Corporate Social Responsibility as a Driver of Financial Performance: An Exploration of South African Companies

Abstract

This study investigates the relationship between corporate social responsibility performance and financial performance among firms listed on the Johannesburg Stock Exchange in South Africa. Utilising a multi-metric approach, the research incorporates corporate social responsibility scores; environmental, social, and governance ratings; and the social pillar score to provide a comprehensive analysis. Data from 104 companies with 624 observations from 2017 to 2022 was analysed. This quantitative study employs a Generalised Least Squares estimation, and the findings reveal a significant positive correlation between corporate social responsibility performance and several key financial metrics, including return on assets, earnings per share, market value added, and Tobin’s Q ratio. The results suggest that companies prioritising corporate social responsibility initiatives are likely to experience improved financial outcomes. Furthermore, the study examines the influence of board characteristics on financial performance, identifying a positive effect of gender diversity and negative impacts from board independence and meeting frequency. Overall, this research contributes to the literature on corporate social responsibility and financial performance by highlighting the importance of corporate social responsibility in driving sustainable business practices and enhancing firm performance within the context of an emerging economy. The findings underscore the need for firms to integrate corporate social responsibility into their strategies to promote long-term success while addressing societal challenges.

1. Introduction

Business analysts forecast companies’ future growth and performance by utilising corporate social responsibility (CSR) ratings to inform their investment recommendations (Ioannou & Serafeim, 2015). Boccia and Sarnacchiaro (2018) assert that these factors necessitate that organisations engage in socially responsible practices, as sustainable, ethical, and responsible actions can yield the highest value for any enterprise. This means that for businesses to operate efficiently in recent decades, they need to incorporate CSR activities as part of their operations. In recent years, rapid transformation and significant global challenges have necessitated companies to uphold a credible image (Barauskaite & Streimikiene, 2021). Barauskaite and Streimikiene (2021) further state that achieving this requires a careful equilibrium among financial advantages for the company, the welfare of the public, and the safeguarding of the environment. Companies striving to achieve this equilibrium progressively implement sustainable and socially responsible practices. The implementation of sustainable and socially responsible policies aimed at tackling social and environmental challenges has emerged as a significant focus in both academic research and practical applications (Ghanbarpour et al., 2024).

Corporate sustainability has emerged as a critical concern in the 21st century, with CSR acting as a significant catalyst for both societal advancement and corporate efficacy (Fosu et al., 2024). Companies are integral to fostering the progress of society and local communities (Fosu et al., 2024). CSR has become a crucial element of modern business strategies, shaping the conduct of companies across the globe (Knudsen & Moon, 2022). A well-structured CSR strategy is thought to contribute to substantial decreases in various business risks, encompassing those related to stakeholders, regulatory compliance, and potential sanctions within the marketplace (Maya, 2024). Therefore, to attain sustainable corporate growth, companies must extend their accountability beyond mere profit generation to encompass environmental stewardship and the fulfilment of associated social responsibilities (Simmou et al., 2023).

Asiedu et al. (2020) argues that even companies driven by profit motives are compelled to shift their focus from solely economic goals to incorporate societal considerations. It is essential to examine the extent to which CSR may conflict with the objective of maximising profits (Ang et al., 2022). This raises concerns about the potential trade-offs between CSR initiatives and financial performance (FP), highlighting the importance of examining the relationship between CSR and FP, as understanding these dynamics is crucial for businesses seeking to adopt sustainable practices while remaining financially viable. The inquiry into the impact of CSR on a company’s FP has been a subject of scholarly debate since the early discussions surrounding CSR (Licandro et al., 2024). Numerous studies have investigated this relationship, yet a definitive agreement on the nature of the association between CSR and FP remains unresolved (An & Albattat, 2023; Li et al., 2023; Saeed et al., 2023). As such, the aim of this study is to fill the existing gap in conflicting evidence by empirically examining the impact of CSR performance on FP in the context of an emerging economy, an area that has been relatively underexplored in the literature (Chininga et al., 2024).

McKeever (2024) states that South Africa is widely recognised as the most unequal nation globally. According to Valodia (2023), South Africa exhibited the highest level of income inequality globally, as evidenced by a Gini coefficient estimated at approximately 0.67. Regardless of whether one assesses income, wealth, or quality of life, a significant proportion of the nation’s economic advantages are held by a limited segment of the population (McKeever, 2024). Crime is also regarded as a significant challenge in post-apartheid South Africa (Obagbuwa & Abidoye, 2021). Unemployment is another contemporary issue in South Africa. The first quarter of 2024 witnessed a significant increase in unemployment among South Africans, with the total number of jobless individuals rising by 330,000, reaching an unprecedented 8.2 million (Scribante, 2024). Events in the last decade that indicate how South Africa is unequal include the Fees Must Fall (FMF) movement and the July 2021 unrest. The FMF movement protests commenced in 2015 (Greeff et al., 2021). According to Greeff et al. (2021), these protests arose from a unified yet varied set of motivations, as university students expressed their discontent regarding elevated tuition fees, issues related to university governance, and insufficient financial support. Universities South Africa projected that the financial repercussions resulting from the FMF protests over an eighteen-month period amounted to approximately ZAR 700 to ZAR 800 million for restoration efforts (Staff, 2016). The protests ended in December 2017 following the government’s introduction of a sliding-scale payment policy designed to reduce the financial strain on underprivileged students (Griffiths, 2019). The social unrest in July 2021 represented a significant episode of violence, emerging amidst the third wave of the COVID-19 pandemic following the “Free Zuma” demonstrations (Phungula, 2024). The social unrest was fuelled by widespread frustration over economic hardship, government corruption, and rising inequality, leading to protests that escalated into riots and looting across many regions. The July 2021 unrest resulted in the death of 337 people (Maeko, 2021). This unrest left two million people unemployed and a ZAR 50 million loss to the South African economy (Erasmus, 2022). The prevalence of these social ills in South Africa is not only concerning. They also suggest that the corporate sector can play a role in addressing them via CSR activities while striving to achieve profitability and sustainable value for all the stakeholders (Serfontein-Jordaan & Dlungwane, 2022).

The impetus for this study resides in the pressing need to thoroughly examine CSR in the South African context, particularly for companies listed on the Johannesburg Stock Exchange (JSE). As highlighted by Nguyen et al. (2022), existing research often fixates on a singular metric of CSR, thereby offering a limited perspective on the relationship between CSR and FP. This study seeks to bridge this knowledge gap by employing a multi-metric approach to CSR. Investigating the correlation between diverse CSR performance metrics and FP within the unique regulatory frameworks of the financial and nonfinancial sectors contributes to a more nuanced understanding of how CSR practices impact companies’ FP (Matemane et al., 2024). Prior studies focusing on CSR in South Africa have primarily emphasised qualitative indicators, illuminating crucial social and environmental challenges (Chininga et al., 2024). In contrast, this research extends beyond qualitative assessments by integrating CSR scores; environmental, social, and governance (ESG) ratings; and social pillar scores, thus spotlighting the quantitative dimensions of CSR initiatives. This approach enables a robust measurement of the effectiveness of such efforts. Methodologically, this study employs the Generalised Least Squares (GLS) estimation technique to account for heteroscedasticity and autocorrelation, thereby enhancing statistical efficiency and reducing the likelihood of erroneous inferences. This study is the first to employ the GLS method in this context. Overall, this research aims to provide valuable insights into the multifaceted relationship between CSR performance and financial outcomes, ultimately contributing to both academic discourse and practical policy considerations.

This article is structured as follows: Section 1 introduces the research topic and its significance. Section 2 reviews the relevant literature on CSR performance and its impact on financial performance with an overview of the applicable theoretical frameworks. Section 3 details the empirical strategy employed in this study. Section 4 presents the findings and discusses their implications. Finally, Section 5 offers conclusions and suggestions for future research.

2. Literature Review and Hypothesis Development

2.1. Theoretical Framework

Nguyen et al. (2022) are of the view that a singular theoretical framework is insufficient to comprehensively account for CSR due to the inherent complexity associated with CSR activities. The relationship between CSR and financial performance is intricate, making it challenging to identify a singular theoretical framework that adequately captures this connection (Homayoun et al., 2023). Two opposing theories are often used to frame the relationship between CSR and firm performance, namely, the shareholder theory and the stakeholder theory (Zeribi-Benslimane & Boussoura, 2007).

2.1.1. Shareholder Theory

The shareholder theory, as pioneered by Friedman (1970), is rooted in Neoclassical economics: CSR has a detrimental effect on a company’s financial performance. Friedman (1970) asserts that the sole obligation of a business is to maximise its profits. In the South African context, the shareholder theory, as articulated by Friedman (1970), faces significant challenges and limitations that question its applicability. However, the study conducted by Sekhon and Kathuria (2019) supports this theory. While the traditional premise emphasises profit maximisation as the primary goal of corporations, this perspective may overlook the complex socio-economic realities and the historical injustices embedded in South Africa’s economic landscape.

Firstly, the aftermath of apartheid has left deep-rooted inequalities in ownership and control of businesses. As noted by Kantor (1998), prior to the advent of a democratic system, major companies were predominantly owned by a select few groups, often in the form of family-run conglomerates. The Black Economic Empowerment (BEE) program, as outlined by Sunday and Kwenda (2021), seeks to address these imbalances by promoting greater racial diversity in the ownership and management structures of companies listed on the JSE. In this scenario, the shareholder theory’s singular focus on profit maximisation could potentially perpetuate the existing inequalities and social injustices, ultimately undermining the long-term stability of the South African economy.

Moreover, the shifting expectations of stakeholders in a rapidly evolving South African marketplace complicate the shareholder-centric approach. Consumers, investors, and communities are increasingly prioritising ethical behaviour, social responsibility, and sustainability. As Lin (2024) argues, CSR initiatives can sometimes be seen as a necessary investment rather than a deviation from profit generation. In South Africa, where issues such as poverty, unemployment, and social cohesion are paramount, adopting a broader stakeholder model that incorporates social and environmental considerations may be essential for achieving sustained financial performance. This broader view encourages companies to build trust and rapport with various stakeholders, which is crucial in a society grappling with the legacies of its past.

Furthermore, in an era where legislative frameworks and corporate governance standards are evolving to promote greater social accountability, adherence to the shareholder theory could lead companies to overlook important regulatory and reputational risks. The need for companies to align their operational practices with national initiatives like BEE accentuates the importance of being responsive to the social and economic contexts in which they operate. Ignoring the imperative to integrate social responsibility into core business strategies may not only pose ethical dilemmas but also jeopardise firms’ market positioning and competitiveness in a socially conscious consumer landscape.

In conclusion, while the shareholder theory provides a foundational understanding of corporate objectives centred on profit maximisation, its applicability in the South African context is limited. The unique socio-economic challenges, coupled with evolving stakeholder expectations and regulatory frameworks, necessitate a more nuanced approach to corporate governance, one that recognises the importance of social responsibility and equity in fostering sustainable business practices. Thus, shifting the focus from singular profit generation to a more integrated model of stakeholder engagement may be foundational for the long-term viability of companies in South Africa.

2.1.2. Stakeholder Theory

Companies influence their external environments by engaging with stakeholders to create conditions advantageous for their operations, which has led to the development of the stakeholder theory (Awa et al., 2024). This theory was introduced by Freeman (1984), who asserts that organisations hold obligations to all their stakeholders. This may result in a sustained beneficial effect on the company’s performance (Aftab et al., 2024). The stakeholder theory posits that CSR positively influences FP by exceeding mere compliance with CSR standards and actively engaging with environmental and social challenges (Abid, 2023). This theory supports the positive relationship between CSR and FP. This is supported by Bristy et al. (2021) whose study revealed that engaging in CSR initiatives does not harm FP. In conclusion, the stakeholder theory, rooted in the concept of CSR, provides a framework that supports the notion that companies’ active engagement with stakeholders and their environmental and social responsibilities can lead to improved FP and sustained competitive advantage.

The stakeholder theory is particularly applicable in the South African context due to the country’s diverse socio-economic landscape, limited access to public services and the legacy of apartheid, which has created a complex web of stakeholder relationships (Hlongwane, 2022). The King IV Report on Corporate Governance for South Africa promotes a stakeholder-inclusive framework, wherein the governing body considers the legitimate and reasonable needs, interests, and expectations of all significant stakeholders while fulfilling its responsibilities, ultimately aiming to serve the long-term interests of the company (IoD, 2016). Failure to adhere to these requirements may result in a determination that they have violated their fiduciary responsibilities (Sartorius & Botha, 2008). This implies that the companies should consider the interests of all legitimate stakeholders and not only shareholders.

In response to the severe impacts of socio-economic issues, the private sector has recently collaborated with the government to accelerate the execution of the Economic Reconstruction and Recovery Plan (ERRP) (Hlongwane, 2022). As a result, numerous companies in South Africa are actively engaging in effective socio-economic development initiatives, recognising that their success is increasingly intertwined with the well-being of various stakeholders, including employees, suppliers, communities, and the environment (Hlongwane, 2022). This interconnectedness underscores the urgent need for businesses to adopt a more inclusive approach to corporate governance and decision making, especially given the pressing challenges of inequality and unemployment that the nation faces. By addressing these issues collaboratively, the private sector not only contributes to national recovery efforts but also enhances its long-term sustainability and success.

2.1.3. The Relationship Between CSR and ESG

To effectively address major global issues, businesses must innovate and operate sustainably by equally considering economic, social, and environmental factors, necessitating a new approach to reporting that includes ESG data (Kaźmierczak, 2022). While CSR has historically involved voluntary efforts to evaluate a company’s impact on society and stakeholders, often without direct links to measurement or regulatory oversight (Passas, 2024), ESG represents a framework utilised by investors and stakeholders to assess sustainability performance through measurable indicators of environmental, social, and governance risks and opportunities (Passas, 2024; Kaźmierczak, 2022). This shift from a CSR-centric model to ESG signifies a growing demand for transparency and structured reporting, driven by institutional investors and regulatory authorities (Passas, 2024; Rau & Yu, 2023).

CSR is frequently regarded as the essential groundwork for establishing ESG criteria. It encompasses a wide array of voluntary actions and obligations organisations undertake to ensure ethical operations and make a positive impact on society and the environment, including social programmes, environmental conservation, and fair treatment of employees and stakeholders (Kocmanová & Šimberová, 2014). Firms proactively participating in CSR by adopting sustainable practices often observe an improvement in their ESG ratings (Garcia & Orsato, 2020), suggesting a significant relationship between CSR activities and ESG performance indicators. Conventionally, CSR is understood as a corporation’s dedication to ethical business practices, considering its effects on society, the environment, and the economy through initiatives like philanthropy and sustainable environmental practices. In contrast, ESG serves as a framework for assessing a company’s actions and policies through quantifiable metrics, facilitating comparisons among organisations and prioritising accountability and performance evaluation in governance, environmental initiatives, and social responsibility, thus linking CSR efforts to broader standards crucial for investors and stakeholders (Xie et al., 2018; Supriyono et al., 2024). ESG management, encompassing areas like carbon reduction and social contribution, has emerged as a new standard for corporate operations, building upon and extending the principles of CSR by actively integrating environmental, social, and governance factors into business strategy (Yun & Lee, 2022). For example, Korean companies are adopting ESG to meet modern demands and enhance their sustainability management, demonstrating how ESG expands CSR by considering a broader range of stakeholders and issues (Yun & Lee, 2022). Ultimately, ESG represents a more evolved and quantifiable framework that builds upon the foundational principles of CSR, translating a company’s voluntary social and environmental commitments into measurable performance indicators crucial for modern investment and stakeholder assessment.

2.2. Empirical Literature on CSR and Financial Performance Nexus

The concept of CSR is not a recent development, as it has been gaining traction for more than ten years (Amahalu & Okudo, 2023). This concept encompasses various definitions and has undergone significant evolution in both its interpretation and implementation (Okafor et al., 2021). The definition of CSR proposed by Carroll remains the most widely recognised and employed in both practical applications and academic research, even though there continues to be unresolved questions and ambiguities surrounding the concept (Barauskaite & Streimikiene, 2021; Han et al., 2019). CSR investment can be understood as an approach that integrates traditional investment objectives with considerations of ESG factors, aiming to secure favourable long-term results (Masongweni & Simo-Kengne, 2024). CSR entails voluntary initiatives undertaken by companies to promote social welfare, extending beyond their own interests and legal obligations (Saeed et al., 2023). Examples of CSR initiatives encompass various actions, such as supporting employee welfare initiatives, providing financial contributions and sponsorships to local communities, and investing in environmental sustainability, among others (An & Albattat, 2023).

An examination of the pertinent literature reveals that the investigation into the correlation between CSR and FP has been a prominent area of interest since the 1970s (Zhou et al., 2021). Previous studies have investigated the connection between CSR and FP; however, a unified perspective has yet to emerge due to the varying definitions and measurement standards associated with CSR (Ang et al., 2022; Rahi et al., 2022). The internal governance structure of companies also influences the relationship between CSR and FP (Ang et al., 2022).

The correlation between CSR and FP has been extensively examined by academics, and three distinct perspectives have emerged. There is an increasing discourse suggesting that participation in CSR positively influences FP, although this assertion remains a subject of contention as contended by Orlitzky et al. (2003). Margolis and Walsh (2001) also observe that a significant portion of the existing literature indicates a positive correlation between CSR and FP. Their findings indicate that around 50 percent of the studies identified a positive correlation between CSR and FP, whereas 25 percent found no significant relationship, 20 percent yielded inconclusive results, and 5 percent indicated a negative correlation. Margolis et al. (2009) reviewed the literature from 1972 to 2007 and the results revealed that most studies have a positive significant correlation between CSR and FP. These findings align with the notion that CSR contributes to the enhancement of corporate reputation and fosters stakeholder support, ultimately leading to improved economic results. Socially responsible companies acquire a competitive advantage, as investors are increasingly inclined to allocate their resources to companies demonstrating greater social commitment (Ramasamy et al., 2007).

Some other authors have observed a positive correlation between CSR and FP. Nyeadi et al. (2018) conducted a study using 100 JSE-listed companies from 2011 to 2013 and found that CSR positively impacts FP in South African firms, particularly in larger firms. The study also discovered that governance performance significantly contributes, while social and environmental components show no notable effect on FP. Chininga et al. (2024) analysed the impact of ESG ratings on the FP of 40 JSE-listed firms within the FTSE/JSE Responsible Investment Index from 2015 to 2019 by employing a two-stage least squares (2SLS) method. The findings revealed that only environmental initiatives significantly enhance FP, while social and governance practices show no measurable effect. These results are corroborated by other authors who discovered that CSR performance has a positive influence on FP (Aftab et al., 2024; Akhand et al., 2024; Bristy et al., 2021; Kaimal & Uzma, 2024). These findings align with the notion that CSR contributes to the enhancement of corporate reputation and fosters stakeholder support, ultimately leading to improved economic results. Socially responsible companies acquire a competitive advantage, as investors are increasingly inclined to allocate their resources to firms demonstrating greater social commitment (Ramasamy et al., 2007).

A second line of inquiry indicates a negative correlation between CSR performance and financial performance (Mansaray et al., 2017). Kabir and Chowdhury (2023) investigated the relationship between CSR and FP among 30 listed banks in Bangladesh over the period from 2006 to 2018 utilising a Panel Vector Autoregression (VAR) approach. Their findings indicated a negative impact of CSR on FP. Similarly, Lin et al. (2019) employed a Panel VAR methodology combined with the Generalised Method of Moments (GMM) on a dataset comprising 1000 firm-year observations from 2007 to 2016 to explore the linkages between CSR and FP. Their analysis also revealed that CSR activities adversely affect the financial performance of companies. Furthermore, Sekhon and Kathuria (2019) assessed the impact of CSR on FP by employing panel data regression analysis on a sample of 137 leading Indian firms from 2008 to 2017. Their results suggested that CSR has either a neutral or negative effect on FP, aligning with the profit maximisation theory posited by Friedman (1970). These studies underscore the complex and often detrimental relationship between CSR initiatives and corporate financial outcomes.

The conventional perspective posits that CSR entails financial burdens, given that certain social initiatives require supplementary expenditures (An & Albattat, 2023). Dzomonda and Fatoki (2020) state that the implementation of CSR initiatives diverts financial resources that could otherwise be allocated to more lucrative investments, thereby adversely impacting overall FP. Consequently, this line of inquiry suggests that while CSR initiatives may enhance a company’s reputation and stakeholder relations over time, they can initially lead to negative results concerning FP, particularly if resources are not effectively managed or aligned with core business objectives.

A third group of researchers identifies no correlation between CSR and FP in companies. du Toit and Lekoloane (2018) examined the relationship between CSR and FP among firms listed on the JSE from 2009 to 2014, employing logistic regression analysis as their methodological approach. Their findings indicated the absence of a clear and direct correlation during the specified period; however, they underscored the significance of factors such as firm size and reputation in the recognition of CSR within the JSE context. Supporting these findings, Sharma et al. (2021) conducted a study analysing a sample of 10 companies using the correlation technique method from 2008 to 2017, and their results similarly revealed no significant relationship between CSR and FP. Moreover, Adamkaite et al. (2023) aligned with these conclusions, as their study of nine Lithuanian energy companies from 2017 to 2020 yielded neutral results regarding the relationship between CSR and FP.

To date, the bulk of studies in this area have relied on different methods for analysing this relationship. Conversely, the GLS method, which offers a powerful framework for addressing non-normality, autocorrelation, and heteroscedasticity, has seen relatively scant application in this context (Tshipa et al., 2018). This study aims to address this knowledge gap by exploring the efficacy of GLS in providing more robust and reliable estimates, thereby enhancing the credibility and generalizability of findings in this area. By adopting this novel methodological approach, this research seeks to contribute to a deeper understanding of the underlying relationships and dynamics, ultimately informing more informed policy and decision making in this critical field.

Based upon the literature review, the researchers have formulated the following hypotheses:

H1. 

There is a positive correlation between high CSR scores and the financial performance of JSE-listed companies.

H2. 

ESG scores have a significant positive impact on the financial performance of JSE-listed companies.

H3. 

The social pillar, a component of ESG, is positively associated with the financial performance of JSE-listed companies.

3. Methodology

3.1. Population, Sample, and Data Source

The population of this study is all the firms that were listed on the JSE from 2017 to 2022. Companies that were delisted during this period were not considered for the study. The research employs panel data that include 104 companies listed on the JSE, covering the years from 2017 to 2022. The duration of the study and the sample selection were also constrained by the accessibility of CSR data.

Financial data were sourced from the IRESS database, which is widely utilised by South African researchers and is esteemed for its reliability (Matemane & Graça, 2023). Data on CSR performance were obtained from the LSEG Workshop, formerly known as the Refinitiv Eikon database. Information regarding control variables, including board characteristics and other factors influencing CSR, was gathered from the integrated reports of the firms and their official websites.

3.2. Variables

3.2.1. Independent Variable

CSR performance is the independent variable in this study. As previously mentioned, this study uses a multicentric approach to measure CSR performance. CSR performance is measured firstly in terms of the CSR score. The study also utilises the ESG score as an alternative measure of CSR performance for robustness checks. The use of ESG scores as an alternative measure of CSR performance is consistent with Bifulco et al. (2023), who argue that investors frequently utilise ESG scores as a primary metric to assess a company’s overall CSR performance. Furthermore, the study also utilises the social pillar of the ESG framework based on Matemane et al. (2024), who posit that the ESG social pillar holds particular significance within the context of South Africa, a developing nation grappling with challenges such as poverty, unemployment, inequality, and various other social issues.

3.2.2. Dependent Variable

This study uses both accounting-based and market-based measures, as there has been a notable deficiency in research conducted in South Africa concerning the relationship between CSR performance and financial performance (Dzomonda & Fatoki, 2020). This study measures FP using the Return on Assets (ROA), Return on equity (ROE), market value added (MVA), Tobin’s Q ratio (TOBQ), economic value added (EVA), and the earnings per share (EPS) ratio like other previous studies that have been conducted in this regard (Amahalu & Okudo, 2023; An & Albattat, 2023; Masongweni & Simo-Kengne, 2024).

3.2.3. Control Variables

This study controls for company specific characteristics to address the issue of omitted variables bias since corporate performance might be explained by factors other than CSR performance. The control variables for firm size and leverage were utilised in the analysis as Evans et al. (2023) state that this will improve the validity of the results. Firm size was measured in terms of total assets, as performed by Saeed et al. (2023).

This study also includes board characteristics as control variables as Ang et al. (2022) stated that internal governance structures of companies also influence CSR performance and FP correlation. Board characteristics included in this study are measured in terms of board size, board gender diversity, board independence, board race, number of board meetings, presence of the CSR committee, independence of the CSR committee, CSR committee members diversity, and the frequency of CSR committee meetings.

3.3. Research Method

This study adopts a quantitative research methodology to investigate the relationship between CSR performance and FP. The analysis is conducted using the GLS regression method, which is suitable for addressing issues of heteroskedasticity and autocorrelation in panel data (Tshipa et al., 2018; Wahba, 2015). The GLS method enhances the reliability and accuracy of the parameter estimates in the presence of these common econometric issues (Fang et al., 2024). To ensure the robustness of findings, the analyses are conducted across two distinct sample structures, which is the full sample of 104 companies and a sub-sample that includes only nonfinancial firms. This distinction allows for an exploration of sector-specific effects, particularly given that financial institutions operate under unique regulatory frameworks that could influence both CSR performance and FP.

4. Results and Discussion

4.1. Descriptive Statistics

The descriptive statistics concerning the financial and CSR performance indicators utilised in this research are presented in

Table 1. Descriptive statistics.

	n	Mean	SD	Min	Max
roa	624	7.118	11.265	−30.41	43.040
roe	624	7.884	30.650	−158.65	77.270
eps	624	760.343	1425.127	−923.300	8843.49
tobq	624	1.119	0.885	0.140	4.530
eva	624	−2.894	15.172	−106.219	29.820
mva	624	1.637	1.433	0.470	9.520
csr_score	624	50.585	29.402	0.39	99.58
esg	624	50.337	17.611	4.14	88.88
soc	624	53.787	21.053	1.62	96.34
bsize	624	11.495	2.753	4.000	20.000
bind	624	61.033	14.440	0.000	90.909
gend	624	28.101	11.873	0.000	69.231
ethn	624	33.755	16.942	0.000	84.615
bmeet	624	6.0691	2.562	1.000	24.000
lev	624	0.560	0.224	0.015	1.004
fsize	624	141,858.700	350,062.000	1038.838	2,004,402.000

Source: Own computation.

. Focusing on the accounting-based performance metrics, the mean values for ROA and ROE are recorded at 7.118% and 7.884%, respectively. It is noteworthy that the firms within the sample exhibit significant performance variability, with the top-performing firm achieving an ROA of 43.040% and the lowest-performing firm showing an ROA of −30.41%. Similarly, the ROE figures range from 77.270% for the highest performer to −158.65% for the lowest. The observed disparity in profitability metrics is further supported by standard deviation values of 11.265 for ROA and 30.650 for ROE. The average EPS across the sample is ZAR 760.343, with the highest and lowest EPS recorded at ZAR 8843.49 and ZAR–923.300, respectively, during the sample period. The substantial standard deviation in EPS indicates considerable variation among the selected JSE firms regarding their earnings distribution decisions for shareholders.

Regarding market-based performance indicators, the average Tobin’s Q for the sample of JSE companies stands at 1.119, accompanied by a standard deviation of 0.885. The Tobin’s Q values among the selected firms range from a minimum of 0.140 to a maximum of 4.530 throughout the sample period. Additionally, EVA and MVA exhibit average values of −2.894 and 1.637, respectively, during the study period. EVA values fluctuate between −106.219 and 29.820, with a standard deviation of 15.172. Conversely, MVA shows minimum and maximum values of 0.470 and 9.520, respectively, with a standard deviation of 1.433. A key conclusion from this analysis is that the selected JSE firms demonstrated varied performance across the financial outcome metrics throughout the study period.

The performance indicators for CSR reveal that the average values for the CSR strategic score (csr_score), composite ESG score, and the social dimension of ESG (socia) are approximately 51%, 50%, and 54%, respectively. Notably, the CSR score for the firms listed on the JSE ranges from a minimum of 0.39% to a maximum of 99.58%, with a standard deviation of 29.4. Regarding other CSR performance metrics, the composite ESG score varies between 14.14% and 88.88%, while the social component of ESG fluctuates between 1.62% and 96.34%. This variation indicates that the firms in the sample exhibit differing levels of CSR performance, as evidenced by the observed minimum and maximum values. The disparities are further substantiated by the standard deviations, which are calculated to be 29.4 for the CSR score, 17.611 for the overall ESG score, and 21.053 for the social pillar of ESG.

Regarding the control variables related to board characteristics, the firms in the sample report an average board size of approximately 11 members, with independent directors, female directors, and black directors constituting about 61%, 28%, and 34% of the board, respectively. It is noteworthy that the highest proportions of independent, female, and ethnic directors within the board are approximately 91%, 69%, and 85%, respectively. On average, the JSE firms convened for board meetings six times per year during the study period, with the number of meetings ranging from a minimum of one to a maximum of 24. In terms of firm-specific control variables, the average leverage ratio for the sample firms is recorded at 0.560, accompanied by a standard deviation of 0.224. The average total assets of the firms under study amount to ZAR 141.858 billion, with a substantial standard deviation of 350,062,000. Furthermore, there exists a significant disparity in firm size among the JSE sample firms, with the smallest and largest asset sizes recorded at ZAR 1.038 billion and ZAR 2004.402 billion, respectively.

4.2. Correlational Analysis

Correlational analysis offers preliminary insights into the relationships between the dependent variable and the independent variables. Additionally, this analysis can indicate the potential presence of multicollinearity among the explanatory variables within the model. To assess the strength of the associations among the variables of interest, the study utilises the Pearson moment correlation coefficient, as detailed in

Table 2. Pairwise correlations for CSR performance and firm performance nexus.

Var	roa	roe	eps	tobq	eva	mva	csr_s	esg	socia	bsize	bind	gend	ethn	bmeet	lev	(fsize)
roa	1.000
roe	0.595 ***	1.000
eps	0.292 ***	0.316 ***	1.000
tobq	0.535 ***	0.271 ***	0.233 ***	1.000
eva	0.312 ***	0.082 **	0.181 ***	0.246 ***	1.000
mva	0.415 ***	0.342 ***	0.202 ***	0.712 ***	0.100 **	1.000
csr_s	0.084 **	0.081 **	0.239 ***	0.030	−0.198 ***	0.146	1.000
esg	0.210 ***	0.110 ***	0.251 ***	0.098 **	−0.093 **	0.148 ***	0.644 ***	1.000
soc	0.206 ***	0.140 ***	0.300 ***	0.125 ***	−0.069 *	0.124 ***	0.630 ***	0.884 ***	1.000
bsize	−0.109 ***	0.092 **	0.308 ***	−0.041	−0.201 ***	−0.038	0.372 ***	0.321 ***	0.368 ***	1.000
bind	−0.033	−0.014	−0.068 *	−0.110 ***	−0.096 **	−0.013	0.217 ***	0.409 ***	0.276 ***	0.093 **	1.000
gend	0.118 ***	0.029	0.103 **	0.057	0.039	0.131 ***	0.208 ***	0.375 ***	0.324 ***	0.064	0.362 ***	1.000
ethn	0.004	−0.023	0.098 **	−0.157 ***	−0.114 ***	−0.027	0.212 ***	0.279 ***	0.277 ***	0.195 ***	0.222 ***	0.416 ***	1.000
meet	−0.155 ***	−0.234 ***	0.100 ***	−0.170 ***	−0.160 ***	−0.170 ***	0.193 ***	0.167 ***	0.176 ***	0.136 ***	0.140 ***	0.119 ***	0.115 ***	1.000
lev	−0.255 ***	−0.118 ***	−0.124 ***	−0.152 ***	−0.255 ***	0.090 **	0.315 ***	0.223 ***	0.262 ***	0.159 ***	0.113 ***	0.153 ***	0.109 ***	0.083 **	1.000
fsize	−0.125 ***	0.180 ***	0.358 ***	−0.108 ***	−0.343 ***	−0.039	0.431 ***	0.395 ***	0.441 ***	0.629 ***	0.143 ***	0.030	0.148 ***	0.209 ***	0.324 ***	1.000

Note: ***, **, and * indicate 1%, 5%, and 10% significance levels, respectively. Source: Own computation.

. In examining column 1, it is evident that the three indicators of CSR performance—csr_s, esg, and soc—exhibit positive and statistically significant correlations with the accounting-based performance measure ROA, with correlation coefficients of 0.084, 0.210, and 0.206, respectively. Similar patterns are observed between the CSR performance indicators and other accounting-based metrics, such as ROE and EPS. Conversely, regarding market-based performance indicators, the composite ESG and the social pillar (soc) show positive and significant correlations with TOBQ, with correlation coefficients of 0.098 and 0.125, respectively. Furthermore, MVA is positively and significantly correlated with ESG (0.146) and soc (0.124) at the 5% and 1% significance levels, respectively. In contrast, the market-based performance measure, EVA, demonstrates negative and significant correlations with csr_s (−0.198), esg (−0.093), and soc (−0.069) at the 1%, 5%, and 10% significance levels, respectively.

The investigation into the relationship between board characteristics and firm performance indicators reveals that board size (bsize) exhibits a negative and statistically significant correlation with performance metrics such as ROA at −0.109 and EVA at −0.201. Conversely, it positively correlates with accounting-based performance indicators, including ROE at 0.092 and EPS at 0.308. This indicates that the relationship between board size and firm performance depends on the specific performance proxy utilised. In contrast, board independence (bind) demonstrates a negative and significant relationship with performance indicators such as EPS (−0.068), TOBQ (−0.110), and EVA (−0.096), significant at the 10%, 1%, and 5% levels, respectively. Additionally, gender diversity (gend) is positively and significantly correlated with firm performance measures, including roa (0.118), eps (0.103), and MVA (0.131) at the 1%, 5%, and 1% significance levels, respectively. The relationship between ethnic diversity (ethn) and firm performance appears to be mixed, varying according to the performance measure employed. Furthermore, board meetings (bmeet) are negatively and significantly associated with all the performance indicators at the 1% significance level. Regarding firm-specific control variables, a negative and significant correlation is observed between the leverage ratio (lev) and all the performance metrics, except for MVA, which shows a positive association with lev at the 5% significance level. Similarly, firm size (fsize) is negatively and significantly correlated with ROA (−0.125) and market-based performance measures such as TOBQ (−0.108) and EVA (−0.343) at the 1% level, while a positive correlation is noted between fsize and accounting-based performance metrics like ROE (0.180) and EPS (0.358) at the 1% level.

The analysis of the correlation coefficients presented in Table 2 indicates that the relationships among the explanatory variables in this study are characterised by weak to moderate correlations. According to Gujarati and Porter (2008), a correlation coefficient of 0.8 is considered the minimum threshold for identifying multicollinearity. The data in Table 2 reveals that the highest correlation coefficient observed is between fsize and bsize, which stands at 0.629, falling below the established threshold. This finding implies that multicollinearity does not pose a significant concern across the estimated models.

4.3. Preliminary Tests to Choose Among the Pooled OLS, Random Effect, and Fixed Effect Models

A critical preliminary assessment in panel analysis involves evaluating the presence of a panel effect within the dataset. This evaluation is essential to ensure that the most appropriate methodology is employed to fulfil the study’s objectives. To conduct this assessment, the study initially applies the Breusch–Pagan LM test, as formulated by Breusch and Pagan. This test operates under the null hypothesis that no panel effect exists within the model. The null hypothesis is deemed rejected if the probability associated with the Breusch–Pagan test is below 0.05 at a 5% significance level. The results of the Breusch–Pagan test are summarised in

Table 3. Breusch–Pagan and Hausman tests.

	test	prob
B-P test (chi bar square)	147.88 ***	0.000
Hausman test (chi square)	18.52 **	0.018

Note: ***, **, and * indicate 1%, 5%, and 10% significance levels, respectively. Source: Own computation.

, which indicates a probability of 0.000, significant at the 1% level, thereby confirming the rejection of the null hypothesis. This outcome suggests the existence of a panel effect, which precludes the application of pooled OLS as an estimation technique. Following the identification of the panel effect, it is crucial to differentiate between the two panel effect models—fixed effect and random effect—utilising the widely recognised Hausman test, as recommended by Hausman (1978). The null hypothesis of the Hausman test favours the random effect model if the test probability exceeds 0.05. The findings of the Hausman test are also presented in Table 3, revealing a Hausman test statistic of 18.52, which is statistically significant at the 5% level (p < 0.05). This result leads to the rejection of the null hypothesis, indicating that the fixed effect model is superior to the random effect model. Consequently, based on the Hausman test results, the study adopts the fixed effect model as the most appropriate estimation technique to analyse the impact of CSR performance on FP among the 104 selected JSE firms in South Africa.

4.4. Multicollinearity and Linearity Tests

To assess multicollinearity among the explanatory variables, this study utilises the variance inflation factor (VIF). In accordance with the findings of Ojeyinka and Matemane (2024), multicollinearity is indicated if the VIF for any variable exceeds 10.0, or if the mean VIF for all the independent variables surpasses 2.0. The results presented in

Table 4. Result of VIF.

Variable	VIF	1/VIF
fsize	1.96	0.511
bsize	1.73	0.577
csr_s	1.39	0.718
gend	1.38	0.723
ethn	1.27	0.786
bind	1.20	0.833
lev	1.19	0.837
bmeet	1.08	0.93
Mean (VIF)	1.40

Source: Own computation.

indicate that the variable fsize exhibits the highest VIF at 1.96, while the overall average VIF is calculated to be 1.40. Consequently, given that both the maximum VIF and the average VIF remain below 2.0, the study concludes that there is no multicollinearity issue present in the estimated models. Additionally, these VIF results corroborate the findings from the correlation analysis, which identified only moderate to weak associations among the independent variables.

4.5. Diagnosis Tests

The residual term of the estimated model demonstrates a departure from the normality assumption, as evidenced by its negative skewness; a high kurtosis of 5.543, which exceeds the threshold of 3.0; and statistically significant results from the Jarque–Bera test which is presented in

Table 5. Skewness–kurtosis-based normality test for CSR performance and firm performance relationship.

Variable	Skewness	Kurtosis	Jacque Bera	Prob (J-B)
Residuals	−0.423	5.543	185.765	0.000

Source: Own computation.

. Furthermore,

Table 6. Wooldridge autocorrelation test.

	Test	Prob
Wooldridge test (F-stat)	16.401 ***	0.001

Note: ***, **, and * indicate 1%, 5%, and 10% significance levels, respectively. Source: Own computation.

presents the Wooldridge test for autocorrelation, which rejects the null hypothesis, revealing that the OLS error term exhibits first-order autocorrelation at a significance level of 1%. Compounding these issues, the results from the Modified Wald test indicate heteroscedasticity present in the residual term of the fixed effect model. Consequently, these violations of fundamental assumptions undermine the appropriateness of using the fixed effect model in this analysis. The GLS method allocates weights to individual observations, enabling the generation of estimators that are Best Linear Unbiased Estimators (BLUE) (Gujarati, 2003). These arguments support the use of the GLS method as the preferred estimation technique, given its efficacy in addressing issues related to non-normality, autocorrelation, and heteroscedasticity in the residuals (Tshipa et al., 2018; Wahba, 2015).

4.6. Regression Results

The findings derived from the GLS methodology regarding the influence of CSR score (csr_s) on FP are presented in

Table 7. Effect of CSR score on firm performance.

Panel A: Full Sample							Panel B: Nonfinancial
	(1)	(2)	(3)	(4)	(6)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
VARIABLES	roa	roe	eps	tobq	eva	mva	roa	roe	eps	tobq	eva	mva
csr_score	0.097 ***	0.095 **	8.374 ***	0.005 ***	−0.012	0.010 ***	0.035	0.058	5.364 *	0.004 *	0.0052	0.006 *
	(0.017)	(0.045)	(2.159)	(0.001)	(0.022)	(0.002)	(0.023)	(0.061)	(2.756)	(0.002)	(0.016)	(0.003)
bsize	−0.480 **	−0.667	44.06 *	0.009	0.137	−0.025	−0.403	−0.379	49.96	0.016	0.066	−0.006
	(0.198)	(0.540)	(25.71)	(0.016)	(0.266)	(0.026)	(0.257)	(0.687)	(30.99)	(0.022)	(0.184)	(0.036)
bind	−0.062 *	−0.080	−20.78 ***	−0.008 ***	−0.062	−0.007	−0.079 **	−0.091	−21.77 ***	−0.009 ***	−0.091 ***	−0.007
	(0.032)	(0.087)	(4.121)	(0.003)	(0.043)	(0.004)	(0.037)	(0.099)	(4.484)	(0.003)	(0.027)	(0.005)
gend	0.177 ***	0.294 ***	22.39 ***	0.015 ***	0.190 ***	0.020 ***	0.216 ***	0.203	25.77 ***	0.019 ***	0.077 **	0.025 ***
	(0.041)	(0.112)	(5.343)	(0.003)	(0.055)	(0.005)	(0.050)	(0.135)	(6.072)	(0.004)	(0.036)	(0.007)
ethn	−0.023	−0.112	−0.141	−0.011 ***	−0.088 **	−0.008 **	−0.074 **	−0.130	−3.562	−0.014 ***	−0.000	−0.010 **
	(0.028)	(0.076)	(3.600)	(0.002)	(0.037)	(0.004)	(0.034)	(0.091)	(4.112)	(0.003)	(0.024)	(0.005)
meet	−0.728 ***	−3.511 ***	11.22	−0.056 ***	−0.515 **	−0.110 ***	−0.870 ***	−4.280 ***	−2.123	−0.077 ***	−0.485 ***	−0.133 ***
	(0.167)	(0.457)	(21.76)	(0.014)	(0.225)	(0.022)	(0.227)	(0.606)	(27.36)	(0.019)	(0.162)	(0.032)
leverage	−15.84 ***	−30.71 ***	−2.166 ***	−0.659 ***	−11.22 ***	0.381	−17.82 ***	−51.30 ***	−2.948 ***	−0.562 **	−4.226 **	0.376
	(2.011)	(5.492)	(261.7)	(0.162)	(2.709)	(0.266)	(2.873)	(7.677)	(346.4)	(0.244)	(2.056)	(0.401)
fsize	−0.083	6.316 ***	359.2 ***	−0.036	−2.493 ***	−0.053	1.466 ***	7.450 ***	555.9 ***	0.046	−0.202	0.060
	(0.366)	(0.999)	(47.58)	(0.030)	(0.493)	(0.048)	(0.523)	(1.398)	(63.07)	(0.044)	(0.374)	(0.073)
Constant	21.37 ***	−15.86 *	−2080 ***	2.333 ***	32.72 ***	2.502 ***	13.78 ***	−8.489	−3.208 ***	1.767 ***	9.240 ***	1.584 **
	(3.236)	(8.837)	(421.1)	(0.261)	(4.360)	(0.429)	(4.954)	(13.24)	(597.3)	(0.420)	(3.545)	(0.691)
Wald test	124.41 ***	119.41 ***	222.81 ***	85.26 ***	124.86 ***	67.29 ***	82.71 ***	131.21 ***	277.43 ***	64.78 ***	31.58 ***	35.44 ***
Prob (Wald test)	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}
Observations	621	621	621	621	621	621	430	430	430	430	430	430
Number of c_id	104	104	104	104	104	104	72	72	72	72	72	72
Standard errors are in parentheses (); figures in {} represent probability value.

Note: *** p < 0.01, ** p < 0.05, * p < 0.1. Source: Own computation.

, with the overall sample and nonfinancial firms categorised as panel A and panel B, respectively. Initially focusing on the primary variable, the analysis indicates a significant positive correlation between CSR score and FP across the complete sample, except for the EVA in model 6. Statistically, an increase in one unit in the CSR score results in enhancements of 0.097, 0.095, 8.374, 0.005, and 0.010 units in ROA, ROE, EPS, TOBQ, and MVA, respectively, at varying levels of significance.

This finding confirmed the previous findings of Aftab et al. (2024), Kaimal and Uzma (2024), and Li et al. (2023), grounded in the stakeholder theory. It also demonstrates that CSR performance enhances a company’s financial performance, thereby corroborating Freeman’s (1984) assertion that a CSR-focused strategy can lead to improved financial results over the long term. The beneficial effect of CSR involvement on FP is evident across both accounting-based financial metrics (models 1, 2, and 3) and market-oriented performance indicators such as TOBQ and MVA in models 4 and 5. Conversely, within the nonfinancial subsample (panel B), the CSR score positively influences financial performance as indicated by EPS, TOBQ, and MVA at a significance level of 10%. This finding implies that a company’s commitment to CSR initiatives fosters profitability and enhances firm value, thereby contributing to improved FP for the selected companies listed on the JSE. Furthermore, when employing the composite ESG score (esg) and the social dimension (socia) as indicators of CSR performance, the results illustrated in

Table 8. Effect of overall esg score on firm performance.

	Panel A: Full Sample						Panel B: Nonfinancial Firms
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
Dep. Var.	roa	roe	eps	tobq	eva	mva	roa	roe	eps	tobq	eva	mva
esg	0.253 ***	0.178 **	16.37 ***	0.013 ***	0.077 *	0.017 ***	0.220 ***	0.234 **	13.94 ***	0.0128 ***	0.066 **	0.016 ***
	(0.028)	(0.080)	(3.812)	(0.002)	(0.040)	(0.004)	(0.039)	(0.108)	(4.871)	(0.003)	(0.029)	(0.006)
bsize	−0.458 **	−0.607	48.99 *	0.010	0.074	−0.018	−0.530 **	−0.466	49.14	0.012	0.016	−0.008
	(0.190)	(0.536)	(25.48)	(0.016)	(0.264)	(0.026)	(0.246)	(0.677)	(30.54)	(0.021)	(0.181)	(0.035)
bind	−0.124 ***	−0.119	−24.42 ***	−0.011 ***	−0.089 **	−0.010 **	−0.140 ***	−0.154	−25.28 ***	−0.012 ***	−0.110 ***	−0.011 **
	(0.032)	(0.090)	(4.267)	(0.003)	(0.044)	(0.004)	(0.038)	(0.104)	(4.680)	(0.003)	(0.028)	(0.005)
gend	0.114 ***	0.255 **	18.77 ***	0.012 ***	0.161 ***	0.017 ***	0.136 ***	0.124	21.63 ***	0.015 ***	0.051	0.020 ***
	(0.041)	(0.115)	(5.461)	(0.003)	(0.057)	(0.006)	(0.0509)	(0.140)	(6.311)	(0.004)	(0.037)	(0.007)
ethn	−0.030	−0.115	−0.391	−0.011 ***	−0.094 **	−0.008 **	−0.089 ***	−0.147	−4.653	−0.015 ***	−0.005	−0.011 **
	(0.027)	(0.076)	(3.593)	(0.002)	(0.037)	(0.004)	(0.033)	(0.091)	(4.101)	(0.003)	(0.024)	(0.005)
bmeet	−0.685 ***	−3.457 ***	15.81	−0.054 ***	−0.538 **	−0.104 ***	−0.814 ***	−4.202 ***	4.126	−0.073 ***	−0.472 ***	−0.126 ***
	(0.161)	(0.455)	(21.63)	(0.013)	(0.224)	(0.022)	(0.219)	(0.602)	(27.13)	(0.019)	(0.161)	(0.031)
lev	−14.54 ***	−29.18 ***	−2.034 ***	−0.596 ***	−11.78 ***	0.549 **	−17.57 ***	−50.74 ***	−2.887 ***	−0.527 **	−4.228 **	0.445
	(1.912)	(5.397)	(256.6)	(0.157)	(2.656)	(0.262)	(2.763)	(7.594)	(342.5)	(0.239)	(2.031)	(0.396)
fsize	−0.415	6.197 ***	346.8 ***	−0.054 *	−2.776 ***	−0.061	0.998 **	7.076 ***	545.2 ***	0.027	−0.374	0.046
	(0.357)	(1.007)	(47.89)	(0.029)	(0.496)	(0.049)	(0.503)	(1.383)	(62.39)	(0.044)	(0.370)	(0.072)
Constant	21.60 ***	−17.13 **	−2.177 ***	2.356 ***	35.03 ***	2.333 ***	16.32 ***	−6.975	−3.229 ***	1.828 ***	10.31 ***	1.570 **
	(3.071)	(8.665)	(412.0)	(0.253)	(4.265)	(0.421)	(4.695)	(12.90)	(581.8)	(0.406)	(3.450)	(0.673)
Wald test	175.65 ***	120.07 ***	227.34 ***	106.05 ***	129.12 ***	65.92 ***	117.21 ***	136.16 ***	284.60 ***	77.13 ***	37.09 ***	40.55 ***
Prob (Wald test)	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}
Number of c_id	104	104	104	104	104	104	72	72	72	72	72	72
Standard errors are in parentheses () while figures in {} represent probability value.

Note: *** p < 0.01, ** p < 0.05, * p < 0.1. Source: Own computation.

and

Table 9. Effect of social component of ESG on firm performance.

	Panel A: Full Sample						Panel B: Nonfinancial Sectors
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)
VARIABLES	roa	roe	eps	tobq	eva	mva	roa	roe	eps	tobq	eva	mva
socia	0.211 ***	0.204 ***	15.56 ***	0.012 ***	0.105 ***	0.012 ***	0.190 ***	0.269 ***	11.84 ***	0.012 ***	0.045 *	0.013 ***
	(0.023)	(0.065)	(3.094)	(0.002)	(0.032)	(0.003)	(0.033)	(0.090)	(4.058)	(0.003)	(0.024)	(0.005)
bsize	−0.513 ***	−0.696	43.72 *	0.006	0.020	−0.019	−0.535 **	−0.546	49.03	0.011	0.028	−0.007
	(0.190)	(0.535)	(25.42)	(0.016)	(0.263)	(0.026)	(0.246)	(0.674)	(30.52)	(0.021)	(0.181)	(0.035)
bind	−0.079 **	−0.096	−21.80 ***	−0.009 ***	−0.082 *	−0.007	−0.111 ***	−0.136	−23.40 ***	−0.010 ***	−0.099 ***	−0.009 *
	(0.031)	(0.086)	(4.100)	(0.003)	(0.042)	(0.004)	(0.036)	(0.010)	(4.516)	(0.003)	(0.027)	(0.005)
gend	0.121 ***	0.239 **	18.51 ***	0.012 ***	0.148 ***	0.018 ***	0.144 ***	0.104	22.25 ***	0.015 ***	0.0592	0.022 ***
	(0.040)	(0.114)	(5.402)	(0.003)	(0.056)	(0.006)	(0.050)	(0.138)	(6.234)	(0.004)	(0.037)	(0.007)
ethn	−0.035	−0.124 *	−0.932	−0.012 ***	−0.101 ***	−0.008 **	−0.093 ***	−0.158 *	−4.893	−0.016 ***	−0.005	−0.011 **
	(0.027)	(0.0754)	(3.580)	(0.002)	(0.037)	(0.004)	(0.033)	(0.091)	(4.106)	(0.003)	(0.024)	(0.005)
bmeet	−0.714 ***	−3.496 ***	13.30	−0.055 ***	−0.561 **	−0.105 ***	−0.887 ***	−4.297 ***	−0.476	−0.077 ***	−0.491 ***	−0.131 ***
	(0.161)	(0.454)	(21.53)	(0.013)	(0.223)	(0.022)	(0.218)	(0.599)	(27.13)	(0.019)	(0.161)	(0.031)
lev	−15.34 ***	−30.19 ***	−2.101 ***	−0.648 ***	−12.36 ***	0.518 *	−17.72 ***	−51.04 ***	−2.896 ***	−0.537 **	−4.247 **	0.437
	(1.913)	(5.394)	(256.1)	(0.157)	(2.651)	(0.264)	(2.756)	(7.558)	(342.4)	(0.238)	(2.036)	(0.397)
fsize	−0.491	5.930 ***	334.5 ***	−0.061 **	−2.958 ***	−0.055	0.854 *	6.634 ***	536.9 ***	0.015	−0.363	0.041
	(0.357)	(1.008)	(47.84)	(0.029)	(0.495)	(0.049)	(0.507)	(1.391)	(63.02)	(0.044)	(0.375)	(0.073)
Constant	22.21 ***	−15.14 *	−2.085 ***	2.356 ***	36.37 ***	2.294 ***	17.34 ***	−3.942	−3,170 ***	1.917 ***	10.24 ***	1.604 **
	(3.071)	(8.662)	(411.2)	(0.253)	(4.256)	(0.424)	(4.709)	(12.91)	(585.0)	(0.407)	(3.478)	(0.678)
Wald test	180.79 ***	125.77 ***	236.43 ***	117.53 ***	137.49 ***	59.02 ***	120.23 ***	141.81 ***	285.14 ***	81.92 ***	34.14 ***	39.37 ***
Prob (Wald test)	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}	{0.000}
Number of c_id	104	104	104	104	104	104	72	72	72	72	72	72
Standard errors are in parentheses (); figures in {} stand for probability value

Note: *** p < 0.01, ** p < 0.05, * p < 0.1. Source: Own computation.

reveal that CSR performance consistently bolsters corporate performance both in Panel A and Panel B. This consistency is observed across both accounting-based and market-based performance measures, applicable to both the full sample and the nonfinancial subsample. In Table 8, statistically, an increase of one unit in the ESG score (esg) results in enhancements of 0.253, 0.178, 16.37, 0.013, 0.077, and 0.017 units in ROA, ROE, EPS, TOBQ, EVA, and MVA, respectively, at varying levels of significance. In Table 9, statistically, an increase of one unit in the social score (socia) results in enhancements of 0.211, 0.204, 15.56, 0.012, 0.105, and 0.012 units in ROA, ROE, EPS, TOBQ, EVA, and MVA, respectively, at varying levels of significance. Thus, it can be concluded that irrespective of the CSR performance metric utilised, CSR practices significantly enhance the performance of firms listed on the JSE. This is consistent with the rational choice and stakeholder–agency theories, and the findings of prior research (Chininga et al., 2024; Fu & Li, 2023; Shan et al., 2024), which found that ESG performance enhances the FP of firms. Masongweni and Simo-Kengne (2024) have also found a positive association between the social pillar score and FP. Thus, a growing body of evidence, supported by established theories, indicates that strong ESG performance is positively linked to improved FP for businesses.

The analysis of board characteristic variables, as presented in Table 7, Table 8 and Table 9, indicates that both board independence (bind) exerts a negative and statistically significant effect on various measures of FP within the aggregate sample (panel A) and among nonfinancial firms (panel B). The results are consistent with those of Khan et al. (2024) and Sahoo et al. (2023), who suggest that independent directors play a passive role and maintain strong connections with company management. Therefore, while promoting independence in board composition is essential for good governance, firms should carefully consider the optimal balance between independence and direct managerial involvement to foster FP. The frequency of board meetings (bmeet) also exerts negative and statistically significant effects on various measures of FP within the aggregate sample (panel A) and among nonfinancial firms (panel B) in Table 7, Table 8 and Table 9. These findings are in contrast with those of Sahoo et al. (2023). This finding suggests that an increase in the proportion of external directors and the frequency of board meetings may detract from FP. While regular meetings are necessary for oversight, excessive frequency may indicate governance issues or a reactive approach to problem-solving, which can hinder a firm’s agility and overall performance. Companies should therefore aim to balance adequate board engagement with maintaining a strategic, outcomes-focused meeting structure.

Conversely, gender diversity (gend) appears to positively influence financial performance metrics across both samples. This effect is robust and consistent across all models detailed in Table 7, Table 8 and Table 9, highlighting that female representation on boards is a significant contributor to FP in South Africa. These results are consistent with studies conducted by Sahoo et al. (2023). In contrast, other board characteristics, such as board size (bsize) and ethnic diversity (ethn), exhibit mixed effects depending on the specific performance measure analysed. For example, findings from Model 1 in Table 7, Table 8 and Table 9 indicate that an increase in board size correlates with a significant decrease in ROA (p < 0.05). This finding supports the study conducted by Sarpong-Danquah et al. (2023), who found that board size negatively impacts FP. At the same time, it shows a positive and statistically significant relationship with the EPS (p < 0.1) for the aggregate sample. In the context of nonfinancial firms (panel B), board size does not emerge as a critical determinant of FP, particularly when CSR scores are utilised as a performance metric, as shown in Table 5. However, Table 8 and Table 9 reveal that board size is negatively and significantly related to ROA at a 10% significance level (model 7 in Table 8 and Table 9), while its influence on other performance indicators remains negligible. These results suggest that while larger boards may dilute decision-making efficiency, they also potentially bring diverse expertise that positively affects FP. Similarly, the effect of ethnic diversity (ethn) on FP is predominantly negative across all model specifications. The influence of the proportion of black directors on FP, as indicated by market-based indices, reveals a correlation where an increase in such representation is linked to a decline in both profitability and market value among the sampled firms in South Africa. This may reflect challenges associated with diverse perspectives, potentially leading to slower consensus in decision making.

In alignment with prior expectations, the leverage ratio exhibits a negative and statistically significant effect on FP across various significance levels, except for the MVA specification (models 6 and 12 in Table 7, Table 8 and Table 9). This adverse effect of the leverage ratio remains consistent across different performance metrics and is evident in both the full sample and the regressions focused on nonfinancial firms. The findings strongly emphasise the detrimental impact of elevated leverage ratios on financial and market performance indicators. Specifically, as illustrated in panel A of Table 7, a one-unit increase in the leverage ratio results in declines of 15.84 units in ROA, 30.71 units in ROE, 2.166 units in EPS, 0.659 units in TOBQ, and 11.22 units in EVA at a 1% significance level. In contrast, a similar increase in leverage ratio adversely affects the ROA, ROE, EPS, TOBQ, and EVA of nonfinancial firms (Panel B, Table 7) by 17.82 units, 51.30 units, 2.948 units, 0.562 units, and 4.226 units, respectively. Interestingly, the analysis indicates that an increase in leverage ratio is associated with a higher MVA of 0.518 units for the entire sample, as shown in Table 8. When examining the nonfinancial companies, the results regarding the leverage ratio’s impact on financial outcomes are consistent with those of the aggregate sample, indicating that firms with high leverage tend to exhibit lower performance, except for the MVA regression.

The examination of the second firm-specific control variable reveals that FP indicators exhibit varying responses to alterations in firm size (fsize). As illustrated in Panel A of Table 7, (fsize) demonstrates a positive and statistically significant influence on ROE in model 2 and EPS in model 3, indicating that larger firms tend to achieve superior performance across the overall sample. Comparable results are observed in Table 8 and Table 9 when utilising the composite ESG and its social component as measures of CSR performance. Conversely, market-based performance metrics, such as TOBQ and EVA, negatively correlate with firm size when analysing the aggregate sample (Panel A). This finding is consistent with the outcomes presented in models 4 and 5 of Table 8 and Table 9. This suggests that the impact of firm size on performance indicators is heterogeneous, contingent upon whether accounting-based or market-based metrics are employed. Furthermore, the relationship between firm performance indicators and firm size within nonfinancial sectors lacks consistency across various specifications. In summary, accounting-based financial performance measures exhibit a positive and significant response to changes in firm size (as seen in models 7, 8, and 9 of Table 7, Table 8 and Table 9). In contrast, the influence of the firm-specific variable on market-based performance appears negligible.

5. Conclusions

The findings of this study provide strong evidence that CSR has a positive and significant impact on firm performance among the companies listed on the JSE in South Africa. The results of the panel data analysis reveal that the CSR score is positively correlated with key financial and market-based performance indicators, including ROA, EPS, MVA, EPS, and TOBQ. The study, furthermore, employed the ESG score and the social pillar score to provide a robust and comprehensive assessment of CSR performance. The ESG score and the social pillar score also revealed that these scores positively influence the FP of the companies in both the financial and market-based indicators, reinforcing the reliability of the findings.

This suggests that companies that prioritise CSR are more likely to experience improved FP and market value. These findings are consistent with the growing body of research highlighting the importance of CSR in promoting sustainable business practices and enhancing firm performance.

The study also sheds light on the role of board characteristics in firm performance. The results indicate that board independence and board meetings have a negative and statistically significant effect on FP, while gender diversity has a positive and significant effect. This suggests that boards with more female representation are better equipped to make decisions that drive business success. These findings have important implications for companies seeking to improve their governance structures and enhance their performance. This study contributes through the inclusion of board characteristics where previous studies (Chininga et al., 2024; Nyahuna & Doorasamy, 2023) had not included them.

Overall, the results of this study contribute to the existing literature on CSR and FP by providing evidence that CSR is a key driver of FP. The study’s findings also highlight the importance of gender diversity in the boardroom in driving business success. As policymakers, investors, and companies seek to promote sustainable business practices, they should prioritise CSR and improve their governance structures, including increasing female representation on their boards.

The implications of this study are far-reaching and have significant policy and practice implications if it is directed toward the leadership of companies, investors, market participants, and regulatory bodies. Firstly, policymakers should consider implementing policies that incentivise companies to adopt CSR practices, such as tax breaks or government contracts. Secondly, investors should consider CSR performance when making investment decisions to promote sustainable business practices. The findings indicate that a company’s emphasis on improving its CSR performance will, in turn, enhance its appeal to various stakeholders. Finally, companies should prioritise CSR and integrate it into their business strategies to improve their financial performance and market value. By doing so, they can contribute to the well-being of society and the environment while enhancing their long-term success.

This study is also subject to limitations. Firstly, the sample of this study is the firms listed on the JSE, which are from different sectors of the economy. Each sector within an economy presents unique risks and opportunities, making it inappropriate to generalise findings across all sectors. Consequently, future research should concentrate on the sector-specific impacts of CSR performance on FP rather than examining all the JSE-listed companies. Secondly, future research should explore the role of board characteristics as potential moderating variables in the relationship between CSR and financial performance. Investigating these interaction effects could provide deeper insights into how specific governance structures amplify or weaken the impact of CSR performance on FP. Thirdly, this study has been conducted for JSE-listed companies. Future research can address this limitation by conducting a cross-country analysis. Lastly, this study does not account for the potential effects of greenwashing, and this limitation presents a valuable opportunity for future research. By incorporating measures such as the Refinitiv ESG Controversies Score or similar indicators, future studies can critically assess the authenticity of CSR disclosures. This will allow for a more nuanced understanding of how the credibility of CSR performance influences FP, particularly in contexts where reputational motivations may drive reporting practices. Notwithstanding this limitation, the current study offers evidence that companies in emerging economies, such as South Africa, experience positive financial returns from their engagement in CSR activities.