reporting. Initially, all the listed Indian firms (both BSE and NSE) are considered. At the outset, we have identified the firms who have published sustainability or responsibility reports in their respective websites. Nevertheless, there are companies who are not publishing their sustainability report continuously and are primarily excluded from the sample. Finally, only 28 firms are found publishing their sustainability report continuously since the end of March 2009 in their respective websites. Thus, the final sample consists of 28 listed Indian companies for a period of six years from 2008–2009 to 2013–2014.

Measurement of Variables

In this study, dependent variable is market to book ratio (MBR), which is used as a measure of firm performance. MBR is computed by dividing the market value (MV) of common stock by book value (BV) of common stock. MV is the product of the number of outstanding equity share and share price at the financial year ending. Since investors’ expectation is determined by the market share, MBR is considered to be more appropriate than the accounting measure of firm performance to examine the influence of CSR (Margolis & Walsh, 2001).

The independent variables in this study are CSR and its components—HR information (including human rights and labour related information), product responsibility (PR) and society related (SO). The disclosure score of CSR and the components of CSR are computed employing content analysis technique based on appropriate GRI framework. The details of content analysis are presented in the next subsection.

Two control variables, research and development expenditure (R&D) as a percentage of total assets and firm size as natural log of sales revenue (LnS), are considered for the study. R&D is used as a proxy for innovation, which is assumed to play a vital role for enhancing firm performance. Likewise, firm size measured by sales revenue takes care of the growth aspect and market share of the firm.

Content Analysis Technique Used in This Study and Its Reliability

According to Guthrie, Petty and Yongvanich (2004), the methodology of content analysis is used to codify different types of information into pre-defined classes. After codification, quantitative scales are obtained to help analyse further (Hackston & Milne, 1996). This is a common technique used in social responsibility literature for analysing corporate disclosure practices (Abbott & Monsen, 1979; Hackston & Milne, 1996). While carrying out the technique of content analysis, it is of the utmost importance to identify a unit of analysis. In most of the previous studies, researchers have used words (Ng, 1985), sentences (Hackston & Milne, 1996), paragraphs (Guthrie et al., 2004) and portions of pages (Unerman, 2000) as units of analysis. However, these units of analysis are losing their relevance after the development of the GRI reporting framework. For measuring CSR, GRI provided a list of specified items under the different categories of CSR such as human rights, decent work and labour performance, product responsibility and society performance. These items are presently used as a unit of analysis (Burhan & Rahmanti, 2012; Chen et al., 2015; Hussain, 2015).

In the present study, content analysis is employed using GRI framework as a base to calculate the disclosure score of CSR and its components (human rights, labour performance, product responsibility and society performance). The reporting framework of GRI is found to be most appropriate for the study due to the imperative that large numbers of firms are adopting this framework for disclosing their non-financial performance. In addition, GRI based report provides an inclusive, clear and reliable measurement system for analysing the firm’s responsibility performance (Marimon et al., 2012). Thus, in the present study, either of the two versions of GRI framework, that is, G3 and G3.1 is used for content analysis basing on their applicability in the published report. The G3 version was published in 2006, which contains 40 items for CSR. The G3.1 framework came in 2011 with five additional items and thus, the total items have increased to 45.

In many of the previous studies, researchers have used binary coding system of content analysis (Patten, 2002; Purushothaman, Tower et al., 2000). Binary coding system detects the absence or presence of information in the published report. Thus, to calculate the disclosure score, binary coding system is used, that is, 1 if the specified item in the framework is presented in the published report or 0 if it is not presented in the report. After obtaining the item-wise score, the overall disclosure score for CSR and its three components are calculated by:

D i s c l o s u r e j = ∑ i = 0 n X i j m j

(1)

where m_j is the maximum expected score for each category (or overall), j is the company, i is the items and X_ij assumes the value of 1, if the item is disclosed and 0 otherwise.

For approving the quality of content analysis, reliability plays a very important role. There are different ways to improve the quality of content analysis (Guthrie et al., 2003). First, identifying, selecting and then defining the disclosure categories from the related literature is of the utmost importance. Second, a reliable coding tool with proper decision rules must be established. Last, there is a need to adopt uniform coding system for reliable content analysis (Potter & Levine-Donnerstein, 1999).

Sustainability report based on GRI is itself reliable for analysing the firm’s sustainability performance (Marimon et al., 2012). In order to measure CSR, the GRI report provides altogether 40 (in G3) and 45 (in G3.1) specified items for all the components of CSR. Moreover, according to the GRI guideline, only the validated contents can be disclosed in the sustainability report. Thus, the present study uses GRI-guided items as a unit of measurement along with a uniform coding system for enhancing the reliability.

Empirical Models

Homoscedasticity, that is, constant variance, is an important assumption of the classical linear regression model. The OLS estimation in the presence of heteroscedasticity may be unbiased and consistent, but cannot be efficient or best linear unbiased estimator (BLUE) (Gujarati & Sangeetha, 2008). Some plausible reasons for the heteroscedasticity are the existence of outliers in the data set and the skewed distribution of the one or more variables in the regression model. In Table 1, the distribution of the variables clearly demonstrates that the distribution of some variables is skewed and outliers are present in the data set. Further, the outcome of Park Test (Park, 1966) and Glejser Test (Glejser, 1969) indicates that heteroscedasticity is present in the data set (results are not reported here). Hence, in order to obtain estimators those are BLUE for the present data set, GLS method of estimation is employed in this study. Unlike OLS, which minimises equal-weighted residual sum of squares (RSS), GLS method minimises a weighted sum of residuals with w_i = 1/σ² acting as weight.

OPEN IN VIEWER

Table 1.

Descriptive Statistics

Variables	Mean	5% Trimmed Mean	Median	Skewness
HR	0.839	0.846	0.873	–.738
SO	0.886	0.893	0.875	–.582
PR	0.832	0.839	0.888	–.620
CSR	0.849	0.851	0.865	–1.13
MBR	7.033	5.919	3.515	2.15
LnS	9.475	8.656	9.946	–2.46
R&D	1.233	0.961	0.239	1.953

Source: Computed by the authors.

Further, to find out the appropriate panel data regression model for the present data set, two widely used tests are conducted: Breusch–Pagan test and Hausman test. The test results advocate in favour of random effects model (test results are shown in the respective tables). The study, thus, uses random GLS method to estimate the coefficients. The following regression models are employed here to test the hypotheses:

M B R i , t = β 1 + β 2 C S R i , t + β 3 L n S i , t + β 4 R & D i , t + w i , t ( M o d e l 1 ) M B R i , t = β 1 + β 2 H R i , t + β 3 S O i , t + β 4 P R i , t + β 5 L n S i , t + β 6 R & D i , t + w i , t ( M o d e l 2 )

(2)

where w_i,t is the composite error term of the two different components: ε_i, is the specific individual or cross-section error component and u_i,t is the cross-section and time series combined error component. Where, ε_i ~ N(0, σ ε 2 ) and u_i,t ~ N(0, σ u 2 ).

Disclosure Trend

To study the disclosure trend of the overall CSR as well as components of CSR (HR, SO and PR) for the selected Indian firms, year-wise average disclosure score based on GRI framework has been computed. The outcomes are shown in Figure 1 and Figure 2 respectively for the overall CSR and components-wise disclosure scores. Figure 1 shows that there is an increasing trend in disclosing the overall CSR for the selected firms during 2008–2009 to 2013–2014. For instance, the average disclosure score has increased from around 78 per cent in 2008–2009 to 91 per cent in 2013–2014. However, during 2012–2013 the disclosure score has remained almost same, may be due to the adoption of GRI 3.1 framework with five additional social-related items. The observed increasing trend of disclosing social-related information indicates that Indian firms are paying more attention in satisfying the needs of the stakeholders and are becoming sensitive towards sustainable development.

OPEN IN VIEWER

Figure 1.

Source: Graph plotted by the authors.

The year-wise average disclosure scores for the three components of CSR, as shown in Figure 2, also reveal an increasing trend over the years. To be specific, the average disclosure level of SO (or community related) information and HR information is found to be increasing throughout the study period as compared to PR related information. Moreover, the average disclosure score of product responsibility (PR) is found to be slightly lower than the other two components from 2009 to 2011, thereby it improves slightly than the HR information. During the study period, the average disclosure score of SO (or community related) information is found to be the highest (nearly 89 per cent) followed by HR (84 per cent) and PR (83 per cent). The observed higher level of disclosure in respect of community related information indicates that Indian firms are focusing more on activities related to social upliftment and communicates more actively to their stakeholders. However, on an average, the observed disclosure scores for all the three components are quite satisfactory. The study found that the CSR disclosure score of Indian firms are considerably higher as compared to the earlier studies (Aggarwal, 2001; Bhatia & Chander, 2014; Ghosh, 2002; Murthy, 2008) may be due to the different methodologies adopted by others.

OPEN IN VIEWER

Figure 2.

Source: Graph plotted by the authors.

Descriptive Statistics

Table 1 shows the descriptive statistics relating to mean, 5 per cent trimmed mean, median and skewness of all the variables included in the regression models. The main aim here is to get some idea about the distribution of the variables and the presence of outliers in the data set. The near equality of mean, median and 5 per cent trimmed mean indicates that the distribution of the variables is not far away from symmetry. A look into the table reveals that the observed values of mean, 5 per cent trimmed mean and median, in case of HR, SO, PR and CSR are not significantly different. As a result, the values of skewness are also not very high. On the other hand, for MBR, LnS and R&D the values of mean, 5 per cent trimmed mean and median are considerably different from each other. For instance, the mean of MBR is 7.033, while the median value is 3.515. Likewise, the mean and median values of R&D are 1.233 and 0.239, respectively, which indicates that outliers are present in the data set. The observed values of skewness for these three variables also demonstrate that the distribution is far away from symmetry. As discussed in the ‘data and methodology’ section, the outcome of Park Test and Glejser Test also indicates that heteroscedasticity is present in the data set. The observed values of descriptive statistics, as shown in Table 1, also exhibit the same. Hence, the use of GLS method instead of OLS is justified. However, the observed mean values of the overall CSR and the three components show that the average disclosure score is quite satisfactory for the Indian firms.

Regression Results

Table 2 demonstrates the regression results of model 1. Model 1 is basically employed to investigate the overall impact of CSR on firm performance. The outcomes reveal that the coefficient estimate of CSR is positive and significant at 5 per cent level after controlling the influence of R&D and LnS. This implies that disclosure of CSR positively influences firm performance. The results, thus, support the theoretical positive relationship between the disclosure of CSR and firm performance. Indeed, in developed economies, many earlier researchers have observed positive influence of CSR on the performance of the firm (Akpinar et al., 2008; Preston & O’Bannon, 1997; Roberts & Dowling, 2002). In Indian context, however, researchers have observed mixed results (Batra, 1996; Singh & Ahuja, 1983). The outcome of the present study is consistent with the studies carried out in developed economies like the USA. Again, the survey report of the Carrots and Sticks (2013) indicates that in India, there is an increasing trend in disclosing non-financial information in the recent times. When a significant number of firms disclose the SO information rigorously over the years, stakeholders also try to evaluate firm performance based on these factors and the positive impact of CSR on firms’ market value is expected to be more. The observed result is, thus, sufficient to accept the first alternative hypothesis (H₁) of this study. Regarding the influence of the two control variables on firm performance, positive impact is found but the results are not significant.

OPEN IN VIEWER

Table 2.

Regression Results: Impact of CSR on Firm Performance

Variables	Unstd. Coefficient	Std. Coefficient	t-Statistic	p-Value
Constant	0.944	–	0.327	0.743
CSR	5.023	0.067	2.516	0.012**
R&D	0.183	0.043	0.743	0.458
LnS	0.166	0.034	0.776	0.438

Source: Computed by the authors.

Notes: Dependent variable: MBR; ** indicates significant at 5% level; model used: random GLS model.

Panel data test results:

Breusch–Pagan test: Chi-square = 354.025 with p-value = 0.000; Hausman test: Chi-square = 2.26512 with p-value = 0.519236.

The influence of the three components of CSR (HR, SO and PR) on the performance of the firm is shown in Table 3. In model 2, all the three components along with the two control variables are considered as explanatory variables. We have checked the multicollinearity problem through correlation matrix. The results indicate that these values are extremely correlated with each other. For example, the correlation coefficient between PR and HR is found to be 0.794 (significant at 1 per cent level by two-tailed test). Again, the correlation coefficient between PR and SO is found to be 0.734 (significant at 1 per cent level). The observed high degree of correlation clearly indicates that the inclusion of all the three variables in one model may produce biased result due to severity of multicollinearity. Hence, we regress each component separately. Again, in model 1, it is observed that the impacts of the two control variables are insignificant. Hence, these variables are excluded from model 2 for specifying the final model. The outcomes are shown in Table 3.

OPEN IN VIEWER

Table 3.

Regression Results: Impact of the Components of CSR on Firm Performance

Variables	Unstd. Coefficient	Std. Coefficient	t-Statistic	p-Value
HR	5.990	0.735	3.199	0.001***
SO	3.541	0.371	1.854	0.065*
PR	3.751	0.564	2.585	0.011**

Source: Computed by authors.

Notes: Dependent variable: MBR; * indicates significant at 10% level; model used: random GLS model. Instead of using all the variables in one equation as described in model 2, here separate regression for each of the three independent variables has been run. Hence, for each run, there is a constant term. However, the values of constant are not shown here.

Table 3 indicates that the coefficients of all the three components are significant and positive. This entails that not only the overall influence of CSR on performance of the firm is positive but the CSR components also individually enhance the firm’s market value. Among the three components, HR (0.735) is the most dominating variable as measured by standardised coefficient in enhancing firm performance, followed by PR (0.564) and SO (0.371). The results, thus, strongly advocate in favour of H₂, that is, the CSR components positively influence the performance of the firm.