Do Audit Committees and Corporate Governance Mechanisms Affect the Bank’s Performance? Empirical Evidence from Panel Data Analysis ()
1. Introduction
The need to strengthen corporate governance principles within the business environment is due to the increasing conflict of interest arising from the principal-agent relationship. The Basel Committee on Banking Supervision (BCBS) lays prominence on the significance of good corporate mechanisms for banks to build trust and confidence among the owners and the general public at large. Banks are deposit-taking institutions are therefore are highly leveraged and therefore, need to stay away from unpleasant occurrences through effective corporate governance mechanisms. Better corporate governance promotes transparency and provides adequate checks and balances between the principal and agent interests through the monitoring of the board. According to Barton, Coombes, and Wong (2004) , good corporate governance practice is an important element in attracting investors, and investors are willing to pay a premium of up to 25% for a well-governed firm.
Most of the reported cases of corporate failures are ascribed to bad corporate governance practices (Appiah, 2013) . Corporate governance failure was identified as one of the causes that led to the recent banking crisis leading to the revocation of six strings of banks in Ghana by the Bank of Ghana in 2018 that were licensed under Act, 2016 (Act 930). It was reported that one of the reasons for the revocation of banking license by the Bank of Ghana in 2018 was due to poor corporate governance practices by the banks. The purpose of the revocation is to ensure safety, stability, and soundness in the financial systems of Ghana and to protect the depositors’ interests (Bank of Ghana [BoG], 2018) . Some of the weaknesses identified by the central banks of Ghana for collapsed banks are poor corporate governance, poor risk management, non-performing loans, undercapitalization, and regulatory lapses. To prevent future recurrence, the central bank put measures in place to restore the stability and resilience of the financial system in Ghana (Dwamena & Yusoff, 2022) .
The need to strengthen corporate governance principles within the business environment is due to conflict of interest arising from the separation of ownership and management of the organization. The Corporate governance structures in the private sector focus on appropriate ways to manage and control a structure of an organization. In the following studies Claessens and Yurtoglu (2013) , Love (2011) , and Rajagopalan and Zhang (2008) opined that good corporate governance reduces agency cost, minimizes information asymmetry, lowers capital cost, builds trust for the stakeholders, and improves a firm’s profitability and value. One of the ways advocated by policymakers and researchers to enhance good corporate governance practices is audit committee independence as a critical element of any corporate governance system (Al-Jalahma, 2022) . Audit committee Independence (ACI) is considered one the most important in agency theory. The Bank of Ghana, the financial sector regulator, released a comprehensive corporate governance code for the banking industry. The release was christened “Corporate Governance Directive 2018”. One of the provisions advocates for the establishment of an audit committee of the board which must consist of non-executive directors. The chairperson of the audit committee shall be an independent director. Members of the audit committee must be competent in accounting, auditing, and finance and shall have oversight of the financial institution’s internal and external audit function. In addition, the Companies Act, 1963 (Act 179) which primary governance structure in the private sector also undergone amendment, the Companies Act, 2019 (Act 992) as well as corporate governance regulations issued by the supervisory bodies like the Bank of Ghana and National Insurance Commission has enhanced corporate governance regimes in Ghana in recent times. It focuses on the appropriate way to manage and control a structure of an organization.
The agency theory espoused by Jensen and Meckling (1976) asserts that audit committees can mitigate the agency conflict associated with the principal-agent relationship. Scholars of agency theory believe audit can mitigate the agency conflict between the principal and agent. The main purpose for establishing an audit committee is to provide oversight over management in respect of the firm’s financial reporting process, compliance with internal controls systems with laws and regulations, and internal and external audit processes. The audit committee is seen as a vital institution that assists the board of directors in terms of transparency and integrity in the financial reporting process (Klein, 2002) . The audit committee has many avenues to monitor and control management information asymmetric to ensure owners get the right information for decision-making (Allegrini & Greco, 2013) . Audit committee AC) is recognized globally as a very important governance mechanism for ensuring the integrity of the financial reporting and audit process are enhanced (DeZoort et al., 2002; Cohen et al., 2008) . An audit committee (AC) is a very important component of good corporate governance. According to Dwamena (2021) audit committee is an essential part of good corporate governance and provides effective oversight of the objectivity, independence, and performance of the auditor and the audit quality work. AC supervises and oversees timely, appropriate, and relevant financial reporting and disclosure. Not surprisingly, market regulators including Ghana Stock Exchange have set in place certain standards or requirements designed to strengthen audit committee effectiveness. Therefore, strengthening the audit committee function can minimize this conflict between the principal and agent relationship.
Despite the theoretical importance of audit committees as advocated by agency theory that it would strengthen corporate governance mechanisms. The empirical evidence on the relationship was mixed and uncertain. Some studies found no significant relationship between audit committee independence and the firm’s performance (Xie, Davidson, & DaDalt, 2013; Osevwe-Okoroyibo & Emeka-Nwokeji, 2021; Davidson, Goodwin-Stewart, & Kent, 2015) . Studies found a positive and significant relationship between audit committee independence and the firm’s performance (Amahalu & Ezechukwu, 2017) while these studies found a negative and significant relationship between audit committee independence and the firm’s performance (Yang & Krishnan, 2005) . The mixed outcome could be attributed to wrong estimation challenges often associated with violating regression assumptions, especially on omitted variables and unobserved heterogeneity challenges. Therefore, I am motivated to use a panel data estimator to handle the issue of unobserved heterogeneity challenges and to include control variables in the model to mitigate the issue of omitted variables in this study. This article will contribute to corporate governance literature in several ways. Firstly, it shows that having a corporate governance mechanism is not enough rather deepening the board and audit committee is paramount for ensuring there check and balance between the principal and agent relationship. For methodological contribution, the study included control variables to ensure the problem of omitted variables is addressed, and secondly, panel data regression was adopted to ensure the right model estimator is selected to control the heterogeneity challenges in this study. The rest of the paper is organized as follows: section 2 reviews related literature, section 3 looks at the data and research methodology issues, section four discusses the empirical findings and section five concludes and makes recommendations for the study.
2. Literature Review
This section reviewed pieces of literature on corporate governance mechanisms and the firm’s performance. The reviews are in two sections. These are: 1) Theoretical review and 2) empirical review.
2.1. Theoretical Review
The main theory underlying this study is agency theory and it is discussed below in this study.
Agency Theory
Agency theory asserts that the ownership and control of firms are entrusted to different individuals, leading to the creation of a conflict of interest between the principal and agent or agency problem (Aguilera et al., 2008) . The shareholders are the owners or the principal of the bank, they appoint managers (agents) to manage their banks with the expectation that managers would exploit the owners’ interests. On the contrary, the managers (agents) may not take the best interests of the owners (Padilla, 2002) . Underpinning the agency problem is the tendency of the agents to satisfy their ambition and forfeit the interest of the principal (Daily, Dalyon, & Cannella, 2003) .
The managers (agents) may seek their personal desires or self-interest and opportunistic behaviour which fall short of the owners’ aspirations or pursuits. Again, managers may seek unwarranted risks which act as a setback against the principal-agent relationship. To control this opportunistic behaviour of managers against the owners, agency theorists advocate the setting up of strong corporate governance structures to minimize the opportunistic behaviour of managers. This implies that managers cannot be trusted, and therefore Jensen and Meckling (1976) assert that since managers are constantly pursuing their interests then the entity should set up a strong corporate governance structure to mitigate the agency problem. Therefore, the idea of setting up a corporate governance structure is to align the goals of owners to that of managers and to set boundaries of effective governance (Jensen & Meckling, 1976) . Lu and Zhu (2020) opined that establishing an effective board of directors to monitor management’s interest and protect the owners’ interest would overcome the problems of separation and ownership between principals and agents.
2.2. Empirical Review
Theoretically, both agency theory and resource dependency theories predict a positive causal relationship between corporate governance mechanism and the bank’s performance and audit committee independence and the bank’s performance. However, some empirical evidence on these studies was mixed and varied. This sub-section reviews some empirical results on corporate governance on the bank’s performance and audit committee independence and the bank’s performance in this study.
2.2.1. Empirical Review of Direct Effect between Corporate Governance and Bank’s Performance
Corporate governance is a system used to assess accountability and provide an adequate internal control system for management. It focuses on creating and developing better corporate governance systems. Corporate governance is essential to create a culture of transparency, integrity, and accountability leading to the creation of value and financial wealth for the banks and the shareholders. Studies conducted on the relationship between corporate governance and performance has revealed mixed outcomes (Mak & Kusnadi, 2005; Detthamrong, Chancharat, & Vithessonthi, 2017) . To date, pieces of the empirical literature on the relationship have no consensus among researchers agreement among researchers (Agrawal & Knoeber, 1996; Xu & Wang, 1997; Lehmann & Weigand, 2000; Gruszczynski, 2006; Berthelot et al., 2010) . For instance, some of these studies reported a positive and significant relationship between corporate governance characteristics and bank performance (Brown & Caylor, 2006; Drobetz et al., 2004; Huang, 2007; Varshney et al., 2012) . However, other studies have opined a negative relationship between corporate governance of the firm’s performance (Doc-Ho & Nguyen, 2014) .
2.2.2. Empirical Review of Direct Effect between Audit Committee Independence and Bank’s Performance
The audit committee provides additional safeguards against fraud and ensures that the financial statements are prepared according to the standards. To enhance the work of the audit committee, the committee must be adequately independent and must have requisite qualifications and experience in accounting, auditing, and financial matters. An independent audit committee reduces the problems of information asymmetry, adverse selection, and moral hazard by monitoring management (Aldamen et al., 2012) . Audit Committee Independence means being more watchful, and more involved in the financial matters of the firm, to get the monitoring done effectively (Mohiuddin & Karbhari, 2010; Qeshta et al., 2021) . The effectiveness of the audit committee members is usually seen through their independence, financial literacy, and expertise and allocating sufficient time to meet regularly to discuss financial and related party issues (Putrajaya Governance Committee [PGC], 2006) . It is a true reflection of the firm’s commitment to good corporate governance practices. Audit committee independence is a widely accepted idea that independent directors can effectively monitor management actions and minimize management opportunistic behavior that conflicts with the principal’s interests. According to Mangena and Pike (2015) , audit committee independence is more likely to minimize management influence. Previous opined there is a positive relationship between audit committee independence and performance (Klein, 2002; Mansi & Reeb, 2002) .
3. Methodology
The study employed a quantitative method involving an ex-post facto design to collect data to test the hypotheses and assess the effect of audit committee characteristics on the bank’s performance using panel data analysis. A purposive sampling methodology was adapted to select 20 banks from the 23 licensed banks from the Bank of Ghana from 2013 to 2022. The study used panel data to pools observations from a cross-sectional unit over ten years to facilitate the investigation of an effect that is not detectable in pure cross-sections or pure time-series studies. The advantage of using panel data analysis is that it can treat problems of omitted variables and unobserved heterogeneity issues usually associated with either pure cross-sectional data or pure time-series data. The analysis is done using STATA (i.e., statistics and analysis software) as the software for this study.
3.1. Research Variables
Three research variables used in this study are classified into broad categories: Dependent, independent, and control variables. The dependent variable was used to assess the bank’s performance, the independent variables were used to assess the audit committee independence and corporate governance mechanisms, the control variables were used to control the potential effects on the bank’s performance and to mitigate the problem of omitted variables.
3.1.1. Dependent Variable (i.e., NIM)
Net Interest Margin (NIM): NIM is the dependent variable used to operationalize bank’s performance of banks. NIM is calculated as the ratio between received and paid interests, all over total assets. The ratio measures the margin a bank makes on its core business of the bank. The researchers used NIM as a proxy for measuring the efficiency in the banking sector. For formula for calculating NIM is expressed in Equation (1):
(1)
3.1.2. Independent Variables (i.e., ACI, CEO, Bsize, and NED)
Four variables were used as independent variables to assess the effect on the bank’s performance these are Audit Committee Independent (ACI), CEO-duality (CEO), board size (Bsize), and non-executive director (NED). These are:
Audit committee Independence (ACI): Audit committee independence is the mediating variable in this study. Audit committee independence is seen as a vital institution that assists the board of directors in terms of transparency and integrity in the financial reporting process (Klein, 2002) . The primary function of the establishment of the audit committee is to enhance the financial reporting quality.
The proxy for measuring the audit committee independence composition is the number of non-executive directors or outsider directors over the number.
ACI = No of Non-executive/No of directors on the audit committee
CEO-duality (CEO): CEO-duality eludes for circumstances where the CEO holds the positions as the Chairman of the board and the CEO of the firm at the same time as undesirable corporate governance practice. Cadbury Committee considers that one person should hold the CEO and another should hold the position of the Chairman of the board and not both. Given that many corporate governance experts find CEO duality an undesirable practice, many studies have found a negative relationship between ACI and the firm performance (…), even though there are some mixed outcomes in the literature. The separation of the position provides an opportunity to assess independently the performance of the CEO and the executive management to require better accountability from management (Ahulu & MacCarthy, 2019; Monks & Minow, 2004) . The study equates a CEO-non-duality variable as “0” and when the CEO also serves as the Board Chairman else “1” if the position is handled by different persons.
Board Size (Bsize): The board size represents the number of directors on the board. A board size of seven or less is considered a small board and those above seven are considered a large board size. Empirical evidence between board size and the firm’s performance was mixed. These studies revealed a positive relationship between board size and performance (Elsayed, 2011; Goodstein et al., 1994; Huang, 2007; Sanda, Mikailu, & Garba, 2010; Saravanan, 2012) while these studies revealed a negative relationship between board size and performance (Yermack, 1996; Eisenberg et al., 1998) . The proxy for measuring the board size is the number of directors on the board.
Bsize = Number of board members
Non-Executive Director (NED): A high number of NED on the board represents board independence. Advocators of good corporate governance argue for a higher number of non-executive directors on the board enhances the independence of the board (Ahulu & MacCarthy, 2019) . The proxy for measuring the board composition is the number of non-executive directors or outsider directors on the board.
NED = No of Non-executive/No of directors on the board
3.1.3. Control Variables (i.e., Fsize and Growth)
Based on pieces of literature on the need to control the specific effect of corporate governance and the bank’s performance, firm size, and growth are included in the regression model as control variables. Control variables affect dependent and independent variables; if not, control affects the study’s outcome. The control variables used in this study are firm size and growth. According to Kiel and Nicholson (2003) , firm size and growth co-vary with many pieces of literature on Corporate Governance and the Bank’s performance nexus.
Bank Size (Size): The bank measures either the bank’s economy of scale or diseconomies of scale for this study. Boone et al. (2007) asserted that as the firm size becomes more extensive and more diversified, the board size increases; therefore, more corporate advice and counsel are needed from the board. We expect the bank’s size to affect the bank’s performance positively (Lehn et al., 2004; Abbasi & Malik, 2015) . As the bank size increases, the bank performance also increases significantly in the case of small and medium-sized banks in the banking sectors. The proxy for size is measured as the logarithm of the bank’s total assets. The logarithm helps get the bank’s total assets due to its capability to standardize values, thus bringing them to the same platform for more efficient analysis.
Growth: Growth represents the rate of growth of the firm. A growing firm can generate enough revenue to finance its operation and vice versa. A growing firm tends to contribute positively to the firm’s performance and vice versa. Pandey (2007) concluded that growth positively correlates with a firm’s performance. Park and Jang (2014) measured the growth using the current year’s sales minus last year’s sales divided by last year’s sales and expressed it as a percentage change in annual sales.
Growth = (Current year’s revenue – Previous year’s revenue)/Previous year’s revenue
3.2. Model Specification
To estimate the effect of corporate governance and audit committee independence on the bank’s performance, the study adopted an equation adopted similar to Hair, Black, Babin, Anderson, and Tatham (2006) . The dataset for this analysis is taken from both cross-sectional and time series observation from the selected firms and it is organized to fit panel data. The advantage of using panel data instead of cross-sectional data or time series data is that it can resolve the problems of omitted variables and unobserved heterogeneity issues usually associated with either pure cross-sectional data or pure time-series data. The model was slightly modified in this study to include the firm’s size and growth as a control variable to mitigate the problem of omitted variables as in Equation (2):
(2)
where:
Min is the dependent variable of this study.
CEO, ACI, Bsize, NED, Fsize, and Growth are the independent and control variables in this study.
β1, β2, β3, β4, β5 and β6 are the regression coefficients to be estimated.
β0 is the constant or the intercepts on the regression equation.
t = is the time series of the study (t = 1, 2, 3, 4…10).
i = is the cross-section (i.e., 18 firms from the banking sector of Ghana).
ε = Unique Error or Error Term.
3.3. Research Hypotheses Development
The following four hypotheses were espoused to assess the relationship between the audit committee, corporate governance mechanisms (CEO, ACI, Bsize, and NED) and the bank’s performance (NIM) in this study:
H01: There is no significant relationship between CEO non-duality (CEO) and the bank’s performance (NIM). Hence CEO non-duality does not influence the bank’s performance for the period selected for the study.
H02: There is no significant relationship between audit committee independence (ACI)) and the bank’s performance (NIM). Hence audit committee independence do not influence and bank’s performance (NIM).
H03: There is no significant relationship between board size (Bsize) and the bank’s performance (NIM). Hence board size do not influence and bank’s performance (NIM).
H04: There is no significant relationship between the non-executive director (NED) and the bank’s performance (NIM). Hence non-executive director does not influence and bank’s performance (NIM).
Each of these hypotheses is accepted or rejected based on the outcome of panel data regression as an analytical tool involving the t-statistic combined with the p-value, at a 5% significance level, is used as the decision criteria.
4. Results and Discussion
This section presented the results of the main aim of this study and followed up the discussion of the results. The section is divided into four main subsections: Descriptive statistics, Pearson correlation analysis, Hausman test specification test, and the panel regression result.
4.1. Descriptive Statistics
To estimate the relationship between corporate governance practices and a firm’s market value, descriptive statistics were used to present graphical summaries of the relationship between the dependent, independent, and control variables. Table 1 presents the mean, standard deviation, minimum, maximum, Kurtosis, and Skewness of the dataset. Descriptive statistics analysis is used to identify any possible irregularities before inferential statistics.
The second column of Table 1 indicates the mean values for NIM, CEO, NED, Bsize, ACI, Size and growth were 0.084, 0.007, 0.753, 7.667, 0.628, 6.365, and 0.151 respectively, for the ten-year under-study. The mean value for NIM was 0.084 or 8.4%. The bank’s net interest margin (NIM) indicates the bank’s profitability and growth. The mean value of 8.4% is an appreciable performance of the banks within the period. Table 1 shows that CEO-non-duality was 0.07. This implies that most of the banks have separated the Chairman position from the CEO position and are also appointed externally to represent the Chair of the boards. Again, the mean value for Non-executive directors (NED) was 0.753 and it implies there was sufficient independence of the board of directors (Yesser, 2011; Harford et al., 2008) . Table 1 shows the average board size was 7.667 and it implies that most of the firms listed in Ghana have an identical and average board size. An average board size (Bsize) is between 7 and 8 directors. A board size below seven is considered as small while a size above 10 is considered as large. The mean value for the Audit Committee Independence (ACI) was 0.628 and it indicates the degree of independence of the audit committee members of the board (Lin et al., 2006; Javeed & Azeem, 2014; Sheikh, Wang, & Khan, 2013) .
The standard deviation measures the spread among the dataset and reveals how close or dispersed the variables are from the means of the dataset. The standard deviation for NIM, CEO, NED, Bsize, ACI, Size, and growth was 0.021, 0.003, 0.130, 1.686, 0.148, 0.997, and 0.609. A high standard deviation means dispersion from the mean is large, and the variable is volatile. In contrast, a low standard deviation implies the variable cluster around the mean and it is stable. The study observed that growth is the most volatile variable among the variable and followed after was net interest margin (NIM) for the period under study. However, CEO, NED, and ACI are the most stable in the dataset. Furthermore, Table 1 shows information about the Kurtosis and Skewness of the variables in the dataset. The Information on skewness and kurtosis determines whether the dataset met the normality assumption (Kline, 2011) . The acceptable skewness values should be between −2 and +2, and the kurtosis should be between −7 and +7 when assessing normality in regression (Byrne, 2010; George & Mallery, 2010) .
The result shows that NIM, CEO, NED, Bsize, ACI, Size, and Growth exhibit a positive skewness and are closer to zero. A positive skewness implies that the dataset is positively skewed and that the right tail is longer than the left. Therefore, the skewness for NIM, CEO, NED, Bsize, ACI, Size, and Growth is approximately symmetrical. The kurtosis for NIM, CEO, NED, Bsize, ACI, Size, and Growth were 3.327, 2.916, 3.130, 2.886, 3.004, 3.074, and 2.996, respectively.
Source: Author’s Stata version 15 Computation.
4.2. Pearson Correlation Analysis
Table 2 shows the Pearson correlation analysis between the variables used in this study. The Pearson correlation analysis was used to determine whether there was a significant relationship between a dependent variable and the independent variables. Again, it is used to identify the existence of a strong relationship among the independent variables for removal from the model. Table 2 shows the Pearson correlation analysis between variables used in this study. It is also used preliminary test for multicollinearity and linearity assumptions among variables before the regression analysis.
Table 2 indicates the correlation between CEO, ACI, Bsize, NED, Bsize, Size, Growth, and NIM were O.603, 0.573, 0.497, 0.624, 0.523, and 0.446, respectively. It shows that there were significant and positive correlations between the independent variables and the dependent variable. According to Pfeifer & Carraway (2000) , the correlation between the variables is considered very low when the correlation coefficient is below 0.20, the correlation between the variables is considered low when the correlation coefficient is between 0.21 and 0.40, the correlation between the variables is considered moderate when the correlation coefficient is between 0.41 and 0.70, and a high correlation when the value or the degree of the association is from 0.71 to 0.91. Table 2, revealed that there is significant relationship between CEO-non-duality (CEO) and bank’s performance (r = 0.603, p < 0.01), audit committee independence (ACI) and bank’s performance (r = 0.573, p < 0.01), board size (Bsize) and bank’s performance (r = 0.497, p < 0.01) and non-executive director (NED) and bank’s performance (r = 0.624, p < 0.01). However, a correlation among the independent variables was not high enough to violate the multicollinearity assumption. According to Darmadi and Gunawan (2013) , a strong correlation among the independent variables signals the presence of multicollinearity challenges between variables in the model. A correlation coefficient index greater than 0.90 indicates the presence of multicollinearity challenges (Asterious & Hall, 2007) . It is worth noting that the correlation among the independent was lower than 0.500 to avoid multicollinearity challenges for the modal. To further confirm the absence of multicollinearity in the model, the study employed Vector Inflation Factor (VIF) was used to ensure the multicollinearity assumption is not violated. When a tolerance value is lower than 0.20 and a VIF value is greater than 10 suggests a multicollinearity problem (Menard, 1995; Myers, 1990) . The tolerance value is the inverse of the Variance Inflation Factor (1/VIF).
Table 3 indicates that the VIF values for the variables are less than 10 and shows a clear indication that the variables are not suffering from multicollinearity problems. The highest VIF among the variable is Growth which is 3.401 with a tolerance value above 0.294. There, none of the variables would be removed from the model because none of them violated the multicollinearity assumption
Table 2. Pearson correlation analysis.
Source: Author’s Stata version 15 Computation.
Table 3. Variance Inflation Factor (VIF) and Tolerance Value.
Source: Author’s Stata version 15 Computation.
for this study. The existence of multicollinearity would not affect how the regression is performed but rather affect the interpretation of the result (Anderson, Sweeney, & Williams, 2009) . Therefore, since there is no independent variable that is strongly correlated, the study proceeds to run the regression model without the need to eliminate any variable from the model for fear of multicollinearity challenge.
4.3. Multiple Regression Results
This sub-section aims to assess the effect of the audit committee mechanism on the bank’s performance. The panel regression analysis was carried out in three stages: 1) to specify the most suitable estimator between the fixed effect model and random effect model for this analysis, 2) to conduct the panel regression analysis and 3) to test the regression assumption underlying the regression model. The regression analysis is inferential statistics used to determine whether the relationship observed in the sample is similar to that of the larger population, therefore, it is important to ensure that the model is goodness of fit to predict the relationship among the variables.
4.3.1. Econometric Techniques for Efficient Estimation
This study employed the Hausman test as the econometrics tool to choose between the fixed effect model and the random effect model for this analysis. The null hypothesis for the Hausman test assumes the specified random effect estimator while the alternate hypothesis assumes the fixed effects estimator (Green, 2003) . The rule of thumb is to reject the null hypothesis and select the Fixed effects estimator as the most suitable estimator for the analysis when the p-value is less than or equal to 5% and else, the study is compelled to select the Random effects estimator as the most suitable estimator this analysis when the p-value is greater than and statistically significant (i.e., when the p-value > 0.05).
The result from the Hausman test presented in Table 4 indicates that the (Chi-square = 1.63, p-value = 0.6484). This means that the p-value is greater than 5%. Since the p-value is greater than 5% (p-value < 0.05), the study failed to reject the null hypothesis and conclude that the preferred estimator model for the assessment is a random effect estimator (Green, 2003) . Therefore, the suitable model to estimate the between the audit committee mechanism and the bank’s performance is the random effect model.
4.3.2. Relationship between Corporate Governance, Audit Committee Independence, and Bank’s Performance
The result from the random effect estimator is presented in Table 5 and shows the relationship between the dependent variable (i.e., NIM) and independent variables (i.e., CEO, ACI, Bsize, and NED). The R2 of the model summary explains the fraction of the variation in the dependent variable that is explained or accounted for by the independent variables. The R2 indicates statistically the percentage of the variance in the dependent variable that the independent variables
Table 4. Test results from hausman test.
Source: Researcher’s Stata version 15 Computation. b = consistent under Ho and Ha; obtained from xtreg. B = inconsistent under Ha, efficient under Ho; obtained from xtreg. Test: Ho: difference in coefficients not systematic. chi2(5) = (b-B)’[(V_b-V_B)^(−1)](b-B) = 1.63. Prob > chi2 = 0.6484. (V_b-V_B is not positive definite).
Table 5. Regression result for Step 1 (CEO, ACI, Bsize, and NED →NIM).
Source: Author’s Stata version 15 Computation.
explain collectively. It is the fraction of the dependent variable that can be explained or accounted for by the independent variables. Therefore, the R2 of 0.7783 or 77.83%, implies that the dependent variable (NIM) is jointly explained by the independent variables (CEO, ACI, Bsize, and NED). It implies that the model can predict or explain the changes in the bank’s performance. When the R2 is higher and closer to 100% or 1.0, the model is better predictive power for the analysis. It implies that the model is a strong determinant of the relationship between sustainability reporting, corporate governance practice, and value. Thus, only 22.17% of the regression model is unexplained by the independent variables. The “rho” gives the proportion of the variation in the dependent variable that can be explained by the εit (Caruso & Cliff, 1997) . Again, Wald chi2(6) value of 591.27 with an associated Prob > chi2 = 0.0000 implies the model is fit to explain the relationship between audit committee characteristics and the bank’s performance and therefore, suggests further that the independent variables were properly selected and jointly used to predict the relationship in the model. The error εit is not correlated with the regressors variables and it assumes that Random Effects Estimator (corr (εit, Xi) = 0.00).
Table 5 shows the regression coefficients (β), and their corresponding standard error, t-statistics, and the significance of all coefficients for the model used to assess the null hypotheses of this study. The estimated coefficient (β) indicates the magnitude and direction of each of the variables to predict the effect of the independent variable on the dependent variable. The study observed that there were positive and significant relationships between CEO, ACI, Bsize, NED, and NIM. Table 5 shows that the coefficient (β), t-statistic value, and p-value between CEO non-duality (CEO) and the bank’s performance were (β = 0.257, t = 3.218, and p < 0.05). This implies that CEO non-duality affects the bank’s performance positively and significantly since the p-value is less than a 5% level of significance. The CEO-non-duality allows the board to depend on better accountability from the CEO and the management team as a whole. This outcome is consistent with the previous studies and agency theory which requires that the CEO position and the board Chairman Position should not be handled by one person as described in corporate governance guidelines. Therefore, based on the result in Table 5 and the explanations thereof, the study failed to reject the null hypothesis (H01) and concludes that CEO non-duality affects the bank’s performance significantly. Therefore, all things being equal, a 1% increase in audit committee independence leads to an increase of 25.7% in the bank’s performance.
Additionally, Table 5 shows that the coefficient (β), t-statistic value, and p-value between audit committee independence (ACI) and the bank’s performance were (β = 0.346, t = 3.490, and p < 0.05). This implies that audit committee independence affects the bank’s performance positively and significantly since the p-value is less than a 5% level of significance. This outcome is consistent with previous studies (Ahulu & MacCarthy, 2019; Kakar et al., 2021; Monks & Minow, 2004) and agency theory. Agency theory asserts that audit committee independence provides means to check on the opportunistic behaviors of management that conflict with the objectives of the firm and the owners. Therefore, the study rejects the null hypothesis (H02) and concludes that audit committee independence affects the bank’s performance. Therefore, all things being equal, a 1% increase in audit committee independence leads to an increase of 34.6% in the bank’s performance.
Again, Table 5 shows the coefficient (β), t-statistic value, and the p-value between board size (Bsize) and the bank’s performance (β = 0.151, t = 4.283, and p < 0.10). This implies that board size affects the bank’s performance positively but insignificantly since the p-value was marginally significant at 10%. This outcome of no significant relationship between board size and a firm’s performance is consistent with some previous studies such as (Adams & Mehran, 2003; Belkhir, 2009; Busta, 2007; Hadi Zulkafli & Abdul Samad, 2007) who opined that board size no significant relationship with firm’s performance. However, it is consistent with previous studies that opined a positive association between board size and firm performance (Dwivedi & Jain, 2015; Abor & Biekpe, 2007; Kiel & Nicholson, 2003) . Therefore, the study failed to reject the null hypothesis (H03) and concludes that board size is insignificantly related to a bank’s performance.
Finally, Table 5 shows the coefficient (β), t-statistic value, and p-value between non-executive directors (NED) and the bank’s performance (β = 0.239, t = 3.126, and p < 0.05). This implies that non-executive director affects the bank’s performance positively and significantly since the p-value is less than a 5% level of significance. This implies that a high number of non-executive director improve the bank’s performance. This outcome is consistent with agency theory and previous studies (Chabachib et al., 2020; Chen et al., 2017; Jiang et al., 2020; Naciti, 2019; Nyamongo & Temesgen, 2013) that opined that non-executive directors affect a firm’s performance positively. Again, Chabachib et al. (2020) conducted a study and concluded that the high presence of non-executive directors on the board enhances the board’s independence to monitor management performance. Therefore, the study rejects the null hypothesis (H04) and concludes that audit committee independence affects the bank’s performance. Therefore, all things being equal, a 1% increase in audit committee independence leads to an increase of 23.9% in the bank’s performance.
4.3.3. Diagnostic Checks and Robustness Test
To ensure is well-specified to estimate the relationship between corporate governance, audit committee independence, and bank performance, the study proceeds to test the following underlying regression assumptions: normality, autocorrelation, endogeneity, and heteroskedasticity tests to provide that the model is suitable to examine the relationship among and also inference can be made from the outcome.
The diagnostic tests for normality, autocorrelation, and heteroscedasticity presented in Table 6 show that the model passes all the diagnostic tests. Therefore, the model has violated any underlying regression assumptions; it is stable
Table 6. Model diagnostic checks and stability test.
Source: Compiled by the author (2023).
and rightly specified as the goodness of fit for the analysis. Again, a test for the model robustness was performed using two different regression methods as recommended by Boozer (1997) and MacCarthy (2021) to enhance the efficacy of the statistical analysis and also allow for the identification of complex behavioral patterns. As observed by Hermalin and Weisbach (2003) most of the challenges in corporate governance analysis are the issues of endogeneity and omitted variables. Whenever these problems occur in regression analysis the outcome is biased and the result becomes inconsistent. To overcome these challenges the study carried second regression and analysis using Generalised Method of Moments (GMM) as second regression analysis. The result of the GMM shows that CEO-non-duality, audit committee independence, board size and non-executive directors are positively related to bank’s performance. This implies that endogeneity assumption is not violated as confirmed in Table 6. Again, to check on the robustness of the result, the study replaced the NIM with PAT. The result obtained using PAT as the dependent variable was similar to NIM. This implies that the output from the random effect estimator is stable and it is devoid of heteroscedasticity and omitted variables challenges in this analysis. Therefore, the study proceeded to estimate the parameters that predict the relationship among the variables based on the result obtained from the random effect estimator in Table 4 and expressed as multiple regressions in Equation (3):
(3)
5. Conclusion and Recommendations
This study was undertaken purposely to assess the effect of corporate governance and audit committee independence on the bank’s performance. The regression assumption tested revealed that none of the underlying assumptions was violated: Normality, autocorrelation, endogeneity, and heterogeneity, and therefore the model exhibits goodness-of-fit to produce unbiased outcomes for this analysis. The study revealed a positive and significant relationship between CEO-non-duality, audit committee independence, non-executive director, and bank performance. These findings are well-aligned with agency theorists’ perspective that advocates for strengthening corporate governance mechanisms and audit committee independence to minimize agency conflicts arising out of the principal-agent relationship. The findings give support to CEO-non-duality. Agency theorists believed having one person to person to occupy the position of both CEO and the Chairman of a board may affect the bank’s performance negatively. However, the study failed to obtain a statistically significant relationship between board size and the bank’s performance. Therefore, the board size does not contribute significantly to the corporate governance mechanism of banks in Ghana. Since the board does not matter but the presence of non-executive directors on the board affects the bank’s performance, it implies that the composition of the board is what matters because it affects the independence of the board. The reason behind this inconsistency is that most of the board sizes are neither extremely large nor extremely low, making it difficult to make a significant effect on the bank’s performance. Overall, this study’s findings are consistent with agency theory that opined that corporate governance, as well as audit committee independence, are essential monitoring and control device that can mitigate information asymmetry, adverse selection, and moral hazard associated with management (i.e., agents) at the expense of owners (principal). The makes significant recommendations to the regulator and the entities (i.e., the banks) based on the findings from this study that would improve corporate governance and the bank’s performance nexus. Firstly, the regulator must ensure that more independent directors are appointed to audit committees and those appointed should have a background in accounting and finance to improve upon the audit committee function. Secondly, the regulator should legislate that all board chairmen should be independent directors and those who violate this provision should be heavily penalized in monetary to serve as a deterrent. Lastly, the banks should ensure that only independent directors chair the audit committee to enhance the independence of the audit committees of banks. There are two main limitations to this study. The first limitation of this study is that the study is a quantitative analysis between board sizes and bank performance. A future study involving mixed methodology could unravel the insignificant relationship between the board sizes and the bank’s performance. To serve as a guide to board composition in Ghana, until then it remains theoretical significance without empirical support in Ghana. The second limitation is that the study assesses only the direct relationship between corporate governance, the audit committee, and the bank’s performance but failed to assess the indirect relationship of audit committee independence through moderating and mediating analysis of the bank’s performance. Therefore, future research should include both moderating and mediating the role of audit committee independence between corporate governance mechanisms and the bank’s performance.