5.1. Hypothesis test
Due to a heteroscedasticity problem, the GLS (Generalized Least Squares) method is used to test the hypotheses. The regression results (Table 5) affirm that the CSR variable affects positively and insignificantly Tobin’s Q (coeff= -1.266; sig = 0.235) and ROA (coeff = 0.128; sign = 0.257). Hence, hypothesis 1 is rejected. These findings notice that the CSR commitment level decreases financial performance. These results do not seem to conform with our expectations, nor with those documented in the already published studies highlighting that CSR engagement positively impacts financial performance (Freeman, 1984; McGuire et al., 1988; Graves and Waddock, 1994; Karagiorgos, 2010; Feng et al., 2017 and Maqbooln and Zameer, 2017; Aftab et al., 2024). However, these results corroborate the one found by Ullmann, 1985; Makni et al., 2009; Aras et al., 2010.
Concerning CG mechanisms, findings, as reported in Table 5 panel B, show that the governance score affects negatively and insignificantly financial performance, as measured through the Tobin’s Q (-0.0001, sig = 0.973), but it affects significantly ROA (-0.0004; sig = 0,092). Consequently, our second hypothesis is partially rejected. These findings are consistent with those released by Cheng (2008), who argues that the more the board size increases, the less likely are firms to take extreme decisions, which leads, in turn, to lower variations in corporate performance. Yet, such results are not congruent with our expectations, nor with the findings of already published studies (Gompers et al., 2003; Brown and Caylor, 2006; O'Connor, 2012; Black et al., 2006; Akbar et al., 2016; Pillai et al., 2017; Maranho and Leal, 2018). These findings may be explained by the aggregate measure of CG mechanisms. So, we could conclude that some mechanisms are not effective in firms. Actually, companies were expected to benefit noticeably from the advantages associated to CG mechanisms, mainly, close and effective monitoring, reduced agency costs and easier access to financial resources.
Our second objective consists in studying the joint effect of CG and CSR on financial performance. As illustrated in Table 6, we notice a positive and significant relationship between the interaction term and performance measured by Tobin’s Q (0.576, p = 0.062) and ROA (0.061, p = 0.004). In fact, the stronger the company applies governance practices are, the more willing to implement CSR practices it is, and consequently, the more effectively it will perform. Thus, our hypothesis 3 is accepted. Our achieved results corroborate the findings attained by previous studies (Khanchel El Mehdi, 2013; Ntim and Soobaroyen, 2013; Pillai et al., 2017; Sahut et al. 2018), stressing that the interaction between CG and CSR helps in further intensifying and accentuating their effect on financial performance. Thus, our study supports already published results maintaining that CSR practices improve the financial performance of the firm through the moderating effect of governance. In fact, the adoption of these practices increases the company’s competitive advantage on the market, while improving its management process. Therefore, CSR activities can have a positive effect on financial performance due to an effective or good CG allowing the reduction of conflicts of interests. Yet, in poorly governed firms, a greenwashing behaviour may be adopted and then poor CSR practices leading to intensify the stakeholders’ conflicts (Jo and Harjoto, 2011). Thus, CSR can have a negative effect on financial performance.
By adopting CSR and good governance practices, the company could promote its reputation and brand image, thereby, attracting new customers. Additionally, integrating CSR and governance in activities and decisions allows firms to establish stronger ties and a permanent co-integration among the different stakeholders, resolve conflicts of interest, minimize agency costs and satisfy the stakeholders' expectations. On the one hand, the interaction between governance and CSR provides the company a greater transparency, respect for shareholder rights, and effective control. On the other hand, an effective system of governance interactively coupled with CSR practices is liable to curb the opportunistic behavior of managers. Particularly, this could be achieved by helping reduce the problems of liquidity and information asymmetry, thereby, inhibiting the manipulation of accounts. It is certain that a company concerned with social, environmental and governance standards, and striving to satisfy them, enjoys greater credibility in the eyes of the population and outsiders, alike, and is therefore met with greater trust from their part. Such a procedure coincides well with the hypothesis assuming that the major expected effect of CSR and good governance practices lies in a considerable attenuation of risk perceived by investors, while maintaining the promotion and boosting of the company’s financial performance.
As explained by Ntim and Soobaroyen (2013), our findings are consistent with the neo-institutional framework that emphasis on the efficiency and legitimation effects of CSR engagement. So, when the CG mechanisms are effective, managers are more likely to invest in CSR activities that can help legitimize corporate actions by aligning with higher-order values and enhancing corporate efficiency. Thus, good CG can be a positive catalyst on the CSR-financial performance relationship by helping to reduce interests’ conflicts among the various stakeholders.
As regards the control variables, results reveal that the indebtedness level, measured through the ratio (total debt / total assets), and in conformity with the sign expected, has a negative and significant effect on financial performance, measured with Tobin’s Q, ROA, at the 1% threshold. As for liquidity, the correlation proves to be positive and significant, with the recorded values of Tobin’s Q (0.468, p = 0.001) and ROA (0.025, p = 0.069). Thus, the higher the liquidity level, the better the company performance. In addition, a positive and significant relationship has been perceived to exist between firm size and financial performance. With respect to the growth opportunity, the relevant coefficients are positive and statistically significant, thereby, corroborating the findings published by Margaritis and Psillaki (2007), Zeitun and Tian (2007) and Nunes et al. (2009). In fact, company growth is generally recognized as a positive signal by shareholders, whose main objective is maximizing wealth,
by pressurizing managers to promote company growth and act more efficiently to achieve a rather high corporate performance.
Table 5
Direct effects Panel A : CSR and financial performance
Variables | Model 1 : Tobin’s Q | Model 1 : ROA |
Coefficient | p-value | Coefficient | p-value |
CSR | 1.266 | 0.235 | 0.128 | 0.257 |
LEV | -0.002 | 0.306 | -0.00028 | 0.297 |
QR | 0.134 | 0.421 | -0.015 | 0.368 |
GO | 1.135 | 0.069* | 0.074 | 0.260 |
Size | -0.399 | 0.000*** | -0.029 | 0.000*** |
year fixed effects | Yes | Yes |
Constant | 6.379 | 0.000*** | 0.250 | 0.021*** |
Wald chi2 | 57.64 | 19.74 |
Prob > chi2 | 0.000*** | 0.0014*** |
Panel B: CG and financial performance |
Variables | Model 1 : Tobin’s Q | Model 1 : ROA |
Coefficient | p-value | Coefficient | p-value |
CG | -0.0001 | 0.973 | -0.0004 | 0.092* |
LEV | -0.002 | 0.377 | -0.0027 | 0.193 |
Liquidity | 0.120 | 0.472 | -0.006 | 0.668 |
GO | 1.107 | 0.077* | 0.118 | 0.039 |
Size | -0.424 | 0.000*** | -0.221 | 0.000*** |
year fixed effects | Yes | Yes |
Constant | 5.494 | 0.000*** | 0.314 | 0.000*** |
Wald chi2 | 55.99 | 30.67 |
Prob > chi2 | 0.000*** | 0.0000*** |
Note: * ** significant at the level 1%; ** significant at the level 5%; * significant at the level 10% |
Table 6
Moderating effects of governance
Variables | Model 2 :Tobin’s Q | Model 2 : ROA |
Coefficient | p-value | Coefficient | p-value |
CSR | -0.59 | 0.443 | -0.933 | 0.101 |
GOV | -0.005 | 0.177 | -0.003 | 0.083* |
CSR*GOV | 0.576 | 0.062* | 0.061 | 0.004*** |
LEV | -0.016 | 0.000*** | -0.0009 | 0.000*** |
Liquidity | 0.468 | 0.001*** | 0.025 | 0.069* |
GO | 8.605 | 0.000*** | 0.335 | 0.000*** |
Size | -0.237 | 0.000*** | -0.0023 | 0.000*** |
year fixed effects | Yes | Yes | Yes | Yes |
C | 8.437 | 0.001*** | 1.133 | 0.000*** |
Wald chi2 | 344.19 | 248.94 |
Prob > chi2 | 0.000*** | 0.000*** |
Note: * ** significant at the level 1%; ** significant at the level 5%; * significant at the level 10% |
5.2. Additional analysis
For a more effective assessment of our results’ robustness, we consider implementing a number of sensitivity tests. The first robustness test is administered by applying the ROE as an alternative measure of financial performance. In fact, the ROE variable helps in determining the firm’s efficiency level on using the shareholders’ funds as a means to generate profits. The results (Table 7) highlight clearly that the effect of CSR engagement on financial performance is negative and significant confirming the first results. Concerning the effect of governance on financial performance, it proves to remain negative and significant. However, the joint effect of CSR and governance improves financial performance (0.475, p = 0.044).
Table 7
Variables | Model 2 : ROE |
| Coefficient | p-value |
CSR | -0.396 | 0.047** |
GOV | -0.404 | 0.054** |
CSR*GOV | 0.475 | 0.044** |
LEV | -0.019 | 0.069* |
Liquidity | 0.564 | 0.035** |
GO | 1.220 | 0.000*** |
Size | -0.0015 | 0.074* |
C | 0.638 | 0.144 |
Year fixed effects | Yes |
Wald chi2 | 148.25 |
Prob > chi2 | 0.000*** |
Note: * ** significant at the level 1%; ** significant at the level 5%; * significant at the level 10% |
Second, we replace CG index by board characteristics. As Table 8 indicates, the interactions of the variables CSR*board size CSR*mandate respective coefficients are negative and statistically significant, suggesting that the board and relating expertise prove to negatively affect the relationship between CSR and firm financial performance. As regards the interaction of the variables CSR*expertise and CSR*duality, a statistically positive relationship between the interaction terms and firm performance exists. This result highlights well that when firms with accumulated director-CEO positions and enjoying a great deal of experience engage in CSR activities, financial performance would certainly decrease. Contrarily, if directors’ expertise and independence are associated with CSR activities, it raise financial performance.
Table 8
Moderating effects of CSR and board characteristics
Variables | Model 1 : Tobin’s Q | Model 1: ROA |
Coefficient | p-value | Coefficient | p-value |
CSR*board size | -0.032 | 0.085** | 0.002 | 0.230 |
CSR*mandate | -0.197 | 0.161 | 0.007 | 0.671 |
CSR*duality | 0.546 | 0.000*** | 0.029 | 0.094** |
CSR*expertise | 0.461 | 0.003*** | 0.039 | 0.028** |
CSR*independence | 0.540 | 0.025** | 0.027 | 0.339 |
LEV | -0.004 | 0.067** | -0.0005 | 0.093** |
Liquidity | 0.394 | 0.027** | -0.010 | 0.623 |
GO | 0.68 | 0.215 | 0.065 | 0.311 |
Size | -0.421 | 0.000*** | -0.036 | 0.000*** |
Year fixed effects | Yes | Yes | Yes | Yes |
C | 5.180 | 0.000 | 0.354 | 0.000 |
Wald chi2 | 180.59 | 43.82 |
Prob > chi2 | 0.0000 | 0.0000 |
Note: * ** significant at the level 1%; ** significant at the level 5%; * significant at the level 10% |
Third, to account for any potential endogeneities and simultaneity problems that may arise due to the presence of a lagged CSR-financial performance association, we proceed with estimating a lagged CSR-financial performance structure and two-stage least squares (2SLS), by incorporating the previous year’s ROA into the model.
To address simultaneity problems that may arise due to the presence of a lagged CSR-financial performance connection, and following McGuire et al. (1988), we consider re-estimating model (1) in the form of a lagged structure using a dynamic panel specified as:
ROAit = α0 + α1 ROAit−1 + α2CSRit + α3Govit + α4CSRit*Govit + α5LEVit + α6QRit + α7GOit + α8Sizeit + εit
Indeed, it is necessary to consider the validity of the instruments using GMM analysis. Then, the condition of exogeneity of the instruments (delayed variables) can be checked by means of the Hansen test. Table 9 shows that Hansen’s over-identification test does not reject the null hypothesis of the validity of the lagged and differential variables as instruments. The level of significance and the value of the coefficients of the delayed variable of ROA (Lag ROA) provide anew justification for the dynamic specification of the model and confirm the need to include these effects. Indeed, ROA depends strongly on their retarded variables and socially responsible engagement.
The results presented in Table 9highlight the presence of dependence or temporal dynamics. Thus, the ROA at period t-1 is strongly and positively correlated with that in (t) (p-value = 0.019). They also confirm the previous results stipulating that the CSR and governance are negatively correlated with ROA. Thus, we mention the existence of a synergy of complementarities, given the dynamic temporal nature between ROA and CSR engagement and governance.
Table 9 : Dynamic estimation of panel data | |
Variables | Coefficient | | P value | |
ROA L1. | 4.649 | | 0.019** | |
CSR | -0.6011 | | 0.041** | |
GOV | -0.0120 | | 0.034** | |
CSR*GOV | 0.0126 | | 0.013** | |
LEV | -0.0001 | | 0.093* | |
Liquidity | 0.1540 | | 0.004*** | |
GO | 0.3060 | | 0.026** | |
Size | -0.0521 | | 0.036** | |
year fixed effects | Yes | | Yes | |
Constant | 0.204 | | 0.011 | |
Saran test (Chi-square, p-value) 415.10 (p = 0.000) |
Hansen test (Chi-square, p-value) 2.59 (p = 0.4594) |
Note: * ** significant at the level 1%; ** significant at the level 5%; * significant at the level 10% |