Corporate governance, capital structure, and firm performance: a panel VAR approach

Abstract

This study aims to examine the interrelationships and interdependencies between corporate governance (CG), capital structure (CS), and firm performance (FP) of companies listed on the Stock Exchange of Mauritius from 2009 to 2019 along with a comparison between financial and non-financial firms. A panel vector autoregression (PVAR) approach is used in this study to determine the relationship dynamics between CG, CS and FP. The findings reveal a positive and significant bidirectional association between CS and FP, supporting the trade-off theory. The results also show that CG and FP jointly help to increase CS while CG and CS jointly boost the profitability of firms. A strong bidirectional relationship with varied signs between CG and CS is found only for financial firms. The results of the forecast error variance decomposition analysis support the selection of FP as the most endogenous variable. Robustness tests also support the findings. This study is the first to examine the dynamic and interdependent relationships using a PVAR model between CG, CS and FP that presents new contributions to the existing CG and CS literature with insights from an emerging economy.

Introduction

Corporate governance (CG), capital structure (CS), and firm performance (FP) are three crucial aspects that are linked to each other. Previous studies on the association between CG and CS rely heavily on agency theory to explain a company's financing decisions (Boateng et al. 2017). Both are linked because agency cost is one of the major elements of CS and CG that mitigates agency conflicts. CS is a CG instrument that can assist a company in developing value by preserving CG efficacy (La Rocca 2007). Good CG is commonly acknowledged to improve company performance (Beiner et al. 2004; Black and Kim 2012; Padachi et al. 2017; Sheaba Rani and Adhena 2017; Mansour et al. 2022). However, FP can also influence the level of CG, often measured using the CG disclosure index (CGI) as a proxy for the overall quality of CG in different countries. For instance, profitable organisations are anticipated to have greater compliance and disclosure levels than unprofitable or less profitable organisations to attract new investors and shareholders (Suwaidan et al. 2021). Moreover, on one hand, CS can influence performance (Doan 2020; Amare 2021) but financial performance, on the other hand, may also have an impact on CS (Abdullah and Tursoy 2021). Organisations with better profitability can more easily obtain debt financing, probably at more competitive interest rates than companies with less profitability. Researchers have discovered that one of the most important elements influencing the CS mix is FP (Iyoha and Umoru 2017; Cevheroglu-Acar 2018).

In recent years, there has been a surge in attention paid to the impact of CG on CS and FP and CS on FP, respectively. However, a detailed analysis of the literature reveals significant shortcomings. First, the bidirectional causation between CG, CS and FP has rarely been considered. Second, most prior studies, although taking into account the dynamic nature of financial performance modelling, have largely ignored the issues of endogeneity and reverse causality in the CG, CS and FP nexus. Third, emerging nations such as Mauritius, and more specifically African countries, have distinct economic, institutional, legal, and political settings than developed countries; therefore, the relationships between CG, CS and FP and their reverse causalities may likely differ from those noted in developed economies. Previous studies in Mauritius on CG by Soobaroyen and Mahadeo (2008), McGee (2009), Mahadeo and Soobaroyen (2012) and Mahadeo and Soobaroyen (2016) focus on the level of compliance, whereas Appasamy et al. (2013) and Padachi et al. (2017) quantitatively study the impact of CG on FP using static models and with limited sample size. Prior studies related to CS in Mauritius have focused on the determinants of CS (Fowdar et al. 2009; Odit and Gobardhun 2011; Gourdeale and Polodoo 2016; Omrawoo et al.2017), and so far, only one study has been conducted on the effect of CS on FP by Seetanah et al. (2014), with limited sample size and no CG variables employed as potential determinants. Fifth, no study has been conducted on the impact of CG on CS and vice versa in Mauritius, and most studies (Herlambang et al. 2018; Chow et al. 2018) have used several CG variables as proxies for CG. The use of CGI as a measure of overall CG quality to assess its impact on CS is rare. Finally, previous studies have largely focused on non-financial companies, while financial firms have often been ignored.

For various reasons, Mauritius is an attractive research setting for examining the interrelationships and interdependencies between CG, CS and FP. In Mauritius, the de facto features of the corporate environment are quite different from the CG structure, which is relatively less mature, from those used in developed countries, making Mauritius an interesting case for this study. Additionally, there are major differences between emerging and developed markets in terms of market and knowledge quality, volatility and size (Al-Malkawi 2008). Moreover, the Mauritian capital market has a concentrated ownership structure based on cross-shareholdings and pyramid ownership structures as well as an inactive market for corporate control, that is, takeovers. Managerial entrenchment often results from concentrated ownership structures (Elghuweel et al. 2017). Furthermore, Mauritius is a heavily indebted country with high-leverage enterprises, as debt from banks is favoured over equity and is a relatively inexperienced equity market. In an emerging economy with strong growth prospects, it is critical to investigate the impact of such a high-leverage structure on FP and vice versa. Mauritian enterprises are regarded as small corporations around the world because of their modest size. Finally, as an emerging economy, Mauritius is rapidly evolving and aspires to become a significant foreign direct investment hub by focusing on an innovatively led framework. Mauritius' continuous growth has brought with it several new difficulties and responsibilities, as well as a closer alignment with foreign investors and global stakeholders, all of which need a stronger focus on better CG and optimum CS which improve FP and help to attract investors. As a result, Mauritius emerges as a crucial motivator to conduct a first-hand study on the relationship dynamics between the CG, CS and FP of listed firms as part of realising the country's vision in competing with international competitors.

Consequently, this study aims to add to the existing body of literature by addressing some of the shortcomings of past studies and offering new empirical findings. First, this study offers evidence on the effects of CG, with a single measure of overall CG quality, on CS and vice versa in an emerging economy, Mauritius, where no such research has been conducted. Second, this study provides new evidence on the relationships between CG and FP, CS and FP, and CG and CS with reverse causalities in a small island emerging country like Mauritius, which has different characteristics compared to developed and larger developing/emerging economies. This will be the first attempt, to the authors’ knowledge, to examine the dynamic and interdependent relationships using a PVAR model between CG, CS and FP that presents new contributions to the existing CG and CS literature. Third, this study investigates the interrelationships and interdependencies between CG, CS and FP in a panel vector autoregression (PVAR) framework which accounts for potential dynamic and endogeneity issues and sheds light on reverse causality between these variables. Finally, this study examines any differences in the interrelationships and interdependencies of CG, CS and FP between financial and non-financial firms.

Therefore, this study aims to explore empirically, using a PVAR approach, the following interrelationships between CG, CS and FP, namely, to determine the direction of causation between CS and FP, CG and FP, and CG and CS in a sample consisting of SEMDEX (29) and DEMEX (13) listed firms from 2009 to 2019 in Stock Exchange of Mauritius (SEM), and the results are compared between financial (4) and non-financial firms (38). Dynamic evaluation is established on the completion of the forecast error variance decomposition (FEVD) and impulse response functions (IRFs). Different ordering of the variables and alternative methods of estimating the PVAR model, that is, XTVAR for robustness checks, are examined.

The remainder of this paper is organised as follows. The theoretical and empirical literature on the associations between CG, CS and FP is presented in “Literature review”. “Research design” describes the data, variables and methodologies of this study. “Empirical results and discussion” discusses the findings. “Robustness analysis” determines the robustness of the findings. The summary and conclusion of this study are presented in “Summary and conclusion”.

Literature review

CG entails mechanisms to ensure that lenders of capital to firms will receive a return on their investment (Shleifer and Vishny 1997). In a CG structure, a company’s timely decision-making policies and practices regulate the obligations and rights of its diverse stakeholders. Agency theory presents board members with professional expertise to meet these requirements. This board also decides on the best mix of debt and equity for a firm’s future performance. The pecking order theory (POT) and trade-of-theory (TOT) of CS provide managers with guidelines in this context. Based on these three prominent theories, this section illustrates the interrelationships between CG, CS and FP.

Theoretical literature

Agency theory

Agency theory underpins the practice of CG. The primary tenet of this theory is that there is a working relationship in the form of a cooperation contract between the party providing the authority (the principal), that is the investor, and the party obtaining the authority (agency), namely the manager. Due to the separation of corporate ownership and control, agency conflicts develop. In an agency relationship, it is natural to anticipate that the agent (manager) will make decisions that are detrimental to the interests of the principal (owner) if economic agents are utility maximisers (Jensen and Meckling 1976). The amount of resources under the manager’s control is just one factor that influences how agency costs affect the composition of the CS. According to Jensen (1986), managers may take on debt to increase the resources under their control, which can result in debt agency costs like bankruptcy costs. CG stands out as a tool that facilitates the alignment of interests between agent and principal in this context. There are grounds to believe that CG and CS are related (Borges Júnior, 2022). This is based on the idea that agency conflict is influenced by CG mechanisms, and that these mechanisms are linked to choices concerning the composition of a firm's funding sources. Moreover, the agency theory proposes that CG and financial decisions have an impact on company value and CS must be viewed as a device that can intervene and drive governance structures within the business, and hence FP (Bashir et al. 2020).

According to agency theory, organisations voluntarily reveal additional information to reduce agency conflicts and the costs that arise from the conflict between managers and shareholders (Lambert 2001; Alves et al. 2012; Ntim and Soobaroyen 2013). As a result, increased mandatory and voluntary disclosures on CG may reduce information asymmetry between agents and owners, allowing shareholders to better supervise the management's conduct (Beekes et al. 2016).

Trade-off theory

The trade-off theory (TOT) accounts for the effects of taxes and the costs of bankruptcy. The theory assumes that firms trade-off the benefits of debt financing (favourable corporate tax structure) against increased interest rates and bankruptcy costs to find an optimal CS—the mixer that maximises a firm's worth. The TOT predicts a positive relationship, because when a company is profitable, it may take on more debt, resulting in larger interest payments that are deducted from taxes (Ponce et al. 2019).

Pecking order theory

Majluf and Myers (1984), in contrast to the TOT, developed the pecking order theory (POT), which implies that there is no optimal CS. Instead, the theory proposes that firms have a preferred funding hierarchy. According to Myers (1984), organisations with high levels of profitability have low levels of debt because they have a large number of internal sources of funding. Because POT predicts that corporations use their resources rather than borrow them, it expects a negative relationship. The validity of the POT has been proved in numerous empirical research (Acaravci 2015; Paredes Gómez et al. 2016).

Empirical literature

Capital structure and firm performance

Causal effect of capital structure on firm performance

It is expected that increasing financial leverage will strengthen management, lower information costs, and reduce inefficiencies, all of which will improve FP (Jensen 1986; Jensen and Meckling 1976). Financing decisions affect the cost of capital, allowing businesses to optimise their financial performance (Majluf and Myers 1984; Abdullah and Tursoy 2019). Empirical research demonstrates that FP can be influenced by the relative usage of both capital sources i.e., the mixture of debt and equity (Saona and San Martín 2018; Abdullah and Tursoy 2019). According to theoretical models, the link between CS and FP is unclear (Miglo 2016). Only a few empirical studies have explored the performance effect of leverage and the results vary. A few studies find that leverage is positively linked to FP, such as Vijayakumaran (2018) in China and Amare (2021) in Ethiopia, while others find it to be negatively linked, such as Li et al. (2018) in European SMEs from Austria, Belgium, Finland, France, Germany, Italy, Portugal, Spain, Sweden, and the UK; and Doorasamy (2021) in East Africa (Kenya, Tanzania, and Uganda). In Mauritius, Seetanah et al. (2014) find a negative impact of CS on the FP of listed Mauritian firms for the period 2005–2011. Abata et al. (2017) report mixed results in another study conducted in South Africa.

Causal effect of firm performance on capital structure

CS may influence FP but the latter, on the other hand, may also have an impact on a company’s CS (Abdullah and Tursoy 2021). The logic of the reverse causal relationship between performance and leverage can also be explained using TOT and POT. Researchers have discovered that one of the most important variables influencing the CS mix is FP (Iyoha and Umoru 2017; Cevheroglu-Acar 2018 Koralun-Bereżnicka 2018). This argument may be explained by the TOT, which states that profitable businesses have lower bankruptcy costs and are, hence, more inclined to borrow (Fama and French 2002). Moreover, high-profit businesses are prone to take on more debt to reap tax benefits (Frank and Goyal 2009). Therefore, FP may favourably influence CS. Previous empirical research findings support the TOT’s claim (Ajibola et al. 2018; Angkasajaya and Mahadwartha 2020; Amare 2021). POT contends that profitable businesses are more likely to rely on the generated surplus to fund their assets rather than external sources (Myers 1984). Consequently, profitability is assumed to have a negative effect on leverage, keeping the investment level stable backed by empirical studies such as Jarallah et al. (2019) in Japan and Doan (2020) in Vietnam.

Reverse causality between capital structure and firm performance

Previous research has investigated the relationship between leverage and FP but fails to account for the reverse causality of CS on FP, and a simultaneous-equations bias may emerge (Iyoha, and Umoru 2017). From 2002 to 2012, Jouida (2018) studies the dynamic relationship between CS, diversification, and FP for 412 financial companies in France using a PVAR model and observes bidirectional causation between CS and FP after controlling for individual fixed factors. She and Guo (2018) examine a sample of 49 global e-commerce businesses from 2012 to 2016 and find a negative reverse causality between FP and CS that is in line with the POT, which nonetheless shifts when the quantity of debt grows. Abdullah and Tursoy (2021) examine the reverse causality between FP and CS of listed German non-financial firms from 1993 to 2016 and find that FP and CS can positively influence each other. Adhari and Viverita (2015) investigate the reverse causality between CS and FP of 215 firms in Indonesia, Malaysia and Singapore from 2008 to 2011 and observe that CS and FP can positively affect each other.

Corporate governance and firm performance

Causal effect of corporate governance on firm performance

CG mechanisms may improve FP, amongst others, through better monitoring resulting in managers investing value maximising projects, lesser wastage of resources in unproductive activities and enhanced protection of investors implying a lower risk of losing their assets with acceptance of lower investment return triggering a lower cost of capital for firms. According to agency theory, there is a positive correlation between CG ratings and FP (Jensen and Meckling 1976). The implementation of appropriate CG mechanisms, as well as voluntary disclosure, will result in a net reduction in agency costs and an increase in FP (Fama and Jensen 1983; Siddiqui et al. 2013). CG disclosure is a critical tool for ensuring that firms’ CG practices are held within the bounds of law in terms of openness and accountability (Isukul and Chizea 2017).

A good CG is commonly acknowledged to improve FP (Padachi et al. 2017; Bhatt and Bhatt 2017; Sheaba Rani and Adhena 2017; Mansour et al. 2022). Other researchers, such as Rajput and Joshi (2014) in India and Adegboye et al. (2019) in Nigeria, find a negative connection between CG and FP, or no relationship by Hassouna et al. (2017) in Egypt, and Braendle (2019) in Austria, or find mixed results by Tariq et al. (2018) in Pakistan, Shao (2018) in China, Griffin et al. (2018) in various countries, and Dao and Nguyen (2020) in Vietnam. In Mauritius, Appasamy et al.( 2013) show that there is a relationship between CG and FP in the insurance sector from 2009 to 2011, whereas Padachi et al. (2017) find a significant positive relationship between the CG and FP of 36 listed firms from 2010 to 2014.

Causal effect of firm performance on corporate governance

Profitable firms have more financial resources to sustain the increased administrative costs in meeting compliance and enhancing their CG level as compared to less profitable firms. Moreover, FP is regarded as an essential determinant of the level of CG through enhanced compliance with the code of CG and disclosure to stakeholders. Most disclosure research indicates a positive link between corporate profitability and CG disclosures (Elfeky 2017; Cunha and Rodrigues 2018). Agency theory contends that high-profit corporate executives reveal specific information to gain individual advantages, justify their salary packages, improve their reputation in the business market, and reinforce their position (Alnabsha et al. 2018). Moreover, profitable organisations are anticipated to have greater compliance/disclosure levels than unprofitable or less profitable organisations to attract new investors and shareholders (Suwaidan et al. 2021). However, according to Ben-Amar and Boujenoui (2007), even those with weak financial performance have strong incentives to do so to attract investments and improve their financial ratios. Furthermore, increased information disclosure may be linked to lower profit levels, because corporations’ legal liability, if any, is lowered if they share unfavourable information or ‘bad news’ about themselves (Skinner 1994). This implies that a negative association may also exist between profitability and corporate disclosure (Zeghal and Moussa 2015; Suwaidan et al. 2021). Although previous research has examined the association between CG and FP, the bidirectional causation between these two variables is seldom considered which may result in simultaneous-equation bias.

Reverse causality between corporate governance and firm performance

Love (2011) observes that some prior studies argue that causality goes from governance to performance but others argue the opposite that causality runs in a reverse direction from performance to governance. There are various reasons to believe that causality can truly happen from valuation to governance. On one hand, organisations with superior operating results or greater market values may decide to adopt better governance methods, which will result in reverse causality. On the other hand, companies with poor performance prefer to adopt additional anti-takeover clauses, which are linked to poorer governance. As an alternative, businesses may embrace stronger governance processes as a predictor of future success or as a means of enforcing insiders' adherence to ethical behaviour. The signaling role of governance will be significant for share prices in this situation rather than governance itself. Reverse causation may also occur through institutional or international investors who are more inclined to companies with higher market values, which may also result in better governance practices.

Lamiri et al. (2008) examine the reverse causality between different board characteristics and FP of a panel of 36 listed Tunisian firms between 2004 and 2006 and their findings conclude that board influences FP and firms change their board structure in response to FP. Perez de Toledo (2011) assesses the relationship between the quality of CG proxied by a CGI and the market value of 106 listed Spanish firms from 2005 to 2007 and shows that CG positively impacts firm value but there is no proof of reverse causality, i.e. firm value influencing CG. Ingriyani and Chalid (2021) examine 51 listed Indonesian manufacturing firms between 2014 and 2018 and conclude that executive compensation, CG and FP are related to each other. They find that CG has a positive effect on FP and that greater FP tends to decrease the number of board of directors and the supervisory function of the commissioners but increase the proportion of independent directors.

Corporate governance and capital structure

Causal effect of corporate governance on capital structure

Managers' choices of CS are among the most important business policy decisions they make (Boateng et al. 2017). This is because leverage decisions are subject to agency problems and have an impact on a firm's riskiness and performance (Jensen and Meckling 1976). Jensen and Meckling (1976) propose that the separation of principal and agent roles in businesses causes conflicts of interest (agency costs) between shareholders and management, leading to the idea of CG. This is where the two notions of CG and CS come together. According to agency theory, managers in low-CG practice organisations are more likely to experience agency problems, therefore, they will be tempted to use sub-optimal leverage to take advantage of free cash flow. Higher levels of leverage have been considered as a good substitute for weaker governance practices (Mwambuli 2019). In this situation, leverage and governance quality are inversely associated, with companies with low-CG practice needing to use more leverage to minimise agency costs and align firm managers' interests with those of shareholders.

Several studies have shown some evidence for the above assertion, indicating that CG frameworks have an important impact on listed companies’ leverage decisions (Morellec et al. 2012). For example, using a survey-based CG index Haque et al. (2011) investigate the link between CG and the leverage pattern of listed non-financial firms in Bangladesh. They discover that firm-level governance quality has a significant impact on a company's leverage, with weakly governed businesses having a greater degree of debt financing. Mwambuli (2019) investigates the role of CG, measured by a CGI, on CS of 32 non-financial listed firms in the East African region from 2006 to 2015 and finds a significant negative effect of CG on CS decisions. Most studies (Herlambang et al. 2018; Chow et al. 2018) use several CG variables as a proxy for CG, and the use of CGI as a measure of overall CG quality to assess their impact on CS is rare.

Causal effect of capital structure on corporate governance

CS can influence CG levels through enhanced compliance with the CG Code and more corporate disclosures. According to Jensen and Meckling (1976) and Masum et al. (2020), firms with high debt are prone to report additional information to satisfy the requests of external capital providers and alleviate borrowers’ concerns about the possibility of transferring resources from debt holders to managers and shareholders. Again, agency theory shows a robust correlation between a company’s CS and disclosure (Jensen and Meckling 1976), because the existence of debt holders in a company’s leverage (particularly in highly geared businesses) intensifies agency problems (and hence increases monitoring costs), which aims to decrease these costs by revealing additional information in their annual reports. These companies should increase their disclosure levels to restore investor and creditor confidence and as a result, minimise the impact of bankruptcy risk.

A substantial positive link between CS and CG disclosure has been observed by Al-Moataz and Hussainey (2013) in Saudi Arabia and Elfeky (2017) in Egypt, and an insignificant positive impact by Zeghal and Moussa (2015) in several countries, Elgattani and Hussainey (2020) in eight countries (Bahrain, Syria, Qatar, Sudan, Jordan, Palestine, Oman, and Mauritius), and Suwaidan et al. (2021) in Jordan. However, a significant negative relationship is found in some studies by Mallin and Ow-yong (2009) in the United Kingdom and Cunha and Rodrigues (2018) in Portugal, but such a relationship is found to be insignificant by Alves et al.(2012) in Portugal and Spain, and Allegrini and Greco (2013) in Italy.

Reverse causality between corporate governance and capital structure

CS affects CG and vice versa. This is true regardless of whether management chooses to use debt as a source of funding to minimise issues with information asymmetry and transaction, increasing the efficiency of its firm governance decisions, or whether the growth in the debt level is required by the stockholders as a tool to discipline behaviour and ensure effective CG.

On one hand, a change in how debt and equity are managed affects CG practices by changing the structure of incentives and management control. If through the mixture of debt and equity, diverse types of investors all converge within the company, where they have different kinds of impact on governance decisions, then managers will typically have preferences when deciding how one of these categories will prevail when defining the company’s CS. More crucially, it is possible to significantly improve CG efficiency through the thoughtful design of debt contracts and equity.

On the other hand, CG also affects CS decisions. Myers (1984) and Majluf and Myers (1984) demonstrate how management makes decisions about a firm's financing following an order of preference; in this case, if the manager selects the financing resources, it can be assumed that the latter is avoiding a decrease in its ability to make decisions by agreeing to the discipline that debt represents. Finance from internal sources enables managers to keep outside parties out of their decision-making processes. Management can prevent outside influences from influencing their decision-making by financing internal resources. De Jong (2002) describes how managers in the Netherlands attempt to avoid utilising debt so that their ability to make decisions is unchecked, while Zwiebel (1996) observes that managers are forced to issue debt through other governance mechanisms since they are unable to freely embrace the “discipline” of debt (cited in La Rocca 2007). Jensen (1986) stated that decisions to increase corporate debt are voluntarily undertaken by management when it aims to ‘‘reassure’’ stakeholders that its governance decisions are ‘‘proper’’. Empirical studies on the bidirectional relationship between CG and CS are scarce.

Corporate governance, capital structure and firm performance

Previous research has investigated the relationships between CG, CS and FP but such research (Roy and Pal 2017; Nawaz K. and Nawaz A. 2019; Shahzad et al. 2022) has analysed each association separately, in one direction, and using various mechanisms of CG.

To the best of our knowledge, no study has investigated the interrelationship and interdependence between these three variables simultaneously, and more so, using a single composite measure of CG level.

The impacts of CG mechanisms and leverage on FP from previous studies have mixed results, and it is interesting to investigate the simultaneous interrelationships and interdependencies between these three variables in a unique economic, political and social contexts of a small and emerging economy like Mauritius, considered as a reference for the main economic aspects (including good governance) in the African region.

Research design

This section discusses the current study’s research design and philosophy, disclosure sources, CGI measurement, data collection, sample selection, PVAR models and the statistical tests that are employed.

Research questions and conceptual framework

The study’s objectives are to analyse:

1. The interrelationships and interdependencies between CG, CS and FP of listed Mauritian firms, and

2. Any differences in the magnitude and impact of their interrelationships and interdependencies between Mauritian financial and non-financial listed firms.

Timeframe and statistical analysis model

A sample of firms listed on the SEM from 2009 to 2019 is examined. The PVAR approach is used to capture the multiple variables involved in the sample, according to the applicable literature discussed in “Literature review”. The STATA 16 software is used to analyse the data to obtain descriptive statistics and the PVAR model (Fig. 1).

Fig. 1.

Dynamic interrelationships and interdependencies between corporate governance, capital structure, and firm performance

Research and sampling design

This study applies the balanced panel data method to examine a sample of companies listed on the SEM from 2009 to 2019. The research data are collected manually, comprising four financial (excluding banks because of the difference in their CG disclosure requirements) and 38 non-financial companies listed on both SEMDEX and DEMEX of SEM. Annual reports before 2009 are unavailable for all 42 firms to have a balanced panel. The years 2020 and 2021 are excluded because of the worldwide economic crisis resulting from the COVID-19 pandemic which will not reflect the true financial performance of the selected listed companies.

Description of the corporate governance disclosure index

The CGI is measured as the ratio of compliance with each of the CG practices of all 42 companies in the sample selected, which consist of six components/sub-indices of CG. The annual reports from 2009 to 2019 are used to determine whether each of the 102 governance provisions recommended in the checklist is true for that company, as per the Mauritian Code of CG. A ‘yes’ response in form of compliance to a respective governance practice is given a value of one and a ‘no’ response in form of non-compliance is given a value of zero. The CGI is calculated by adding these values to each company annually. Given that when the index has a large number of items and both weighted and unweighted indices’ scores produce similar results (Chow and Wong-Boren, 1987; Sharma, 2014), an unweighted index is used to evaluate disclosure levels as per previous studies (Cunha and Rodrigues 2018; Masum et al. 2020). Furthermore, the unweighted approach gives each disclosure item in the annual report equal weighting and is best suited to resolve the problem of subjectivity bias (Healy and Palepu 2001). Cronbach’s alpha test of reliability of the above six sub-indices forming the CGI has a score of 0.866, which shows that the sub-indices are reliable indicators for measuring the extent of CG.

Panel data vector autoregression (PVAR) models

A summary of the variables’ descriptions and measurements based on previous literature and utilised in this study is shown in Table 1.

Table 1.

Description of variables

Variables	Description	Details	Sources
ROE	Firm performance	Net profit after tax to total shareholders’ equity	Bhatt and Bhatt (2017) Dao and Nguyen (2020)
CGI	Overall corporate governance disclosure index	The ratio of overall corporate governance disclosure items	(Cunha and Rodrigues (2018) Masum et al. (2020)
CS	Firm leverage	The ratio of total debt to book value of equity	Renders et al. (2010)

As per the definition in prior studies, three variables CGI, CS and ROE are used in this study. Bhagat et al. (2008) argue that developing a CGI is beneficial, because it incorporates the different components of a company's governance structure into a single number that can be utilised to assess governance efficiency. The interrelationships/interdependencies between CG, CS, and FP are investigated using the PVAR methodology. The PVAR method is particularly well suited to this research because it strives to model the evolution of a system of interest variables—CG, CS and FP—in a set of firms that differ significantly in various dimensions such as financial, non-financial, size, age, industry type, and listing status. The PVAR approach is a mixed econometric methodology that blends the standard VAR method, in which all variables in the model are considered endogenous, with the panel data technique, which permits the explicit insertion of a fixed effect in the structure (Shank and Vianna 2016). PVAR accounts for both static and dynamic interdependencies (Canova and Ciccarelli, 2013). This setup also enables us to investigate the Impulse Response Functions (IRFs) of various shocks and how they influence other imbalances. In this study, the model in the PVAR approach is limited to only endogeneous variables and control variables are excluded in line with prior studies (Shank and Vianna 2016; Comunale 2017; Jouida 2018; Traoré 2018; Apostolakis and Papadopoulos 2019; Trofimov 2021).

In a generalised method of moments (GMM) framework, the PVAR model selection, estimation, and inference are used, as proposed by Abrigo and Love (2016). Considering Abrigo and Love (2016), the following k-variable homogeneous panel VAR of order p, with panel-specific fixed effects characterised by the following system of linear equations:

$$Yit=\mathbf{Y}it-1\mathbf{A}1+\mathbf{Y}it-2\mathbf{A}2+\cdots \cdots ·+\mathbf{Y}it-p+1\mathbf{A}p-1+\mathbf{Y}it-p\mathbf{A}p+\mathbf{X}it\mathbf{B}+\mathbf{u}i+\mathbf{e}it$$

$$\text{i}\in \{1,2,...,42\},\text{t}\in \{2009,2010,...,2019\},$$ 1

where Y is a (1 × k) vector of dependent variables, X is a (1 × l) vector of exogenous covariates, and ui and e are (1 × k) vectors of dependent variable-specific panel fixed-effects and idiosyncratic errors, respectively. The (k × k) matrices A1, A2, …, A⁠p − 1, A⁠p, and the (l  × k) matrix B are the parameters to be estimated. It is presumed that the innovations have the following attributes: E (eit) = 0, E (e’iteit) = Σ, and E (e’iteis) = 0, and for all t > s.

According to Abrigo and Love (2016), the PVAR describes in Eq. (1) has problems with dynamic interdependencies and cross-sectional heterogeneities. Consequently, the fixed-effects variable μi is the only variable that captures the heterogeneity between various units. Because the individual effect term Ai is linked to the error term in dynamic panels, the ordinary least-squares (OLS) method cannot be used, because estimation by OLS leads to biased coefficients (Jouida 2018). To address this issue, PVAR models are determined using an equation estimated with the GMM, in an 11 year study of 42 listed Mauritian companies. This method has numerous benefits. Arellano-Bond is used to generate unbiased fixed-effects average coefficients for the short panels (N > T). As a result, the findings control for all time-invariant characteristics that are often addressed in empirical research. On the left side of each equation is the first difference of an endogenous variable and on the right side is the p lagged first difference of all endogenous variables.

Panel unit root test

The Augmented Dickey and Fuller (ADF) (1981), Levin, Lin, and Chu (LLC) (2002) and Im, Pesaran, and Shin (IPS) (2003) tests for data stationarity reveal that all the series of variables used in this model are stationary at level, because the p-values are below the 5% level. Given the absence of a unit root, it is possible to investigate the causation between the three variables.

Selection order criteria

For the study of the PVAR models, the steps of Abrigo and Love (2016), who present a package of controls on STATA, are followed. The optimal lag order in the panel VAR specification and moment condition is used to perform the PVAR analysis. Since the first-order panel VAR (one lag) has the smallest MBIC (Bayesian information criteria, Schwarz, 1978), MAIC (Akaike information criteria, Akaike, 1969), and MQIC (Hannan-Quinn information criteria, Hannan and Quinn, 1979), the data in Table 2 support this option.

Table 2.

Selection order Criteria Sample: 2013 – 2018 No. of obs = 252 No. of panels = 42

lag	CD	J	J p-value	MBIC	MAIC	MQIC
1	0.999	23.778	0.643	− 125.516	− 30.222	− 68.566
2	0.999	15.799	0.607	− 83.731	− 20.201	− 45.764
3	0.999	10.868	0.285	− 38.897	− 7.132	− 19.914
4	0.999	–	–	–	–	–

J denotes Hansen’s (1982) J statistic of overidentifying restrictions; maximum likelihood-based model-selection criteria (M), namely, the Akaike information criteria (AIC)(Akaike 1969), the Bayesian information criteria (BIC)(Schwarz 1978; Rissanen 1978; Akaike 1977), and the Hannan–Quinn information criteria (HQIC)(Hannan and Quinn 1979)

Empirical results and discussion

Descriptive statistics

Table 3 shows the normal statistical characteristics of the main and other variables, including the mean, minimum, maximum, and standard deviation, of the sample of 42 listed companies.

Table 3.

Descriptive statistics

Variables	Observations	Mean	Standard deviation	Minimum	Maximum
ROE	462	0.222	1.342	− 2.070	22.651
CGI	462	0.819	0.119	0.216	0.971
CS	462	1.016	1.141	0.002	8.368
Age	462	17.271	7.728	1	35
Size	462	8907.019	12,202.82	39.5	68,984.17
Listing	462	0.68	0.467	0	1

CGI denotes the ratio of the overall corporate governance disclosure items, CS denotes firm leverage, and ROE denotes firm profitability, AGE denotes firm age, SIZE denotes firm size by total assets, and LISTING denotes a firm’s listing status with SEMDEX firms = 0 and DEMEX firms = 1

As shown in Table 3, the average return on equity ratio (ROE), a proxy for FP, is 22.2%. The CGI, as a proxy for CG, for all 42 listed firms ranged from 21.6 to 97.1%, with a mean of 81.9% and a standard deviation of 11.9%. The mean value of the CGI of the 38 non-financial firms is 81%, while that of the four financial firms is 90.6%. The results indicate that listed Mauritian firms are highly compliant with the Mauritian CG Code, with financial firms being more compliant than non-financial firms. The average CS levels of Mauritian companies are almost 101.6% of their equity (financial firms: 131.2% and non-financial firms: 98.5%) demonstrating that they are highly leveraged firms. ROE is on average 22.2% for the whole sample and 7.1% and 165.4% for non-financial and financial firms, respectively.

Correlation analysis

The correlation analysis between the three variables in Table 4 and the variance inflation factors (VIF) for both CS and CGI is 1.00 which implies that there is no evidence of multicollinearity.

Table 4.

Pearson’s and Spearman’s correlation matrices of the dependent and independent variables for the whole sample

(obs = 462)	ROE	CGI	CS
ROE	1.000
CGI	0.081*	1.000
CS	0.152***	0.061	1.000

CGI denotes the ratio of the overall corporate governance disclosure items, CS denotes firm leverage, and ROE denotes firm profitability

*, **, *** indicate significance at the 10, 5, and 1% levels, respectively

PVAR results and discussion

PVAR results

Table 5 shows the coefficients from the PVAR model by using ‘GMM-style’ instruments for CG, CS and FP. In this model, all variables are at a level and considered endogenous. Table 5 indicates that, in the CG equation, CG responds positively and significantly to its own lag for all firms, including both non-financial and financial firms. The CG responds positively and significantly to the lag of CS only for financial firms. CS has a positive impact on CG because high levels of leverage of financial firms raise agency costs, which encourages managers to reveal more information in an attempt to lower these costs. Moreover, financial companies with high debt ratios are exposed to significant monitoring costs or specific restrictive covenants, which force them to reveal more information and also reassure their lenders to extend or lengthen the debt contract time. It can also be noted that CG responds negatively and substantially to the lag in FP, except for non-financial firms, where it is positive but insignificant. The negative impact of FP on CG is consistent with the findings of Zeghal and Moussa (2015) and Suwaidan et al. (2021), because even Mauritian firms with weak financial performance have strong motivations for CG disclosures to attract investment and enhance their financial ratios. The positive effect of FP on CG disclosure for non-financial firms, supporting the agency theory, is because they may be aiming to attract new investors and shareholders.

Table 5.

Panel VAR estimates

	All firms		Non-financial firms		Financial firms
	Coefficients	P value	Coefficients	P value	Coefficients	P value
Dependent variable: CGI
CGI L1	0.968***	0.000	0.968***	0.000	0.996***	0.000
CSL1	0.003	0.301	0.006	0.350	0.003***	0.000
ROE L1	− 0.001*	0.096	0.010	0.417	− 0.0003***	0.000
Dependent variable: CS
CGI L1	1.096***	0.000	1.184***	0.000	− 1.691***	0.000
CSL1	0.443***	0.000	0.681***	0.000	0.651***	0.000
ROE L1	0.093***	0.000	1.258***	0.002	0.034***	0.000
Dependent variable: ROE
CGI L1	− 0.222	0.126	− 0.410***	0.000	0.906	0.400
CS L1	0.152***	0.001	0.041*	0.072	1.050***	0.000
ROE L1	0.543***	0.000	0.137**	0.048	0.449***	0.000
Number of observations	378		342		36
Number of groups/firms	42		38		4

CGI denotes the ratio of overall corporate governance disclosure items, CS denotes firm leverage, ROE denotes firm profitability, CGIL1 denotes one lag in the ratio of overall corporate governance disclosure items, CSL1 denotes one lag in firm leverage, and ROEL1 denotes one lag in firm profitability

*, **, *** indicate significance at the 10, 5, and 1% levels, respectively

In the CS equation, all coefficients are significant, as indicated by their values at the 1% level. CS responds positively and significantly for the whole sample and non-financial firms to the lag of CG but negatively and significantly for financial firms. The positive effect of CG on CS implies better governed non-financial firms are in a better position to obtain more debt. As regards financial firms, CG impacts negatively on CS because firms with low-CG practice need to use more leverage to minimise agency costs and align firm managers' interests with those of shareholders. The CS responds positively and substantially to its own lag in all three cases. The lag in FP has a substantial positive effect on CS for the whole sample and two sub-samples. This implies that profitable Mauritian firms are more inclined to borrow more because of low bankruptcy costs and reap more tax benefits and support TOT.

Regarding the FP equation, FP responds significantly and positively to its own lag and the lag of CS for the whole sample and two sub-samples. CS has a positive impact on FP, because an increase in firm leverage is expected to decrease information costs, lessen inefficiency, strengthen management and thus enhance FP. Nevertheless, the response of FP to CG differs; it is negatively connected to the lag of CG for the whole sample and non-financial firms which contradicts the findings of Padachi et al.(2017). This significant negative relationship for non-financial firms can be the result of the prevalence of highly concentrated ownership among the listed Mauritian companies which often leads to managerial entrenchment (Elghuweel et al. 2017) that can have adverse impacts on management behaviour and incentives. Another possible reason for the negative impact of CG on FP can be that directors of the board and its board committees may not be having a total commitment to the cause of the company because of other commitments which limit their contribution. For financial firms, FP responds positively but insignificantly to CG lag.

There is no indication of any reverse causation between CG and FP for all firms, including non-financial and financial firms. Causality runs negatively and significantly in just one direction for the whole sample and financial firms—from FP to CG and not vice versa. For non-financial companies, however, causality only flows negatively and significantly in one direction, from CG to FP which is in line with the studies by Rajput and Joshi (2014) and Adegboye et al. (2019) and not vice versa. Additionally, CS has a positive and significant impact on FP and vice versa, demonstrating a strong bidirectional relationship between CS and FP for all firms, including non-financial and financial firms, and supporting the agency cost hypothesis and CS trade-off theory, which contradicts the findings of Jouida (2018) with varying relationships but consistent with the findings of Abdullah and Tursoy (2021) and Adhari and Viverita (2015). Moreover, the analysis reveals a unidirectional relationship between CG and CS for the whole sample and non-financial firms, because CG has a positive significant influence on CS, implying that better governed firms have more debt, but not vice versa. For financial firms, however, substantial bidirectional correlations between CG and CS have been established with varied signs. For instance, CG has a major negative effect on CS, meaning that better governed financial firms have less leverage (Haque et al. 2011; Mwambuli 2019), and CS has a considerable positive impact on CG in line with the findings of Al-Moataz and Hussainey (2013) and Elfeky (2017) and which supports the agency theory.

Overall, the findings reveal that better governed firms can increase their leverage to boost FP which in turn helps to obtain further debt. Therefore, in an emerging economy with steady economic growth and growth opportunities, profitable and better governed firms are prone to finance their investments with additional debt that significantly improves their profitability. However, as a firm increases its debt holdings, the probability of financial distress increases and creditors are less likely to re-finance or renegotiate. In this context, given the importance of debts and good CG, policymakers can consider measures relating to monetary (interest rates) and fiscal (tax) policies to make loans from financial institutions more accessible and attractive/competitive and to further improve CG standards that jointly help to improve FP. However, policymakers can also consider policy measures for steady positive economic growth because any decline may increase bankruptcy risks with serious repercussions due to the presence of highly leveraged companies.

Granger causality

In Table 6, the null hypothesis that all lags of all variables can be excluded from each equation in the PVAR system is evaluated in the final row, which displays the joint probability of all lagged variables in the equation.

Table 6.

Panel VAR-Granger causality Wald test

Equation/excluded		df	All firms		Non-financial firms		Financial firms
			Chi2	Prob > chi2	Chi2	Prob > chi2	Chi2	Prob > chi2
CGI
CS	1		1.068	0.301	0.873	0.350	247.949***	0.000
ROE	1		2.771*	0.096	0.658	0.417	370.691***	0.000
ALL	2		3.841	0.147	0.951	0.622	724.889***	0.000
CS
CGI	1		14.037***	0.000	13.751***	0.000	28.003***	0.000
ROE	1		25.615***	0.000	10.043***	0.002	144.336***	0.000
ALL	2		29.571***	0.000	17.365***	0.000	260.090***	0.000
ROE
CGI	1		2.337	0.126	27.660***	0.000	0.708	0.400
CS	1		11.977***	0.001	3.247*	0.072	322.990***	0.000
ALL	2		14.734***	0.001	30.676***	0.000	3196.255***	0.000

CGI denotes the ratio of overall corporate governance disclosure items, CS denotes firm leverage, and ROE denotes firm profitability

*, **, *** indicate significance at the 10, 5, and 1% levels, respectively

Table 6 shows that the joint significance Chi-square statistics in the last row show that all three variables i.e., CS, CG and FP variables are granger-caused by all the lagged variables for financial firms only. CS and FP variables are jointly and significantly granger-caused by all the lagged variables for all firms including non-financial and financial firms. However, the CG variable is not jointly and significantly granger-caused by all the lagged variables for the whole sample and non-financial firms. In general, the findings reveal that CG and FP jointly help to increase leverage, and CG and CS jointly boost the profitability of firms. Policy measures can, therefore, be focused on jointly improving CG standards and leverage to boost FP (Fig. 2).

Fig. 2.

Stability test for all firms, non-financial firms, and financial firms. All Eigen values are strictly less than one and lie inside the unit circle

Stability test

PVAR satisfies stability conditions for the whole sample and two sub-samples (Table 7).

Table 7.

Eigen value stability condition

All firms			Non-Financial Firms			Financial Firms
Real	Imaginary	Modulus	Real	Imaginary	Modulus	Real	Imaginary	Modulus
0.975	0	0.975	0.980	0	0.980	0.982	0	0.982
0.621	0	0.621	0.754	0	0.754	0.773	0	0.773
0.359	0	0.359	0.054	0	0.054	0.341	0	0.341

Panel variance decompositions

After a particular amount of time, forecast error variance decompositions (FEVDs) demonstrate the percentage of the total variation in one variable explained by the shock of another variable. As a result, they show the size of the total effect of one variable on another. A cumulative effect over ten years is provided. Table 8 reports the FEVDs of the baseline PVAR model and for the financial and non-financial sectors from 0 to 10 years. The implied FEVDs are derived using the causal ordering proposed by Abrigo and Love (2016). Based on the FEVD estimates, Table 8 indicates that for the baseline model and non-financial firms, the three variables are explained mainly by variations in their own variables. However, for financial firms, FP explains only 33.6% of its own lag, but explains 61.2% of the variation in CS. For all firms, 13.6% of the variation in CS is explained by CG and 13.2% of the variation in FP can be explained by CS. FP explained 5.5% of the total variance in the CS. The variation in CG is not explained by CS. Similarly, the variation in CG cannot be explained by FP and vice versa. For non-financial firms, CG and FP explain approximately 16 and 8% of the total variance in CS, respectively. CS explains only a small portion (1.8%) of the variance in CG. The 18.8% and 10.5% variations in FP can be explained by CG and CS, respectively. For financial firms, 8.6% of the variation in CG can be explained by CS, while 2% and 5% of the variation in CS can be explained by FP and CG, respectively. The 61.2% and 5.2% variations in FP can be explained by CS and CG, respectively. The FEVD results support the selection of FP as the most endogenous variable.

Table 8.

Forecast error variance decomposition

Model	Response variable and forecast horizon (from 0 to 10 years)	Impulse variables
		CGI	CS	ROE
All firms	CGI
	0	0	0	0
	10	0.997	0.003	0.000
	CS
	0	0	0	0
	10	0.136	0.809	0.055
	ROE
	0	0	0	0
	10	0.000	0.132	0.868
Non-financial firms	CGI
	0	0	0	0
	10	0.978	0.018	0.004
	CS
	0	0	0	0
	10	0.160	0.757	0.083
	ROE
	0	0	0	0
	10	0.188	0.105	0.707
Financial firms	CGI
	0	0	0	0
	10	0.910	0.086	0.004
	CS
	0	0	0	0
	10	0.050	0.929	0.021
	ROE
	0	0	0	0
	10	0.052	0.612	0.336

CGI denotes the ratio of the overall corporate governance disclosure items, CS denotes firm leverage, and ROE denotes firm profitability

Panel impulse response functions

Based on the calculated model, 200 Monte Carlo draws are used to calculate IRF confidence intervals. The IRFs are estimated using the same ordering as that used in PVAR. Figure 3 depicts the reactions to a one-standard-deviation shock. The impulse response function describes how one variable reacts to changes in another variable in the system while keeping all other shocks equal to zero.

Fig. 3.

Orthogonalised Impulse Response Function (IRF) for all firms. OVCGDI (CGI) denotes the ratio of the overall corporate governance disclosure items, TDBTNEV (CS) denotes firm leverage, and ROE denotes firm profitability

In terms of levels, the IRF plot in Fig. 3 shows that the shocks to CG have a constant insignificant positive effect on the FP of all Mauritian-listed firms over the 10 years. Shocks to the CG created a smaller but constant positive significant response to CS over a longer period. However, the shocks to the CS create a positive significant response in FP and fall to zero after approximately 8 years. Shocks to the CS create a smaller but constant response to the CG over a longer period. Moreover, shocks to FP result in an insignificant but negative response to CG over a longer period and a significant positive response to CS in the first 2 years which eventually dies to zero after 10 years period. The positive response from CS to a shock in FP is in line with TOT. For all firms, the Granger causality Wald test results together with the time path of the impulse response provide robust statistical evidence for the existence of a strong positive bidirectional association between CS and FP.

For non-financial firms in Fig. 4, a shock in CG results in a positive response to CS which is relatively lower in the first two periods, increases until period 6, and then starts declining at a negligible rate but remains positive. A shock in the CG has a negative effect on the FP over the 10 years. A shock in FP has a positive but negligible impact on CG over the years and a positive effect on CS, with the impact being significant in year 1 and gradually declining but remaining positive over the years. A shock in CS has a positive but negligible impact on CG but improves insignificantly over the 10 years. The response of FP to a shock in CS is negative in period zero, positive in period 1 to period 7, and then restarts a negative impact. For non-financial firms, the Granger causality Wald test outcomes, together with the time path of the impulse response, provide solid statistical evidence for the existence of strong positive reverse causality between CS and FP.

Fig. 4.

Orthogonalised Impulse Response Function (IRF) for non-financial firms. OVCGDI (CGI) denotes the ratio of the overall corporate governance disclosure items, TDBTNEV (CS) denotes firm leverage, and ROE denotes firm profitability

For the financial firms in Fig. 5, FP responds positively and significantly up to 7 years, returning to negative from period 8 to a shock in CG. A shock in CG has a significant positive effect on CS until year 5 and then turns negative from period 6. A shock in CS has had a significant positive impact on FP over the 10 years with the initial 4 years being much higher. The CG also responds positively and significantly to a shock in the CS over a long period. Conversely, a shock in FP negatively affects CG for a longer period, similar to all firms, while it has a positive impact on CS for a period above 10 years, with the initial 2 years being much higher. For financial firms, the Granger causality Wald test results together with the time path of the impulse response present robust statistical evidence for the existence of a strong bidirectional correlation between CS and FP, and CS and CG, respectively.

Fig. 5.

Orthogonalised Impulse Response Function (IRF) for financial firms

Robustness analysis

Another estimator XTVAR

The least-squares dummy variable estimator XTVAR, developed by Cagala and Glogowsky (2014), is used to compare the results of PVAR to assess its robustness. The results of the IRFs using this approach indicate that for all firms, the sign of XTVAR results is the same with different magnitudes of the coefficients, but the lags of FP on CG, CG on CS, and CS on FP show no significance compared to PVAR results. Moreover, there are a few differences between the IRFs and the variance decomposition analysis results of XTVAR and PVAR for CG and CS on FP.

The XTVAR and PVAR results are, therefore, almost similar for the whole sample and non-financial firms, although some minor differences are noted for financial firms and confirm the robustness of the results.

Alternative orderings of variables

Instead of CGI → CS → ROE in the baseline model, alternative orderings are tried, that is, ROE → CS → CGI, and the variance decomposition analysis for CG and CS on FP is similar for the whole sample and non-financial firms, respectively. Conversely, differences are noted in the variation in FP and CS explained by shocks in FP and CS, respectively.

Moreover, for all firms, the signs of the effects in general shapes of IRF are similar, except for the response of CG to the shock of FP and vice versa, which are slightly different. For non-financial firms, the signs and general shapes are similar. Concerning financial firms, the sign of the effect and general shapes of IRFs are similar except for the response of CS and FP to a shock in CG having slightly different shapes and the response of CG to a shock in FP.

Overall, the sensitivity analysis indicates that the findings are strongly robust to an alternative estimator, XTVAR, and the alternative ordering of the variables.

Summary and conclusion

This study aims to analyse the interrelationships and interdependencies between CG, CS and FP. In this study, the PVAR with a GMM framework is used and the analysis is based on IRF and FEVD. First, there is no evidence of reverse causality between CG and FP for all firms, including non-financial and financial firms, because, for the whole sample and financial firms, the causality runs negatively and significantly in one way only—from FP to CG and not vice versa. However, for non-financial firms, causality runs negatively and significantly in one way only, from CG to FP, and not vice versa. Second, CS positively and significantly affects FP and vice versa, showing a robust bidirectional relationship between CS and FP for the whole sample and two sub-samples, respectively, and strongly supporting TOT. Third, the findings show that for the whole sample and non-financial firms, there is a unidirectional relationship between CG and CS, since there is a substantial positive effect of CG on CS but not vice versa. However, strong bidirectional correlations between CG and CS are found for financial firms with varying signs. For instance, CG has a strong negative impact on CS, supporting agency theory, implying that better governed financial firms have less leverage, and CS has a significant positive effect on CG which also supports agency theory. Fourth, the FEVD results support FP as the most endogenous variable. For the baseline model and non-financial firms, the three variables are explained mainly by variations in their own variables, whereas for financial firms, 61.2% of the variation in FP is explained by CS. Variation in CG is in all three cases, negligibly explained by other variables, FP and CS.

For the whole sample and non-financial firms, the Granger causality Wald test results, together with the time path of the impulse response, provide robust statistical evidence for a strong positive bidirectional correlation between CS and FP. For financial firms, a strong bidirectional relationship exists between CS and FP, CS and CG, respectively. Overall, the results remain robust after using XTVAR as an alternative approach to PVAR and the alternative ordering of the variables for the whole sample and non-financial firms. Some discrepancies are noted in the case of financial firms; however, in general, the results remain robust.

This study contributes to the existing CG and CS literature and FP effects by employing a PVAR method that overcomes the endogeneity issue due to the possible existence of reverse causality or interdependencies between the three variables which is being done to the best of the authors’ knowledge for the first time for this topic. Moreover, statistical analysis is potentially useful to policymakers in their endeavour to articulate CG recommendations and policies. The implications of the present study can also be of significant value to investors (both local and foreign) and managers, enlightening them that, in this small island emerging economy, although firms are highly geared, more debt has a significant positive impact on FP. Investors while having their investment and credit decisions may also rely on the results that better governed Mauritian financial firms have better financial performance although its effect is short-lived and reduces its significance with the elapse of time but remains positive. In general, better governed firms are more able to have more debt and with growth opportunities in an emerging economy, such additional debt can be invested in viable/profitable projects to increase profitability (results show a significant positive impact of CS on FP). Policymakers can consider the monetary policies to make loans more easily accessible at competitive interest rates to firms, because as a firm increases its debt holdings, the probability of financial distress increases and creditors are less likely to re-finance or renegotiate and also take measures to improve CG standards that jointly help to further improve their profitability. Improving CG standards through better information disclosure not only reduces the cost of capital of loanable funds but also tends to attract more institutional investors and financial analysts to firms. However, policymakers can explore their fiscal measures to ensure consistent positive economic growth, since any slowdown can increase the risk of bankruptcy, which can have major ramifications due to the presence of heavily indebted enterprises.

However, the current study is constrained by certain inherent constraints and limitations: first, the restrictions that are imposed by utilising annual CG reports and accounts; second, this study is confined to listed companies. The small size of the Mauritian economy and its stock exchange limits the possibility of increasing the sample size more specifically for financial firms. Moreover, the quantitative analysis does not provide much insight into the unobserved and unmeasured factors that may influence the interactions, which can be overcome through qualitative methods.

The limitations listed above imply that future studies can build on and use current research as a foundation. Future studies may focus on using qualitative methodologies to better understand the interrelationships between CG, CS and FP, particularly for variables where no connection is detected. Furthermore, the current study may be expanded to compare with other emerging economies and estimate alternative profitability indicators (market performance and Return on Assets). Longitudinal studies can also be undertaken by comparing pre-COVID and post-COVID-19. The dynamic interrelationships between concentrated ownership structure or other CG components, CS and FP can also be examined in future research.

Author contributions

RKR bears full responsibility for the submission and confirms that the authors listed on the title page have contributed significantly to the work. Specifically, RKR has written all parts of the manuscript (introduction, literature), collected, analysed, and interpreted the findings of the PVAR approach on the interrelationships and interdependencies among corporate governance, capital structure, and firm performance of listed Mauritian companies. BS has reviewed the manuscript to put it into perspective. The authors read and approved the final manuscript.

Funding

The authors received no financial support for the research.

Data availability

All data and materials used in this study from the annual reports of the listed firms are available upon subscription on https://www.african-markets.com/en/annual-reports.

Declarations

Conflict of interest

The authors declare that they have no competing interests that are relevant to the content of this article.

Ethical approval

Not applicable.