Balancing bonding and monitoring costs in remuneration governance: evidence from South Africa Open Access

Executive directors’ remuneration (EDR) has been attracting attention worldwide from researchers, standard setters, activists, the media, shareholders, employees and the public (Stathopoulos and Voulgaris, 2016; Van Zyl and Mans-Kemp, 2023). This study relied on empirical evidence from South Africa to improve our understanding of remuneration governance. Although South Africa is known for the quality of its corporate governance codes (the King Codes) as well as for being the first to mandate integrated reporting (Setia et al., 2015), it is also known for the policy of apartheid (between 1948 and 1994) that excluded most of the South African population from economic opportunities, resulting in the highest income inequalities in the world (Mariotti and Fourie, 2014). Post-apartheid South Africa, however, struggles to correct such inequalities, with excessive EDR and growing pay gaps, often blamed as contributing factors (Lemma et al., 2020). Coupled with a concerted effort to improve EDR guidelines in South Africa (which includes voting rights on remuneration to shareholders), it is not surprising that stakeholders (including shareholders) are increasingly publicly criticising and voting against remuneration policies in South Africa (Zondani and Viviers, 2025).

Several theoretical lenses proved helpful in developing this study and interpreting our results. With agency theory (Jensen and Meckling, 1976) that has been at the centre of corporate governance research for the last few decades, optimal contract theory or incentive theory (Bebchuk and Fried, 2003) views EDR as the solution, as opposed to managerial power theory that sees EDR as part of the problem. Supported by the optimal contract theory, corporate governance codes, like the King IV Corporate Governance Code (King IV), recommend that EDR be linked to firm performance as bonding costs (Duh, 2017) to counter goal incongruencies. Despite such guidance, levels of EDR and pay-gaps remains controversial, specifically in developing countries, and often attract media attention, but for the wrong reasons (Faku et al., 2023).

In addition to bonding costs, agency theory also suggests monitoring costs like the disclosure of corporate governance compliance (Duh, 2017) to alleviate managerial power. In South Africa, King IV introduced an innovative three-part remuneration report with a clear focus on transparency [Institute of Directors South Africa (IoDSA), 2016] on 1 November 2016. Although the South African Johannesburg Stock Exchange (JSE) remains the largest stock exchange in Africa, it is important to note that on average 25 firms per year delisted from the JSE between 2015 and 2022 (Larkin, 2022). Costs associated with regulatory requirements are often cited as one of the underlying reasons. In addition to King IV being mandatory to all JSE-listed firms, recent amendments to the South African Companies Act require that firms disclose pay gaps in future. The value of such governance practices is, however, not always clear. Data from the USA, for example, shows an increase in pay gap (CEO salary to median salary) of 60–1 in 1992 when remuneration disclosure requirements were first introduced to a current 200–1 (Cohen, 2024). Others, like Harvey et al. (2020), identified the lack of transparency and accountability as a key driver of widening pay gaps in emerging markets.

Our study was concerned with the interconnectedness of the two governance practices discussed above: EDR linked to performance, bonding costs and the requirements of corporate governance codes, as monitoring costs. Although such governance practices share a common goal, they do have different characteristics, roles and functions (Panayi et al., 2021). Oh et al. (2018) argue that governance practices might have different effects depending on how they are combined, resulting in the development of two alternative hypotheses (Panayi et al., 2021): the substitution hypothesis and the complementarity hypothesis.

Specifically, we examine the moderating effect of remuneration governance disclosures (referred to as DSCORE in the remainder of this article) on the EDR–performance relationship. If the South African EDR–performance relationship becomes weaker in the presence of higher levels of DSCORE, therefore, yielding a negative moderating effect, it would suggest a substitution effect between the two governance mechanisms. This would support the suggestion by Panda and Leepsa (2017) that bonding and monitoring costs move in opposite directions. Possible explanations that support the substitution hypothesis include the cost-benefit trade-off and the unintended consequences of remuneration disclosures, often referred to as the Lake Wobegon Effect (Cohen, 2024). On the other hand, a positive moderating effect of DSCORE would support the complementarity hypothesis. The complementarity hypothesis validates the bundles approach (Schnyder, 2012) which implies synergistic effects among governance mechanisms.

What can be learnt from current empirical evidence? Despite regulatory pressures and increasing stakeholder demands for transparency, the extent and quality of remuneration disclosures vary significantly among firms (Lang, 1998; Watson et al., 2002; Huang et al., 2013; Kang and Nanda, 2016; Orhun, 2019; Ronoowah and Seetanah, 2024). Although there is also no shortage of empirical studies that have examined EDR–performance relationships, empirical evidence remains largely inconclusive in both developed and developing country settings (Tosi et al., 2020; Padia and Callaghan, 2021; Mnyaka-Rulwa and Akande, 2024; Siwendu et al., 2024). Although some studies have explicitly examined the moderating effect of corporate governance on the EDR–pay relationship, results remained mixed and have mostly relied on data from developed countries and single corporate governance proxies. Other possible reasons for inconclusive results include inconsistent proxies used to measure EDR and firm performance, contextual differences in terms of governance settings and sample inclusion and exclusion strategies, as well as inherent methodological differences. Marais and Strydom (2018) document the positive moderating effect of remuneration committee independence on the EDR–performance relationship, but only when performance is measured as total shareholders return, and only for a sample of JSE-listed consumer and technology firms. Another study that used data from JSE-listed firms, Ntim et al. (2019), found both positive and negative moderating effects of governance structures on the EDR–performance relationship.

What can be derived from JSE-listed firms from an emerging market environment? Despite the mandatory nature of the King Codes for JSE-listed firms, research indicates varying levels of remuneration disclosures (Fan et al., 2011; Ntim et al., 2012; Bokpin, 2013; Mans-Kemp and Viviers, 2018). Insights derived from this study not only inform policy decisions within South Africa but also provide valuable guidance for other emerging markets facing similar governance challenges. Furthermore, South Africa’s unique socio-economic context, characterised by significant income inequality (Steenkamp et al., 2019; Lemma et al., 2020) and a history of corporate scandals (Rossouw and Styan, 2018; Maroun and Cerbone, 2020), underscores the importance of effective corporate governance mechanisms. Consequently, stakeholders in South Africa are increasingly demanding greater accountability from firms. This heightened scrutiny creates a fertile ground for exploring how remuneration disclosure can serve as a mechanism to enhance legitimacy and align executive interests with the broader interests of stakeholders (Hassan et al., 2009; Hoang et al., 2020; Bennedsen et al., 2022; Cullen, 2024).

How does this study contribute to the accounting and organisational change literature? While several studies have explored the relationship between EDR and firm performance (Olaniyi and Olayeni, 2020; Bezuidenhout, 2021; Padia and Callaghan, 2021) and the effect of corporate governance thereon (Lee et al., 2008; Jatana, 2023), the unique contribution of this study lies in its focus on the moderating role of DSCORE in a developing country context. Existing literature predominantly examines this relationship in developed economies, often overlooking the interplay between bonding (EDR–performance link) and monitoring (governance disclosure) costs. This study, therefore, contributes in three ways. Firstly, it examines the substitution and complementary effects of bonding and monitoring costs, contributing to the theoretical discourse on corporate governance trade-offs, a perspective underexplored in prior studies. Secondly, it uses advanced analytical methods, including textual analysis (LIWC) and Johnson–Neman (J–N) analysis, to provide a refined understanding of conditions under which governance mechanisms influences the EDR–performance relationship. Thirdly, by leveraging data from South Africa, a country characterised by advanced corporate governance frameworks (e.g., King IV) and one of the world’s highest income inequalities, the study provides unique insights into the role of governance practices in shaping organisational behaviour in emerging markets.

This study aimed to provide valuable insights to policymakers, regulators and corporate governance practitioners. By illuminating the impact of remuneration disclosure on the pay–performance relationship, the findings can inform efforts to enhance transparency, align executive incentives with long-term firm performance and strengthen the overall remuneration governance framework within JSE-listed firms. Specifically, policymakers and regulators can use these insights to refine and enforce disclosure requirements, ensuring that firms provide meaningful and comprehensive information on EDR. Corporate governance practitioners, including boards of directors and remuneration committees, can leverage these findings to design and implement effective remuneration structures that align EDR with firm performance metrics. By linking compensation to long-term value creation and aligning it with shareholder interests, firms can foster sustainable growth and mitigate agency problems. Overall, the outcomes of the study have the potential to catalyse positive changes in remuneration practices, governance standards and performance outcomes among JSE-listed firms, serving as a model for improvement in remuneration governance across other markets facing similar challenges.

King IV consists of 17 principles. Principle 14 states that:

This study was concerned with fair remuneration, that is, EDR linked to performance and transparency, namely, DSCORE. Therefore, this section focuses on these two areas.

The link between EDR and firm performance is central to the debate on whether EDR structures effectively reward top management for enhancing firm performance and thereby increasing shareholder value. The EDR–performance relationship is rooted in agency theory (Jensen and Meckling, 1976), which posits that there is an inherent conflict of interest between principals (shareholders) and agents (executive directors). To align the interests of executives with those of shareholders, remuneration packages are often designed to include performance-based incentives such as annual bonuses and stock options (Wei and Rowley, 2009; Bouteska and Mefteh-Wali, 2021). These mechanisms are intended to motivate executives to make decisions that will enhance firm performance. Empirical studies on the pay–performance relationship within emerging market settings have, however, yielded mixed results (Kang and Nanda, 2017). While some studies report a positive relationship, others report a disconnect. For instance, Unite et al. (2008) found a strong positive link between EDR and firm performance among firms listed on the Philippine Stock Exchange. Similarly, Aslam et al. (2019) identified a positive relationship between EDR and firm performance among non-financial firms listed on the Karachi Stock Exchange. Their findings corroborate those of Lee and Isa (2015) in their study on the Malaysian banking sector, thus, highlighting a clear positive association between EDR and firm performance. Contrary to these results, based on their study of Indian firms, Raithatha and Komera (2016) questioned the efficacy of performance-based EDR in influencing executive behaviour towards stakeholder capitalism. Watto et al. (2023) echoed the same sentiments in their study on Pakistan-listed financial firms, after finding no evidence of a link between EDR and firm performance based on accounting measures.

Several factors can influence the EDR–performance relationship, including the structure of the EDR package, and the overall corporate governance environment. For instance, robust governance mechanisms that enforce transparency and accountability can enhance the effectiveness of the pay-for-performance philosophy. Transparent disclosures, in particular remuneration disclosure as a critical internal governance mechanism, can play a crucial role through moderating the EDR–performance relationship (Clarkson et al., 2011). Transparent disclosures enable stakeholders to scrutinise the alignment between EDR and firm performance more effectively, potentially leading to the design of remuneration contracts that are aligned to stakeholder interests (Hassan et al., 2009; Elmagrhi et al., 2020). Several studies have confirmed that transparent remuneration disclosure enhances the EDR–performance relationship through mitigating agency problems and facilitating alignment of executive incentives with stakeholder interests (Cheung et al., 2010; Hassan and Mohd-Saleh, 2010; Chung et al., 2015; Elmagrhi et al., 2020).

Remuneration disclosure involves publicly revealing information about EDR, serving as a key corporate governance mechanism for promoting transparency and accountability (Isukul and Chizea, 2017; Assenso-Okofo et al., 2021). Corporate governance frameworks, such as King IV, emphasise the importance of remuneration disclosure and provide specific guidelines for disclosing executive remuneration.

Resolving the agency problem, particularly considering managerial power, requires initiatives beyond simple contractual agreements (Choe et al., 2014; Ghrab et al., 2021; Ullah et al., 2021). This is where various corporate governance mechanisms play a vital role. Transparent disclosure enhances a firm’s legitimacy by demonstrating adherence to social and regulatory expectations (Dumay et al., 2015; Masum et al., 2020; Sciulli and Adhariani, 2023), particularly in contexts like the JSE, where stakeholders are vigilant about corporate governance practices (Hoang et al., 2020). Aligning with these expectations helps firms maintain and enhance their legitimacy, securing continued support from investors, regulators and the public (Hermalin and Weisbach, 2012). Integrated reporting on EDR also enhances the credibility of the management team and the board of directors, fostering a favourable view among investors and other stakeholders (Chung et al., 2015). Conversely, opaque or insufficient disclosure can lead to mistrust, negative media attention and potential backlash from shareholders.

Several studies have examined compliance with corporate governance requirements in South Africa and similar emerging market settings (Steenkamp et al., 2019; Dao and Nguyen, 2020; Nsour and Al-Rjoub, 2022; Ronoowah and Seetanah, 2024), with many reporting shortcomings in remuneration disclosures (Steyn and Cairney, 2016; Madlela and Cassim, 2017; Mans-Kemp and Viviers, 2018; Steenkamp et al., 2019). However, none of these studies have explored the relationship between executive remuneration and the corresponding remuneration disclosures, nor the effect of transparent remuneration disclosures as a critical governance mechanism on the EDR-performance link following King IV. The aim of this study was to contribute significantly to filling this gap.

Considering the discussed perspectives, theoretical arguments and empirical evidence; to better understand the interplay and implications of remuneration disclosure on corporate governance practices in South Africa, the following hypotheses were developed:

A sample of 100 JSE-listed firms over a six-year period (2017–2022) was selected. The sample size was determined to ensure adequate industry representation while maintaining the feasibility of detailed data collection. Firms were stratified by industry, and random sampling was used to reflect the industry composition of the JSE. The minimum listing period of three years was applied to mitigate survivorship bias and ensure variability. After accounting for dropped observations where data was unavailable, the analysis was conducted using an unbalanced panel comprising 334 firm-year observations. See Table 1 for a summary of the sample selection process. Data for the study was sourced from the Iress and Bloomberg databases and integrated annual reports.

Data was collected from 2017 to coincide with the King IV implementation date (King IV replaced King III in its entirety). King IV introduced significant changes, particularly for remuneration disclosure. Unlike King III, which followed an “apply or explain” regime, King IV adopted an “apply and explain” framework, mandating remuneration governance transparency in three specific areas: the remuneration background statement, the remuneration policy and the implementation report. This shift required firms to provide detailed, narrative-based disclosures that emphasise the alignment of remuneration practices with corporate strategy and stakeholder interests.

King IV also introduced the principle of fair and responsible remuneration, requiring disclosure of wage gaps and fostering a holistic view of governance that integrates financial and non-financial aspects. These changes represent a notable breakpoint in South Africa’s governance landscape, making 2017 a critical starting point for examining the impact of enhanced remuneration disclosure on firm performance and executive pay.

To test the study hypotheses (as discussed in the literature and hypothesis development section), we specified the following linear function ^[1] in line with Aslam et al. (2019), Elmagrhi et al. (2020) and Ntim et al. (2019):

All variables are defined in Table 2. To further clarify: PERFORM_it representing firm performance, is represented by two proxies, return on assets (ROA) and total shareholders’ return (TSR); DSCORE_it represents the remuneration disclosure score; and CONTROL_it refers to the set of control variables (SIZE, LEV, ISHARE, DSHARE, REMIND and BIG4A).

Given the panel structure of the data, both fixed effects (FE) and random effects (RE) models were considered. The choice between these models was determined using the Hausman test, which compares the consistency and efficiency of the estimators. The best model fit (FE or RE) is reported in all Tables that summarise regression results in this paper (Tables 5 to 8).

King IV (via Principle 14) emphasises that EDR should promote the achievement of outcomes across the short, medium and long term [Institute of Directors South Africa (IoDSA), 2016]. Guided by this principle, this study separately examined short-term incentive remuneration (EDRS), long-term incentive remuneration (EDRL) and a composite measure (EDRI) comprising EDRS and EDRL. Fixed and guaranteed remuneration, which are typically unresponsive to firm performance levels, were excluded from the analysis. Both cash-based and share-based components of EDRS and EDRL were considered. For share-based incentives, only the value of shares vesting during the reporting year was included, ensuring objectivity and practicality. This approach was necessary because not all firms disclosed sufficient information for valuing options. However, all firms provided full disclosure of the value of share-based incentives vesting in each reporting period, enabling consistent and reliable data collection. EDRS and EDRL data were meticulously hand-collected from integrated annual reports.

King IV recommends a three-part disclosure in the remuneration report: the remuneration background statement, the remuneration policy and the implementation report [Institute of Directors South Africa (IoDSA), 2016]. We conducted a content analysis to assess the level of remuneration disclosures in the remuneration report. For practical reasons, we relied on a computerised text analysis method, Linguistic Inquiry and Word Count (LIWC), to measure the level of DSCORE per firm year. While Tausczik and Pennebaker (2010) describe LIWC as a tool capable of detecting psychological meaning in text, this study primarily used LIWC’s word counting capabilities to measure the extent of DSCORE. This distinction is important, as the study’s purpose was not to interpret the psychological meaning of text but to provide a systematic and replicable measure of disclosure levels. For this study, a wordlist ^[2] was constructed based on the remuneration disclosure requirements outlined in King IV. The wordlist includes terms and phrases explicitly mentioned in the King IV guidelines, as well as synonymous terms and phrases commonly used in remuneration discussions. Step-by-step LIWC analysis procedure provides a summary of how LIWC was used to measure DSCORE.

1. Text preparation

   • The integrated annual reports were downloaded in PDF format

   • Remuneration report sections were extracted from the reports

2. Customisation of LIWC dictionary

   • The standard LIWC dictionary was augmented with the self-constructed wordlist derived from king IV remuneration disclosure requirements

   • Inputting of specific terms and phrases into LIWC’s dictionary function, creating a customised category labelled “remuneration disclosure”

3. Text analysis using LIWC

   • The extracted remuneration reports were uploaded into LIWC software

   • LIWC processed each document, identifying and counting the occurrences of the words and phrases from the custom “remuneration disclosure” category

   • For each report, LIWC generated a summary output, which included the percentage of total words that matched the remuneration disclosure wordlist

Note(s):LIWC = Linguistic Enquiry Word Count.

Source(s): Created by authors

DSCORE reflects the proportion of words in the firm’s report that pertain to remuneration disclosure specifications that comply with King IV recommendations:

This percentage score provides a standardised measure of the extent to which each firm discloses information related to remuneration, facilitating comparisons across firms and over time. Our model was based on Cooke’s method, which is a DSCORE unweighted index, implying that all the information in the remuneration report was deemed equally important, hence carrying the same weight (Cooke, 1989). By following this methodology, the study provided a systematic and replicable approach to quantifying remuneration disclosure following King IV guidelines, leveraging the capabilities of LIWC and a carefully constructed wordlist. Both the relevant King IV guidelines and the wordlist are available in the  Appendix.

While LIWC offers systematic and replicable analysis, it has nolimitations. LIWC relies on pre-defined dictionaries to identify and quantify words associated with specific concepts, such as remuneration disclosures in this study. As a dictionary-based tool, LIWC focuses on the volume of words matching the dictionary but does not account for the context or meaning of the text. Consequently, it may miss instances where the quality of disclosure is high but expressed using terms not captured by the dictionary, or it may overestimate disclosures where the relevant terms are used without substantive meaning. This limitation means that LIWC measures the extent, rather than the depth or quality, of remuneration disclosures. While this study acknowledges the potential shortcomings, the findings are interpreted with caution, emphasising the need for future research to integrate more advanced natural language processing techniques, such as sentiment analysis or machine learning models, to capture both the quantity and quality of disclosures comprehensively (Hajek and Henriques, 2017; Liu et al., 2025).

The remuneration reports analysed in this study were sourced from the integrated annual reports of JSE-listed firms, which were typically available in PDF format. Even though LIWC performs optimally with text documents, due to practical reasons, the PDF files were not converted into plain text (txt) format, which requires the use of specialised software. As such, and as presented in Table 1, 169 firm-year observations were excluded because the PDF documents contained embedded images, scanned pages or formatting issues that rendered the text unreadable by LIWC. This limitation is acknowledged as a potential source of bias, as the excluded observations may differ systematically from the analysed sample. Future studies could explore alternative methods, such as manual transcription or the use of more advanced textual extraction tools such as optical character recognition software, to address this issue (Hajek and Henriques, 2017; Liu et al., 2025).

Given their distinct nature and following the recommendation in King III Practice Notes [Institute of Directors South Africa (IoDSA), 2012], this study used one market-based and one accounting-based performance measure. More specifically, return on assets (ROA), which indicates the firm’s efficiency in generating profits from its assets was adopted as the accounting-based measure, while total shareholders’ return (TSR), which measures the total returns received by shareholders through stock price appreciation and dividends over a specific period, was adopted as the market-based measure of firm performance. Although these proxies have been widely used in prior studies, for example, Aslam et al. (2019), Elmagrhi et al. (2020) and Bussin et al. (2023), a literature review by Al-Matari et al. (2014) showed ROA to be the most popular accounting-based performance measure used in corporate governance literature between 2002 and 2012. Furthermore, recent studies, for example, Bruna et al., 2022; Ricca et al., 2023; Siwendu et al., 2024; and Grey et al., 2024 confirm the continued popularity of ROA as an accounting-based measure of firm performance. Industry publications by the Global Equity Organisation (2016) and Ernst and Young (2016) further confirmed the use of share returns by firms to award EDR, as well as the use of a mix of both accounting and market-based measures by firms in awarding EDR. Although such industry publications have also shown the importance of non-financial performance measures, this study did not use such measures for two reasons: the unavailability of data thereon on databases and the diversity of such measures. The exclusion thereof is, however, admitted as a study limitation.

Consistent with prior EDR-performance studies, the current study incorporated several control variables to address potential omitted variable bias. The following control variables were included in the model: firm size, measured by total assets, to control for the scale of operations (Blanes et al., 2020); leverage, to account for the firm’s debt levels (Ortiz-Molina, 2007); firm ownership, which captures the ownership structure of the firm, controlling for its effects in the firm’s decision-making processes (as external governance); and as managerial power (Benamraoui et al., 2019; Farooq et al., 2024). Finally, we also controlled for remuneration committee independence and audit quality. Consistent with the recommendations of Hünermund and Louw (2020), we did not articulate any specific expectations for the control variables or discuss their results (for example, level of significance and direction of relationship).

The empirical results are presented in this section.

Table 3 presents the descriptive statistics for the study. Normality plots and histograms were examined for all variables, and for six variables, the natural logarithm was used to reduce skewness in distributions: EDRS, EDRL, EDRI, ROA, TSR and SIZE. Descriptive statistics for these variables are presented prior to any logarithmic transformations that were used in all further analyses.

Table 3 displays significant cross-sectional variation for all variables, consistent with the sample selection strategy. The table reveals that the mean values of all incentive remuneration components are substantially higher than their respective median values, indicating a positively skewed distribution of EDR. Table 3 further shows that EDRL contributes relatively higher to EDRI compared to EDRS. This stresses the importance of considering the composition of remuneration packages in subsequent analyses. Although untabulated, DSCORE remained relatively stable for all years included in the sample, showing that although firms are maintaining a baseline level of transparency, there is limited evidence of significant improvement in disclosure practices

Table 4 presents the Pearson correlation coefficients among all study variables. Besides the expected high correlation between dependent variables that were used in different regression models, the highest correlation coefficient between all the other variables was 0.593 (between REMIND and SIZE), suggesting the absence of multicollinearity. Also, results confirmed that all variance inflation factor (VIF) values fall below the commonly accepted threshold of 10, further indicating that multicollinearity was not a significant concern in the models.

Table 5 presents regression results for each of the dependent variables examined as proxy for EDR (EDRS, EDRL and EDRI), as well as the two proxies used for firm performance (ROA and TSR), resulting in six different regression models. In all regressions, both the F-test for fixed effects and the Hausman test confirmed that the random effects model showed the best fit compared to the pooled ordinary least squares and fixed effects models that were considered as alternatives.

Although the purpose of this study was not to develop a model that best explains variations in EDR, the $R2$ values, which range between 7% and 11%, are consistent with studies examining the impact of governance mechanisms. For example, Zhang et al. (2020) reported Adjusted $R2$ values ranging from 5.8% to 14.9% in their study of corporate employment decision-making in the US market. The relatively lower $R2$ values in this study may reflect the inherent complexities and heterogeneity of governance mechanisms in developing countries, where institutional frameworks, socio-economic disparities and firm-specific factors introduce additional variability. While the lower $R2$ values limit the explanatory power of the models, the focus of this study remains on the statistical significance and moderating effects of DSCORE. Future studies could enhance the explanatory power by incorporating additional variables, such as non-financial metrics (e.g. ESG measures) or qualitative data to better capture the complexity of governance dynamics.

Inconsistent with Padia and Callaghan (2021) who examined the EDR-performance link following King III (2010–2017), we document a significant positive association between ROA and all three EDR proxies: EDRS, EDRL and EDRI. In support of the H1, our results further show a significant negative independent relationship between DSCORE and both EDRS (Model 1) and EDRI (Model 5) when firm performance is measured by ROA. This suggests that in firms where remuneration practices are more transparently disclosed, executives tend to receive lower EDR. One explanation for our results is that increased transparency may result in greater scrutiny by stakeholders, limiting the potential for excessive or unjustified EDR. Similar results were reported by Elmagrhi et al. (2020) from their UK-based study covering 2008–2013, who concluded that better governed firms on average are more likely to constrain EDR.

Our analysis, however, also reveals the lack of significant independent associations between DSCORE and any form of incentive remuneration when market-based performance metrics are used. The absence of significant relationships with market-based performance metrics suggests that the effect of transparency on executive remuneration may be context-dependent, varying with the type of performance metrics used. This finding implies that the governance role of disclosure may be more effective in certain performance contexts, particularly those that are closely tied to financial outcomes and under the direct influence of management. It is also important to note the significant negative association between ISHARE and EDR in Models (1) to (4), indicating the importance of external governance mechanisms, in additional to internal governance mechanisms (such as our DSCORE) to ensure that EDR is fair and responsible. Another interesting result is the significant positive association between Big4A and EDR in all models reported in Table 5. Elmagrhi et al. (2020) also noted discernible evidence that firms that are audited by a Big-4 audit firm pay significantly higher EDR.

Models (1) and (5) further show the significant negative moderating effect of DSCORE consistent with study’s H2. Such findings, therefore, indicate that as the level of DSCORE increases, the EDR–performance relationship weakens. One explanation for the abovementioned results is that detailed disclosures increase scrutiny from stakeholders, leading to more conservative or risk-averse behaviour by executives to avoid negative perceptions. Thus, executives are motivated to make decisions that protect the firm’s assets and operational stability rather than aggressively pursuing higher returns. On the other hand, increased transparency might align executive actions more closely with long-term sustainability rather than short-term asset utilisation efficiency. Hence, executive directors may prioritise stable, consistent performance over potentially higher but riskier returns, leading to a weaker pay–performance relationship at least in the short term. Overall, these results provide support for the substitution H2, indicating that as suggested by Panda and Leepsa (2017), bonding (EDR linked to performance) and monitoring costs (DSCORE) move in opposite directions. As a result of the costs associated with implementing and disclosing remuneration governance, firms may argue that less transparency is required if EDR is more closely aligned with firm performance. It is also possible that stakeholders (including remuneration activists) may be less likely to critique transparency if EDR is closely aligned with performance. On the other hand, increased levels of DSCORE may result in increased EDR levels (not linked to performance) because of the Lake Wobegon Effect. It should, however, be noted that when adjusted for heteroskedasticity ^[3] the p-value of the interaction term in Model (5) increased from 0.03–0.21.

The regression models were evaluated for overall significance using the Wald chi-squared test, given the use of the RE panel models. The results indicated that all models were statistically significant at the 1% level, confirming that the independent variables collectively explain a significant portion of the variation in EDR. Although the study primarily focuses on the significance of individual coefficients, the model-level significance emphasises the reliability of the estimated relationships.

As opposed to the extant literature that interpreted only the statistical significance of the interaction term (DSCORE * PERFORM), this study aims to contribute by improving our understanding of the moderating effect of DSCORE by conducting a J–N analysis to pinpoint the specific ranges of DSCORE where the moderating effect is significant.

In line with our results reported in Table 5, Figure 1 reveals a significant negative moderating effect on the relationship between EDRS and ROA up to a maximum disclosure score of 3.676, beyond which any further increases in remuneration disclosures have no statistically significant moderating effect.

A closely related result is depicted in Figure 2, where any increase in the level of remuneration disclosure up to a maximum score 3.763 weakens the relationship between EDRI and ROA. It, therefore, appears that on average only firms with below average remuneration governance substitutes recommended governance practices with EDR linked to performance.

Models (2), (4) and (6) in Table 5, on the other hand, suggest that DSCORE has no significant moderating effect on the relationship between EDR and TSR. Although such results do not support the complementarity H3, it provides only weak support for the substitution hypothesis. One possible reason is that TSR, as a market-based measure of firm performance, is influenced by many factors beyond management control, such as market trends, economic conditions and investor sentiments.

These findings suggest that the impact of remuneration disclosure on the perceived effectiveness of TSR in driving EDR is less pronounced. From another perspective, executive directors may already be highly motivated to maximise TSR due to its direct impact on shareholder value and their remuneration. Thus, detailed remuneration disclosure may not substantially change this motivation, as market-driven factors play a significant role in determining TSR.

To ensure the robustness of the findings, separate analyses were conducted. First, the six regression models were re-estimated after excluding the COVID-19 years (2020 and 2021). Although the results (as reported in Table 6) remained consistent with the main results for ROA, in support of study H3, DSCORE positively moderates the relationship between EDR and TSR for both Models 2 (EDRS) and 6 (EDRI) (albeit both at the 10% level). Given the levels of share price volatility during the COVID-19 years, our results, therefore, provide support for both the substitution and complementary hypothesis and suggest that corporate governance may play a different role for different types of performance measures.

Secondly, industry-specific analyses were performed by examining consumer and primary industry firms separately. To allow meaningful comparison, all industries were categorised into two categories: consumer industries and primary industries. Following Nel and Esterhuyse (2019), consumer industries consist of consumer goods, consumer services, telecommunications, health care and financials, while primary industries consist of basic materials, industrials, oil and gas and technology. Due to the familiarity bias among stakeholders and individual investors (De Vries et al., 2017), shareholder activists are presumably more active in companies categorised as consumer industries. It is evident from Table 7 that DSCORE negatively moderates the relationship between performance (ROA) and EDR for the consumer industry. Although this is in line with the main findings, the moderating effect is more significant compared to the full sample (at 1% for the consumer industry) and is also significant for EDRL. As depicted in Table 8, DSCORE does not significantly moderate the relationship in any of the models performed for the primary industry. Investors and other key stakeholders that invest in primary industries may place less emphasis on disclosures, relying instead on industry-specific knowledge or other non-disclosure signals.

Finally, lagged performance variables were considered to address endogeneity concerns, specifically endogeneity caused by reverse causality between EDR and performance. Our untabulated results showed that in line with our main results, lagged DSCORE negatively moderates the relationship between lagged ROA and EDRS, but now at the 1% level (as opposed to the 10% level reported in the main results). Although our results showed that lagged DSCORE does not negatively moderate the relationship between lagged ROA and lagged EDRI (as opposed to the 5% level reported in the main results), when adjusted for heteroskedasticity, DSCORE was also found not to significantly moderate the relationship between ROA and EDRI in the main results. Our untabulated results showed no significant moderating effect for any of the other models using lagged DSCORE and lagged performance variables. In all lagged regressions conducted, Durbin-Watson test statistics were between the acceptable range of 1.5 and 2.5.

This study examined whether transparent remuneration disclosure (DSCORE) moderates the relationship between EDR and firm performance in JSE-listed firms, considering the enhanced remuneration disclosure recommendations introduced by King IV in South Africa. More specifically, it examined whether firms link EDR to firm performance (bonding costs) and excel in DSCORE (monitoring costs) as substitutes or as complements. Although this study has limitations, and the findings reveal somewhat conflicting results, our results provide support for the substitution hypothesis using advanced textual analysis software (LIWC) and the J–N analysis method.

DSCORE was found to have a statistically significant negative moderating effect on the EDR–performance relationship for our accounting-based proxy for performance, namely, return on assets. This negative moderating effect was more pronounced for EDRS and as evident from our J-N analysis on average only prevalent at lower levels of DSCORE. Resource availability and external pressures may explain these findings. Firms that are less experienced in corporate governance, and firms that are more cost-benefit conscious, coupled with less institutional monitoring, appear to invest less in DSCORE. This is specifically pronounced for those firms that have successfully linked EDR to firm performance.

Despite LIWC’s strength in systematic and replicable analysis, its key limitation is reliance on pre-defined dictionaries, which measure word frequency in remuneration disclosures but not their quality or contextual relevance. For instance, it may overlook subtleties in tone or strategic phrasing that contribute to the substantive quality of the disclosure. Future research should consider combining dictionary-based methods with advanced NLP techniques to better capture the richness and meaning of corporate disclosures. Approaches such as sentiment analysis or semantic interpretation could complement LIWC by addressing its limitations, thereby enhancing the robustness of governance research and providing deeper insights into how remuneration disclosures influence organisational practices and stakeholder perceptions.

Despite discretion allowed to governing bodies in where King IV disclosures should be made, this study examined only the integrated annual report. Since empirical evidence exists that firms that rely on their corporate websites as supplementary information source mostly do so with corporate governance-related issues (Nel, 2019), it is proposed that future studies may consider analysis of other sources as well.

Another limitation of this study is a relatively low reported $R2$ that imposed limitations on the interpretation of our results. Given King IV recommendations that firms should balance financial and non-financial performance metrics in creating remuneration packages, future studies could consider the inclusion of, for example, ESG metrics in analysis.

Overall, this study emphasises the complex role of remuneration disclosure in the governance landscape of JSE-listed firms. Beyond the practical implications for policymakers and governing bodies, the findings offer significant opportunities for advancing future research in accounting and organisational change. Specifically, this study emphasises the need to explore the dynamics of governance mechanisms, such as remuneration disclosures, as catalysts for organisational accountability and legitimacy in diverse contexts.

Future research could build on this study by investigating how varying governance frameworks, particularly in emerging economies, shape organisational behaviours and decision-making processes. Additionally, the substitution and complementary effects of governance mechanisms warrant further examination using interdisciplinary approaches, integrating insights from organisational change theories, behavioural economics and institutional theory.

The study also highlights methodological avenues for future research. The use of advanced textual analysis tools like LIWC, combined with robust statistical methods such as J–N analysis, offers a replicable framework for assessing governance impacts. Researchers could extend these methods to other organisational change phenomena, such as sustainability reporting and integrated reporting, to evaluate their broader implications for organisational transformation and performance. By addressing these avenues, future studies can deepen the understanding of governance practices as drivers of organisational change, particularly in contexts with significant socio-economic challenges and evolving regulatory environments.

The authors extend their sincere gratitude to Juan Ontong from the School of Accountancy at Stellenbosch University for his guidance on the use of computerised text analysis methods.