From risk to reward: Unveiling the multidimensional impact of financial risks on the performance of Ghanaian banks Authors: Benjamin Blandful Cobbinah, Wen Yang, Francis Atta Sarpong, George Nyantakyi Published: 2024 ABSTRACT The Ghanaian banking sector, grappling with a spectrum of financial risks, presents a compelling case study for understanding the dynamics of risk and profitability in emerging markets. This study seeks to fortify the financial performance of Ghanaian banks through an innovative application of benchmark regression analysis, focusing on critical financial risk and performance metrics. Employing an explanatory research methodology, we harnessed a panel regression model to scrutinize secondary data extracted from the annual income statements of 23 banks, spanning nearly two decades from 2006 to 2023. Our analytical arsenal encompassed the fixed effects model, the Generalized Method of Moments (GMM) within a fixed-effects framework, and Pooled Ordinary Least Squares (POLS) to ensure methodological rigor and robustness. Our findings illuminate the relationship between financial risks and Return on Equity (ROE). Specifically, credit risk was found to exert a significant positive influence on ROE, suggesting that judicious credit extension can be a conduit for profitability. In contrast, liquidity risk was identified as a determinant negatively impacting ROE, underscoring the imperative for effective liquidity management to safeguard banks' long-term solvency. Additionally, market risk was observed to have a positive association with ROE, indicating that strategic exposure to market volatility can potentially enhance financial performance. The implications of this study are manifold, offering actionable insights for policymakers and stakeholders in the banking sector. It advocates for a recalibrated approach to risk management, where banks proactively embrace credit risk while vigilantly addressing liquidity challenges. The study further posits that by forging strategic interconnections, banks can augment their risk management capabilities, thereby bolstering their financial performance and ensuring resilience in the face of market volatility. Introduction Financial risk, with its potential to induce unforeseen volatility in earnings, poses a significant challenge to the stability and performance of financial institutions, including those within Ghana's banking sector. This risk encompasses a spectrum of hazards such as market, credit, liquidity, operational, and legal risks [1]. Effective risk management is not only crucial for institutional performance but also engages a wide array of stakeholders, from regulatory bodies to customers [2,3]. The 2008–2009 financial crisis served as a stark reminder of the consequences of inadequate risk management, leading to increased regulatory scrutiny and a decline in investor confidence globally [4,5]. A constellation of metrics can assess financial performance, with Return on Equity (ROE) being a prominent example [6]. Discrepancies in financial risk reporting have been documented in various contexts, such as in Bangladeshian enterprises [2], and the detrimental impact of liquidity risk on asset returns in Nigeria has been observed [7]. A panel regression analysis of Ghana Stock Exchange (GSE)-listed commercial banks revealed a positive correlation between credit and market risks and ROE, while liquidity risk was inversely related to banking performance [8]. In response to sector-specific challenges, Ghana's banking industry has witnessed several transformative initiatives. A pivotal reform was the 2017 Bank of Ghana mandate, which raised the minimum capital requirement for commercial banks to GHS 400 million, thereby bolstering their financial fortitude [9]. Concomitantly, the government has targeted the non-performing loans predicament through mechanisms such as the establishment of a Credit Reference Bureau, enhancing the credit assessment process [10]. The Bank of Ghana has promulgated directives for the adoption of risk management methodologies, including the Basel II and III frameworks, aiming to refine the banking sector's risk management proficiency. While the impact of these measures on the industry's stability and performance is yet to be fully realized, they mark critical strides in tackling the sector's exigencies. The crux of Ghanaian banking success and resilience hinges on robust financial risk management. Regrettably, the Central Bank of Ghana has had to rescind the licenses of five banks in recent years, an outcome of their deficient risk management protocols [11]. It is imperative for Ghanaian banks to institute comprehensive risk management frameworks that adeptly identify, quantify, monitor, and mitigate risks [8]. Neglecting to do so could precipitate financial instability, investor disenchantment, and, in the direst cases, institutional collapse. The Ghanaian banking landscape has grappled with a surge in non-performing loans and bad debts, posing significant challenges to financial stability. A pivotal 2018 report from the Bank of Ghana underscored the imperative to manage credit risks, which accounted for a staggering 78.9 % of total loans and advances. Alarmingly, some institutions were found to have a capital deficit of GHC 7.4 billion, a figure egregiously below the mandated minimum of GHC 400 million. Despite the Central Bank of Ghana's prudential guidelines, which were designed to bolster commercial banks' return on assets, the sector's health demands further amelioration [12]. On August 1, 2018, the Bank of Ghana took decisive action, revoking the licenses of five banks due to risk management malpractices. This move precipitated the creation of a consolidated bank, highlighting the dire need for robust risk management frameworks to prevent such occurrences. Empirical evidence suggests a correlation between insufficient risk management and adverse outcomes such as bank failures, takeovers, and diminished profit margins. Liquidity risks have also imperiled some banks, hindering their capacity to meet financial obligations and impacting their returns on equity. Moreover, interest rate risks have been identified as a source of financial losses and suboptimal returns on assets. This study aims to shed light on the critical risk factors impacting the profitability of Ghanaian banks and to propose innovative solutions for performance enhancement. By employing innovative benchmark regression analysis, including fixed effects, the Generalized Method of Moments (GMM), and the 3SLS model, the research meticulously estimates the influence of financial risk on the financial performance of Ghanaian banks. The robustness of these models lies in their ability to elucidate the complex dynamics between credit risk, liquidity risk, market risk, and the banks' performance metrics. The findings of this study are instrumental in informing stakeholders about the risk factors exerting the most significant impact on bank performance. It offers valuable insights for regulators, policymakers, and bank management to craft effective risk management strategies. These strategies, in turn, can fortify the stability and sustainability of Ghana's banking sector, ultimately contributing to the broader economic growth and development of the nation [13]. This study aspires to enrich the discourse on financial risk management within the emerging markets, with a particular focus on the African context, which remains an under-explored area of research. Existing literature in Africa has predominantly concentrated on isolated financial risks, such as operational risk [7] and credit risk [[14], [15], [16]]. However, a comprehensive examination of the synergistic effects of credit, liquidity, and market risks on banks' financial performance is conspicuously absent from the African narrative. Moreover, the aforementioned studies have extended their gaze to diverse sectors including insurance companies and supermarkets, rather than isolating the banking industry. The dearth of research is particularly pronounced regarding the impact of market risk [2,[17], [18], [19]] and liquidity risk [20,21] on financial performance within Africa, with Ghana being no exception. The majority of studies on the ramifications of credit risk have predominantly centered on Asian banking institutions [[22], [23], [24]]. Given the struggles of numerous Ghanaian banks with financial risks—ranging from liquidity to credit and market risks—these challenges have precipitated a failure to meet financial obligations, culminating in diminished returns on equity. This scenario underscores the exigency for an investigation into the influence of credit risk, liquidity risk, and market risk on the Return on Equity (ROE) of Ghanaian banks, and the subsequent implications for their growth and expansion. It is imperative for Ghanaian banks to prioritize robust risk management strategies to ensure not only their long-term viability but also to bolster the nation's economic growth and development. In this vein, the analysis of financial risk factors impacting the performance of Ghana's commercial banks assumes a heightened significance. The paper is structured as follows: Section 2 provides a critical review of pertinent empirical evidence, identifying existing research gaps. Section 3 delineates the study's methodology and data sources. Section 4 presents a thorough discussion of the empirical findings and their implications for the banking sector. The paper culminates in Section 5, which offers conclusions, policy recommendations, and potential directions for future research endeavors. Literature review Theoretical underpinning A robust theoretical framework is essential for research as it provides a roadmap for understanding the relationships between variables under investigation [25]. This study anchors its analysis in three principal theories: the finance distress theory, the shiftability theory of liquidity, and the extreme value theory (EVT). These theories not only offer insights into the dynamics of financial risks such as credit, market, and liquidity risks but also underscore their transient effects on firm performance [26]. Finance Distress Theory: This concept is pivotal for identifying potential financial challenges. Financial distress is marked by an entity's inability to meet debt obligations, among other indicators [27]. It is crucial for managers to monitor these signs proactively to avert operational disruptions [[28], [29], [30], [31]]. The theory posits that financial stability, characterized by cash flows exceeding debt service requirements, is vital for long-term profitability and sustainability [32]. Shiftability Theory of Liquidity: This theory addresses the liquidity risk faced by enterprises, particularly focusing on the assets that banks can readily transfer or sell to maintain liquidity [33]. The concept of "shiftability" suggests that banks with strong ties to other financial institutions may not need to maintain high levels of liquidity, as they can rely on these relationships in times of need [34]. However, this theory also acknowledges that relying solely on asset shiftability may not be sufficient to ensure the liquidity of the financial system [33]. The notion of Shiftability postulates that all banks must maintain such assets that can be transferred to the central bank, which serves as the lender of last resort. This theory states that if banks have a significant amount of assets that they can transfer to other banks for cash in an emergency without incurring any losses, they do not need to rely on maturities [35]. In instances where liquidity is constrained, banks may need to transfer their assets to another bank with higher liquidity [36]. By retaining transferable financial assets in the open market, such as government securities, and having self-liquidating bonds accessible to banks connected with assets, it is postulated that banks can diversify the nature of their assets. However, it is contended that the hypothesis may be flawed in that the shiftability of assets alone does not suffice to supply liquidity to the financial system [33]. Extreme Value Theory (EVT): EVT is instrumental for assessing extreme market events, which is critical for market risk management and internal risk assessment [38]. The incorporation of EVT can enhance the precision of Value at Risk (VaR) calculations, a key tool for financial risk assessment [39]. The study will apply EVT to evaluate the impact of extreme market movements on shareholder returns. This study delves into the efficacy of liquidity management on shareholder returns within commercial banks, contextualized within the regulatory mandates of the Basel III framework. Specifically, it examines the impact of net stable funding ratios and liquidity coverage regulations on bank performance. The Shiftability Theory posits that a bank's liquidity can be enhanced by holding assets that can be swiftly liquidated in times of need, particularly when supported by a responsive central bank and discount market [37]. However, this theory's validity is scrutinized, as it potentially overlooks the significance of liabilities as an alternative source of liquidity, especially during periods of financial stress [33]. In parallel, the Basel III framework's guidelines are instrumental in shaping liquidity policies that this study aims to assess for their influence on shareholder returns. The study also incorporates the Extreme Value Theory (EVT) to analyze the behavior of extreme market movements and their implications for risk management. EVT provides a sophisticated lens for evaluating the tails of return distributions, thereby enhancing the precision of Value at Risk (VaR) calculations [38,39]. The study leverages EVT to refine the VAR estimation, employing a dynamic approach that integrates EVT with GARCH models, as demonstrated in recent literature during significant market disruptions, such as the pandemic outbreak [40]. While EVT is adept at capturing extreme events, it is recognized that its focus on distribution tails may lead to distinct VAR predictions for left and right tails [41]. The conditional EVT (C-EVT) framework is employed to estimate VaR quantiles and generate out-of-sample forecasts, offering a nuanced assessment of market risk and its capital requirements [42,43]. The selection of study variables and the empirical analysis are grounded in these theoretical perspectives, which are further elaborated in subsequent sections to demonstrate their applicability to the research objectives. Nexuses between credit risk, liquidity risk, market risk, and bank performance Effective credit risk monitoring and regulation are paramount for mitigating default rates and enhancing bank performance [44,45]. Credit risk emanates from a variety of financial instruments, including loans, interbank lending, and trade finance [14], and its management is vital for reducing non-performing loans and bolstering profitability [15,46]. Post-credit extension, there is an inherent risk that borrowers may default on their obligations, leading to an increase in banks' gross non-performing asset ratio [47]. Accurate forecasting of credit risk and ongoing assessment of borrowers' creditworthiness are essential for financial institutions to safeguard their assets, make informed lending decisions, and maintain a robust lending operation [48]. The nexus between non-performing loans and financial performance, measured through Return on Assets (ROA), has been well-documented [49,50]. Studies have revealed a significant impact of credit risk on a bank's financial performance, with credit risk factors such as impairments and non-performing loans showing positive correlations with bank profitability metrics like ROA and Return on Equity (ROE) [21]. Conversely, elevated levels of non-performing loans and provisions can erode profitability [22,23]. Interestingly, a higher loan-to-asset ratio, within a balanced risk framework, can potentially increase interest income and profitability [24]. Drawing on the reviewed literature, this study posits that effective credit risk management is pivotal to the profitability of banks in Ghana. The negative implications of non-performing loans on profitability are well-established, underscoring the necessity for robust credit risk management practices to enhance bank profitability. Given this backdrop, the study hypothesizes that.H1Credit risk significantly affects the performance of banks in Ghana. Credit risk significantly affects the performance of banks in Ghana. Liquidity, as the capacity to fulfill short-term financial commitments, is a fundamental measure of a company's financial health [21]. The banking sector is particularly vulnerable to liquidity risks, which can arise from unforeseen cash outflows or the inability to convert assets into cash at a reasonable cost [47]. The Basel Committee's 2008 assessment highlighted that financial institutions are considered liquid if they can access funds at a reasonable cost, indicating the importance of liquidity risk management in the industry [52]. Empirical research has explored the relationship between liquidity and financial performance. For instance, a study by Ref. [52] examining Ghanaian publicly traded banks from 2005 to 2010 reported a decline in both liquidity and profitability, with a weak positive correlation that was statistically insignificant. This finding suggests a complex relationship between liquidity and profitability that may vary by context. Cross-sector comparisons provide additional insights. For example [16], found that operational and liquidity risks negatively impacted the financial performance of supermarkets in Nairobi, while [7] observed a positive relationship between operational risk and Return on Assets (ROA) in Nigerian insurance companies, but a negative impact from liquidity risk. Extending the analysis to overseas Chinese banking corporations in Singapore [20], found that operational efficiency, which includes effective liquidity management, positively affects profitability. Similarly [51], reported that for Western Balkan commercial banks, certain liquidity management indicators, such as the loan-to-total deposits ratio and cash reserves, were positively correlated with ROA, while the current ratio showed a negative correlation. The ability of commercial banks to convert short-term deposits into long-term loans is a continuous challenge that necessitates vigilant monitoring and effective liquidity risk management strategies. Given the critical role of liquidity in financial performance, it is imperative for institutions to regularly assess their liquidity positions and to have contingency plans in place to address potential shortages. Building on this literature, the current study aims to evaluate the impact of liquidity risk on the financial performance of commercial banks in Ghana. The study hypothesizes that.H2Liquidity risk has a positive impact on the performance of commercial banks in Ghana Liquidity risk has a positive impact on the performance of commercial banks in Ghana Market risk is a pivotal factor influencing the financial stability of international organizations, encompassing foreign exchange, interest rate, and stock price risks [17]. Interest rate risk, in particular, gauges the effect of interest rate fluctuations on a bank's profitability or net worth. Conversely, foreign exchange risk materializes when unpredictable shifts in exchange rates alter the domestic currency amount required to service foreign currency-denominated loans. The extant literature presents a mixed perspective on the relationship between market risk and financial performance. While some studies, such as [2], have identified a positive correlation between financial risk disclosures—including interest rate and foreign exchange risks—and company performance, others like [18] have reported a detrimental effect of market risk on the financial performance of non-financial companies. In the Ethiopian banking sector [53], found a significant positive impact of interest rate, credit, and liquidity risks on profitability. However [19], observed no correlation between market risk and bank return on assets, indicating that the influence of market risk can vary by industry and risk type. Given these incongruent findings, this study aims to scrutinize the relationship between market risk and financial performance within the Ghanaian banking industry, with a specific focus on foreign exchange, interest rate, and stock price risks. The study posits the following hypothesis.H3Market risks positively influence the performance of commercial banks in Ghana. Market risks positively influence the performance of commercial banks in Ghana. Conceptual framework The conceptual framework of a study is a critical component that guides the research design and analysis. This study's conceptual framework is developed based on the research hypotheses seen in Fig. 1. The study's primary objective was to investigate the impact of liquidity, credit, and market risk on the return on equity (ROE) of commercial banks in Ghana. The framework also considered the interconnections between these variables.Fig. 1Conceptual model.Fig. 1 Conceptual model. The financial performance of banks is commonly gauged by Return on Equity (ROE), which reflects the return on shareholders' investments [54]. To safeguard operations and investor funds, commercial banks must maintain an adequate level of assets [55]. Credit risk materializes when customers default on their contractual obligations, affecting financial statements when non-performing loans escalate to significant levels [54]. Liquidity risk, a critical component of bank stability, is typically assessed using the liquidity ratio—current assets divided by current liabilities [58]. Market risks, including exchange rate and interest rate fluctuations, are examined as indicators of market risk [57]. The interest rate, in particular, is a pivotal market risk measure due to its potential to disrupt the alignment between interest paid on deposits and interest earned on loans [57]. The net interest margin, indicative of interest rate risk, can influence a company's performance, either positively or negatively [54]. Financial leverage, measured by the Degree of Financial Leverage (DFL), helps determine the appropriate debt level and its impact on operating income, with a higher DFL implying greater financial risk [57]. Foreign exchange risk, another dimension of market risk, is assessed by its impact on a bank's daily operating profit or loss, as it can affect ROE through currency fluctuations [58]. Managing liquidity risk is essential as it can significantly influence ROE. Similarly, credit risk, quantified by the volume of non-performing loans, is a critical factor that can impact ROE [58]. The findings of this research contribute valuable insights into the impact of financial risk on commercial banks' performance and enhance our understanding of effective financial risk communication. Given the dynamic nature of financial markets, it is imperative to periodically revisit the relationship between financial risk and bank performance to account for emerging research and potential interconnections between variables [56]. Materials and methods This study employs an explanatory research design to investigate the impact of financial risks on shareholder value, growth, and development in Ghanaian commercial banks. The primary objective is to establish cause-and-effect relationships between financial risks—credit risk, liquidity risk, and market risk (including interest rate and foreign exchange risks)—and bank performance as illustrated in Table 1. The analytical model developed for this study is informed by previous research demonstrating the efficacy of explanatory designs in uncovering the underlying reasons for observed relationships [[59], [60], [61]].Table 1Variable measurement.Table 1VariableMeasurementSourceDependent variableFinancial PerformanceReturn on equity (ROE)Central Bank of Ghana (BoG)Return on Assets (ROA)Central Bank of Ghana (BoG)Independent VariablesCredit RiskNon-performing loans/DepositsCentral Bank of Ghana (BoG)Liquidity RiskCurrent asset/Current liabilityCentral Bank of Ghana (BoG)Market RiskInterest Rate Risk (Interest Income -Interest Expenses/Total Earning Asset)Central Bank of Ghana (BoG)Foreign Exchange Risk(Exchange gain/Exchange Loss)Central Bank of Ghana (BoG)Control variablesBoard SizeLogSizeCentral Bank of Ghana (BoG)Board IndependenceLogBODCentral Bank of Ghana (BoG)OwnershipLogOWNCentral Bank of Ghana (BoG)Age of FirmLogAGECentral Bank of Ghana (BoG)LeverageLogLEVCentral Bank of Ghana (BoG) Table 1Variable measurement.Table 1VariableMeasurementSourceDependent variableFinancial PerformanceReturn on equity (ROE)Central Bank of Ghana (BoG)Return on Assets (ROA)Central Bank of Ghana (BoG)Independent VariablesCredit RiskNon-performing loans/DepositsCentral Bank of Ghana (BoG)Liquidity RiskCurrent asset/Current liabilityCentral Bank of Ghana (BoG)Market RiskInterest Rate Risk (Interest Income -Interest Expenses/Total Earning Asset)Central Bank of Ghana (BoG)Foreign Exchange Risk(Exchange gain/Exchange Loss)Central Bank of Ghana (BoG)Control variablesBoard SizeLogSizeCentral Bank of Ghana (BoG)Board IndependenceLogBODCentral Bank of Ghana (BoG)OwnershipLogOWNCentral Bank of Ghana (BoG)Age of FirmLogAGECentral Bank of Ghana (BoG)LeverageLogLEVCentral Bank of Ghana (BoG) Variable measurement. The study evaluates the financial performance of 23 Ghanaian banks over the period from 2006 to 2023. The banks selected for this study are described in Appendix A. Data were collected from the banks' annual financial statements, audited reports, and information provided by the central bank of Ghana. To mitigate the risk of aggregation bias, panel data analysis was utilized, a method endorsed for its ability to control for unobserved heterogeneity across entities [63,64]. The study's sample is comprised of these 23 banks over the specified seventeen-year period, providing a robust dataset for analysis. The research period was chosen to reflect recent industry trends and regulatory changes within Ghana's banking sector. The explanatory research design, even in the absence of experimental control over variables, facilitates a deep understanding of the social and economic issues at hand, aligning with the approach advocated by Ref. [62]. Variable measurement Dependent variable The study utilizes Return on Equity (ROE) as the primary metric to assess the financial performance of commercial banks in Ghana. ROE is a critical indicator in the banking sector due to the asset and income-intensive nature of banks [65]. It measures the return on shareholders' equity and is a key benchmark for evaluating a bank's profitability in relation to the capital employed. In addition to ROE, other metrics such as Return on Investment (ROI) provide insights into a bank's efficiency in generating profits from invested funds. Operational profit margins and capitalization levels also influence ROE, reflecting the bank's operational efficiency and financial structure. The Risk-Weighted Capital Adequacy Ratio (RWCAR), or Tier 1 capital adequacy ratio, is another important measure that reflects the strength of a bank's capital position. A strong capital position, as indicated by a higher RWCAR, can support sustainable growth in ROE. However, it is also important to note that an increase in ROE can be driven by higher earnings or returns, but it may also be a consequence of increased risk, such as taking on additional debt. The calculation of ROE is performed by dividing net income by total equity, providing a direct measure of the bank's profitability. This metric is essential for understanding the effectiveness with which banks generate profits from their equity and for comparing the performance of banks within the industry.(1)ReturnofEquity=NetIncomeStakeholder′sEquity Independent variables Credit risk is the potential for a borrower to fail to meet their debt obligations, leading to loan default. Default events can impair a bank's ability to recover the principal and interest owed, resulting in financial losses, cash flow disruptions, and increased costs associated with collection efforts. The magnitude of credit risk is often reflected in borrowing costs, such as yield spreads, which tend to be higher for borrowers perceived to carry a greater risk of default. This risk metric is a critical component of a bank's risk assessment process and is integral to setting the terms and conditions of lending agreements.(2)CreditRisk=nonperformingloansDeposit Liquidity risk emerges when a commercial bank encounters difficulty in meeting its short-term financial obligations or providing loans and services due to insufficient liquidity [66]. This risk can stem from various factors, including a downgrade in the institution's credit rating, sudden cash outflows, challenges in securing trades or loans from counterparties, or a broader market liquidity crunch. The interplay between market and funding liquidity further intensifies liquidity risk, complicating the bank's ability to liquidate collateral and raise funds. When liquidity is scarce, the impact of other financial risks is amplified. In such scenarios, a bank may be compelled to seek alternative borrowing sources to fulfill payment obligations if a counterparty fails to meet theirs. Should the bank be unable to secure the necessary funding, it may face the risk of default, which in turn can escalate credit risk [66].(3)LiquityRisk=CurrentassetCurrentliability This study encompasses two pivotal components of market risk: interest rate risk and foreign exchange risk. Interest rate risk materializes from fluctuations in interest rates, which can significantly affect the valuation of financial instruments, particularly for bondholders. The vulnerability of bonds to interest rate changes is influenced by factors such as coupon rates and the time to maturity. To quantify interest rate risk, the Net Interest Margin (NIM) is employed as a proxy, reflecting the spread between the interest income earned by banks and the interest expenses paid to lenders, adjusted for interest rate variations. A higher NIM suggests a bank is less susceptible to the adverse effects of interest rate risk [55]. Foreign exchange risk emerges when a financial transaction is executed in a currency other than the organization's functional currency, exposing the organization to potential losses due to unfavorable exchange rate movements prior to transaction settlement. Additionally, this risk is pertinent when a subsidiary operates and maintains financial records in a currency different from the parent company's reporting currency. Foreign exchange risk is a critical consideration for investors and businesses engaged in international trade and investment. While this risk can impact the value of international operations, entities can implement various strategies to hedge against or mitigate its effects.(4)InterestRateRisk=Interestincome−InterestExpensesTotalEarningAsset(5)ForeignExchangeRisk=ExchangegainExchangeloss Control variables Drawing from a comprehensive literature review and practical insights into financial risks, which encompass credit, liquidity, and market risks, this study recognizes that certain business characteristics may influence both the dependent variable and the three primary independent variables under investigation. To mitigate the potential for omitted variable bias, which can arise from multicollinearity, the study incorporates seven control variables that are theorized to have a significant association with financial performance in the banking sector [67,68]. The control variables selected for this analysis are age, size, ownership, board composition, Return on Assets (ROA), and leverage. Each of these variables represents a distinct aspect of a bank's operations and strategic positioning that could potentially affect its financial risk exposure and performance. For instance, the age of a bank may reflect its maturity and stability in the market, while size can indicate economies of scale and market power. Ownership structure can influence governance practices and decision-making processes, board composition may affect strategic direction and oversight, and ROA is a measure of profitability that can be impacted by risk management effectiveness. Leverage, on the other hand, reflects the capital structure and the degree of financial risk undertaken by the bank. By controlling for these variables, the study aims to provide a more accurate estimation of the relationships between financial risks and bank performance, ensuring that the findings are robust and not confounded by other influential factors. Model specification and estimation This study employs a combination of multiple panel regression analysis and descriptive statistics to scrutinize the impact of financial risks on commercial banks in Ghana. The analytical process leverages STATA's advanced capabilities for multiple linear panel regression models, which are well-suited for panel data and allow for the examination of both fixed and random effects over time. The dataset comprises 17 years of financial records from 23 commercial banks in Ghana, ensuring a robust sample size for the analysis. Prior to the regression analysis, the data underwent meticulous organization and preliminary computation using Microsoft Excel. This prepared the data for import into both SPSS and STATA, where it was subjected to rigorous statistical analysis. Four key financial risks were examined within the scope of this study: return on equity (ROE), credit risk, liquidity risk, and market risk. To provide a comprehensive overview of the data distribution and central tendencies, descriptive statistics including mean, standard deviation, maximum, and minimum values were calculated and are reported [69]. The regression analysis was designed to reveal the influence of the independent variables—credit risk, liquidity risk, and market risk—on the dependent variable, ROE, which serves as an indicator of the profitability of Ghanaian commercial banks. Mathematically, the relationship can be expressed in the following general form:(6)Yi,t=β0+β1X1i,t+β2X2i,t+β3X3i,t+Ei,tWhere X1, X2, and X3 represent the independent variables that are decomposed into Credit Risk (CR), Liquidity Risk (LR), and Market Risk (MR), respectively. These independent variables predict the outcome of the dependent variable, ROE.(7)ROEi,t=β0+β1CRi,t+β2LRi,t+β3MRi,t+Ei,t In line with previous studies and assuming a linear relationship, the following regression model was used to test the hypotheses and determine the relationship between the dependent variable (ROE), independent variables (CR, LR, and MR), and control variables(8)ROEit=β0+β1CRi,t+β2LRi,t+β3MRi,t+∑i=1nβControlsit+εitWhere, ROE = Return on Equity; CR= Credit Risk; LR = Liquidity Risk; MR = Market Risk; SIZE = company size; BOD = Independence of the Board; OWN = Ownership; AGE = Age of the Firm; ROA = Return on Assets; LEV = leverage; ε = Error term. The study included the entity fixed effects and year fixed effects in the equation for robustness check. Fixed effects in the model can help control for unobserved heterogeneity that may affect the relationship between the independent variables and the dependent variable, as seen in the following equations:(9)ROEi,t=β0+β1CRi,t+β2LRi,t+β3MRi,t+γit+δit+∑i=1nβControlsit+εitWhere ROE = Return on Equity; CR= Credit Risk; LR = Liquidity Risk; MR = Market Risk; SIZE = company size; γ = bank fixed-effects; δ = ; year fixed-effect; BOD = Independence of the Board; OWN = Ownership; AGE = Age of the Firm; ROA = Return on Assets; LEV = leverage; ε = Error term. To account for unobserved heterogeneity among firms that remains constant over time, the bank fixed-effects (γit) were introduced. These fixed effects are represented by a set of dummy variables that assume 1 for each bank and 0 for others. Similarly, to capture the unobserved heterogeneity common to all banks over time, the year-fixed effects (δit) were introduced. A set of dummy variables represents these fixed effects, with each year assigned a value of 1 and others assigned a value of 0. The study further employed the Generalized Moment of Methods for analyzing the panel data collected as guided by Refs. [70,[83], [84], [85], [86]] study using GMM. The study's models are mathematically computated as:(10)lnROEi,t=β0+β1lnCRi,t+β2lnLRi,t+β3lnMRi,t+∑i=1nβControlsit+εitWhere β0 is the intercept, β1, β2, β3 are the coefficient of explanatory variables, CR, LR and MR, respectively, where i is entities (banks), and t represents the year (t = 1, 2, 3, 4, 5), ε is the error. For the controls, it represents the controlled variables such as lnSIZE, lnBOD, lnOWN, lnAGE, lnROA and lnLEV. Main results and discussions The findings of the regression study revealed that the financial performance of commercial banks in Ghana, as measured by the dependent variable (ROE), was influenced by several independent variables, including credit risk, liquidity risk, and market risk. The study found that credit risk had a significant negative impact on the financial performance of commercial banks, while market risk had a positive effect. However, the study found no discernible effect of liquidity risk on the financial performance of commercial banks in Ghana. To ensure the accuracy and reliability of the regression model, the researchers conducted several diagnostic tests before running the model. These tests included the normality test, multicollinearity, autocorrelation and heteroscedasticity, panel unit root test, and Hausman test for random and fixed effects. These tests were crucial in identifying potential issues that may affect the quality of the dependent and independent variables used in the study, as noted by Refs. [71,72,83,86]. Diagnostic test Normality, multicollinearity, and autocorrelation To ensure the robustness of the panel regression analysis, the study conducted a series of diagnostic tests to assess normality, multicollinearity, and autocorrelation. The Jarque-Bera test was utilized to evaluate the normality of the data distribution, a fundamental assumption for regression analysis [[73], [74], [75]]. The presence of multicollinearity was investigated using the Variance Inflation Factor (VIF) and tolerance values. A VIF above 10 suggests a significant linear relationship between the independent variables, indicating multicollinearity [73,76]. Multicollinearity can distort the regression coefficients and their interpretation. The tolerance value and VIF analysis indicated no multicollinearity in the dataset, as all VIF values were below the threshold of 10 [71,72]. The Durbin-Watson test was employed to detect any autocorrelation in the residuals of the regression model. A score between 1.5 and 2.5 typically indicates no autocorrelation [79]. The test results, as presented in Table 2, confirmed the absence of autocorrelation in the data.Table 2Diagnostic test.Table 2lnROElnCRlnLRlnMRlnSIZElnBODlnOWNlnAGElnROAlnLEVlnROE1lnCR0.139∗∗1lnLR0.049−0.4091lnMR0.0190.073−0.0321lnSIZE0.072∗0.201∗0.019∗∗−0.119∗1lnBOD0.0290.1030.1080.0470.0221lnOWN0.1230.0830.019∗0.038∗∗−0.1140.0481lnAGE0.0180.169∗∗0.2130.0140.2500.284∗∗0.0141lnROA0.1420.0580.0490.0050.0360.1030.0330.052∗∗1lnLEV0.0050.109∗−0.0020.015∗0.148∗∗0.249∗∗0.1450.140∗0.019∗1Jarque-Bera7.2754.2647.4245.4397.2768.3558.9584.7795.6504.701Prob.0.003∗∗0.002∗∗0.1030.000∗∗0.001∗∗0.015∗∗0.002∗0.003∗∗0.000∗∗0.0004∗∗VIF2.4892.0183.0192.0582.1441.9212.4023.2493.012Tolerance (1/VIF)0.4020.4960.3310.4860.4660.5210.4160.3080.332DW2.080Breusch-Pagan test0.149Note: ∗∗∗, ∗∗, and ∗ denote significance at 1 %, 5 %, and 10 % levels, respectively. Table 2Diagnostic test.Table 2lnROElnCRlnLRlnMRlnSIZElnBODlnOWNlnAGElnROAlnLEVlnROE1lnCR0.139∗∗1lnLR0.049−0.4091lnMR0.0190.073−0.0321lnSIZE0.072∗0.201∗0.019∗∗−0.119∗1lnBOD0.0290.1030.1080.0470.0221lnOWN0.1230.0830.019∗0.038∗∗−0.1140.0481lnAGE0.0180.169∗∗0.2130.0140.2500.284∗∗0.0141lnROA0.1420.0580.0490.0050.0360.1030.0330.052∗∗1lnLEV0.0050.109∗−0.0020.015∗0.148∗∗0.249∗∗0.1450.140∗0.019∗1Jarque-Bera7.2754.2647.4245.4397.2768.3558.9584.7795.6504.701Prob.0.003∗∗0.002∗∗0.1030.000∗∗0.001∗∗0.015∗∗0.002∗0.003∗∗0.000∗∗0.0004∗∗VIF2.4892.0183.0192.0582.1441.9212.4023.2493.012Tolerance (1/VIF)0.4020.4960.3310.4860.4660.5210.4160.3080.332DW2.080Breusch-Pagan test0.149Note: ∗∗∗, ∗∗, and ∗ denote significance at 1 %, 5 %, and 10 % levels, respectively. Diagnostic test. Note: ∗∗∗, ∗∗, and ∗ denote significance at 1 %, 5 %, and 10 % levels, respectively. White's test was used to examine the presence of heteroscedasticity, which occurs when the variability of the error term differs across observations [73]. The null hypothesis of homoscedasticity (constant variance of the error term) was not rejected, as the p-value exceeded the standard significance level of 0.05, suggesting no heteroscedasticity in the model (Table 2). The condition index, another measure of multicollinearity, was also considered. A condition index below 15 indicates no collinearity, while an index above 30 suggests significant multicollinearity. The study's correlation matrix, as depicted in Table 2, revealed significant correlations between ROE and the independent variables, further supporting the validity of the regression analysis. The statistical software EViews was used to perform these diagnostic tests, ensuring that the regression model was free from common statistical issues that could compromise the reliability of the findings. Table 2 presents the results of diagnostic tests conducted on the panel data, including a correlation matrix, normality assessment via the Jarque-Bera test, multicollinearity assessment with Variance Inflation Factor (VIF) and tolerance values, and tests for autocorrelation (Durbin-Watson) and heteroscedasticity (Breusch-Pagan) [[77], [78]]. The correlation matrix reveals several significant relationships among the variables. For instance, lnSIZE shows significant correlations with lnCR, lnLR, and lnMR at the 0.05 level, and lnOWN is significantly related to lnAGE and lnLEV at the 0.01 level. These correlations are important to consider when interpreting the effects of independent variables on the dependent variable, lnROE. Normality of the data was assessed using the Jarque-Bera test, with all p-values indicating that the data do not significantly deviate from a normal distribution. Specifically, the probabilities range from 0.0004 for lnLEV to 0.103 for lnLR, with most values being highly significant, suggesting normality. In terms of multicollinearity, the VIF values are all below the threshold of 10, suggesting that multicollinearity is not a concern in the regression model. The highest VIF value of 3.249 is observed for lnLEV, while the lowest VIF value of 1.921 is for lnBOD. Tolerance values, which are the reciprocals of the VIFs, further support this conclusion, with all values well above the acceptable threshold. The Durbin-Watson statistic of 2.080 indicates the absence of autocorrelation in the residuals, which is within the acceptable range of 1.5–2.5. Lastly, the Breusch-Pagan test result of 0.149 with a p-value greater than 0.05 provides evidence against the presence of heteroscedasticity in the model. Overall, the diagnostic tests' outcomes in Table 2 confirm that the statistical assumptions for the panel regression analysis have been met, ensuring the validity of the study's regression results. Panel unit root test Given that the majority of the variables exhibited non-stationarity prior to the regression analysis, the Conductance for a Unit Root (CADF) test was employed to determine the stationarity of the variables [70]. The results, as presented in Table 3, indicate that at the 5 % level of significance, all variables became stable after taking the first difference, signifying the absence of unit roots at the first differenced form.Table 3Stability test.Table 3VariablesCADFPPIntegrationLevel1st Diff.Level1st Diff.lnCR−1.302−2.449∗∗−0.339−2.049∗∗I (1)lnLR−3.149−1.859∗∗∗−1.469−1.639∗∗I (1)lnMR−3.252−3.049∗∗−2.038−2.068∗∗I (1)lnFP−2.417−2.149∗∗−1.039−2.138∗∗I (1)lnSIZE−2.329−2.048∗∗−2.058−3.022∗∗I (1)lnBOD−2.035−3.043∗∗−1.492−2.152∗∗∗I (1)lnOWN−3.405−3.592∗∗−2.005−1.408∗∗I (1)lnAGE−1.405−2.683∗∗−1.593−2.329∗∗I (1)lnROA−2.035−1.836∗∗−2.149−2.083∗∗I (1)lnLEV−1.845−2.893∗∗−0.255−2.202∗∗∗I (1)Note: ∗∗∗, ∗∗, and ∗ denote significance at 1 %, 5 %, and 10 % levels, respectively. Table 3Stability test.Table 3VariablesCADFPPIntegrationLevel1st Diff.Level1st Diff.lnCR−1.302−2.449∗∗−0.339−2.049∗∗I (1)lnLR−3.149−1.859∗∗∗−1.469−1.639∗∗I (1)lnMR−3.252−3.049∗∗−2.038−2.068∗∗I (1)lnFP−2.417−2.149∗∗−1.039−2.138∗∗I (1)lnSIZE−2.329−2.048∗∗−2.058−3.022∗∗I (1)lnBOD−2.035−3.043∗∗−1.492−2.152∗∗∗I (1)lnOWN−3.405−3.592∗∗−2.005−1.408∗∗I (1)lnAGE−1.405−2.683∗∗−1.593−2.329∗∗I (1)lnROA−2.035−1.836∗∗−2.149−2.083∗∗I (1)lnLEV−1.845−2.893∗∗−0.255−2.202∗∗∗I (1)Note: ∗∗∗, ∗∗, and ∗ denote significance at 1 %, 5 %, and 10 % levels, respectively. Stability test. Note: ∗∗∗, ∗∗, and ∗ denote significance at 1 %, 5 %, and 10 % levels, respectively. Table 3 shows that in their level form, the variables lnCR, lnLR, lnMR, lnFP, lnSIZE, lnBOD, lnOWN, lnAGE, lnROA, and lnLEV were not stable (had unit roots), as evidenced by their CADF and PP test statistics not reaching the significance levels. However, upon first differencing, all variables achieved stability, as indicated by the significant CADF and PP test statistics at the 1 %, 5 %, and 10 % levels. The decision to use the fixed effects model in estimating the relationship between financial risks and financial performance was validated by these findings, as they ensured that the variables were stationary in their first differenced form. This approach mitigates the potential issues associated with non-stationary data in regression analysis. Hausman test for random and fixed effects To determine the most suitable panel data model, researchers conducted the Hausman test to distinguish between the random-effects and fixed-effects models [69]. The results of this test, detailed in Table 4, indicate a preference for the fixed-effects model over the random-effects model when there is no evidence of individual effects being correlated with the explanatory variables. The chi-squared statistic from the Hausman test was 1.09, with an associated p-value of 0.9855, which is well above the common 5 % significance threshold. This suggests that there is no significant difference between the fixed and random effects estimates, contrary to the initial conclusion stated.Table 4Test of Hausman for random and fixed effects.Table 4Variables(b) fixed(B) random(b-B)Differencesqrt(diag(V_bV_B)) S.E.lnCR−0.0239−0.02530.00140.0043lnLR0.0528−0.04940.10220.0028lnMR0.23580.2417−0.00590.0201lnSIZE−0.2856−0.55840.27280.0018lnBOD0.33160.28990.04170.0032lnOWN0.01290.0838−0.07090.0021lnAGE0.03560.0649−0.02930.0032lnROA0.04160.0370.00460.0013lnLEV0.09070.07010.02060.0018 Table 4Test of Hausman for random and fixed effects.Table 4Variables(b) fixed(B) random(b-B)Differencesqrt(diag(V_bV_B)) S.E.lnCR−0.0239−0.02530.00140.0043lnLR0.0528−0.04940.10220.0028lnMR0.23580.2417−0.00590.0201lnSIZE−0.2856−0.55840.27280.0018lnBOD0.33160.28990.04170.0032lnOWN0.01290.0838−0.07090.0021lnAGE0.03560.0649−0.02930.0032lnROA0.04160.0370.00460.0013lnLEV0.09070.07010.02060.0018 Test of Hausman for random and fixed effects. Upon reviewing the Hausman test results, it is evident that the differences in coefficient estimates (b - B) for the variables lnCR, lnLR, lnMR, lnSIZE, lnBOD, lnOWN, lnAGE, lnROA, and lnLEV are each within their respective standard errors, as shown in the table. The small differences and high p-value indicate that the fixed-effects model is not statistically superior to the random-effects model, and the choice between them may be based on other considerations such as the study design or theoretical framework. Summary statistics The study quantitatively assessed the financial performance of Ghanaian commercial banks, leveraging Return on Equity (ROE) as a pivotal metric as seen in Table 5. The natural logarithm of ROE (lnROE) demonstrated a mean of 2.492, accompanied by a standard deviation of 0.458, reflecting moderate fluctuations in profitability. Credit risk, operationalized through the natural logarithm of credit risk (lnCR), exhibited a heightened mean of 4.7669, coupled with a positive skewness, indicating a propensity for high values in select cases.Table 5Descriptive statistics for the data.Table 5VariableMeanSDMinMaxSkewnessKurtosislnROE2.4920.4580.21740.5759−0.33430.3433lnCR4.76690.5442.35885.73012.87292.9263lnLR2.1720.6890.93172.3244−0.00120.834lnMR6.48641.1533.91632.4449−0.07651.0199lnSIZE4.38051.1940.88223.70110.5742.9841lnBOD4.21450.8690.29091.70131.5740.7863lnOWN3.84751.1753.09675.20111.97992.2891lnAGE3.02151.1972.09622.89020.97150.8792lnROA2.95051.3050.49312.39020.82262.2783lnLEV2.49551.1012.1763.30651.4632.782 Table 5Descriptive statistics for the data.Table 5VariableMeanSDMinMaxSkewnessKurtosislnROE2.4920.4580.21740.5759−0.33430.3433lnCR4.76690.5442.35885.73012.87292.9263lnLR2.1720.6890.93172.3244−0.00120.834lnMR6.48641.1533.91632.4449−0.07651.0199lnSIZE4.38051.1940.88223.70110.5742.9841lnBOD4.21450.8690.29091.70131.5740.7863lnOWN3.84751.1753.09675.20111.97992.2891lnAGE3.02151.1972.09622.89020.97150.8792lnROA2.95051.3050.49312.39020.82262.2783lnLEV2.49551.1012.1763.30651.4632.782 Descriptive statistics for the data. Liquidity and market risks, represented by lnLR and lnMR, respectively, showed means of 2.172 and 6.4864, with lnMR portraying greater variability, suggesting diverse strategies in market exposure management. Bank size, as denoted by lnSIZE, revealed a mean indicative of modest scale, yet with evidence of some significantly larger entities. Governance structures, including board size (lnBOD) and ownership (lnOWN), displayed moderate dimensions, with variations suggesting a range of organizational forms. The natural logarithm of age (lnAGE) conveyed a mean that points to an average tenure for the banks within the sample, while the standard deviation underscored the diversity in their maturity levels. Return on assets (lnROA) and leverage (lnLEV), with means of 2.9505 and 2.4955, respectively, depicted a broader spectrum of asset management and capital leverage approaches, underpinned by a higher variability in ROA. The descriptive statistics underscore the banks' generally profitable status, with variations in risk exposure and management practices that collectively reflect a nuanced financial landscape in Ghana's banking sector. Empirical results Baseline regression The empirical analysis was conducted utilizing a fixed-effects panel regression model to scrutinize the impact of financial risks on the performance of commercial banks in Ghana as seen in Table 6. The regression models, as depicted in equations (7), (8), (9), assumed a linear relationship between the dependent variable—Return on Equity (ROE)—and the independent variables, which included Credit Risk (CR), Liquidity Risk (LR), and Market Risk (MR). The models were adjusted to incorporate control variables such as company size (SIZE), board independence (BOD), ownership (OWN), firm age (AGE), return on assets (ROA), and leverage (LEV), alongside entity and year fixed effects to control for unobserved heterogeneity.Table 6Effect of financial risk on bank performance.Table 6VariablesModel 1Model 2Model 3Model 4Model 5Main Dependent Variable is ROElnCR1.7763 ∗∗0.0142 ∗∗0.2259∗∗0.2773∗∗0.1370∗(2.751)(2.2263)(2.765)(2.152)(1.984)lnLR−0.0011∗∗−0.0248∗∗−0.1119∗∗−0.0241∗∗−0.0068∗∗(-3.024)(-2.371)(-2.0229)(2.256)(3.058)lnMR2.0621∗∗0.1347∗∗0.0241∗∗∗0.1501∗1.0423∗∗(2.313)(2.008)(2.814)(2.511)(4.072)lnSIZE0.0961∗∗0.3219∗∗0.2251(2.552)(2.213)(0.552)lnBOD−0.2929∗∗∗−0.2153∗∗−0.1769∗∗ (2.335)(-4.535)(-2.122)lnOWN0.09620.3821∗∗∗0.2331∗∗(1.045)(3.047)(2.036)lnAGE0.2171∗∗0.4116∗0.0732(2.0361)(2.023)(1.8191)lnLEV0.26130.30110.3342∗∗∗(1.181)(1.483)(4.123)Constant0.0301∗∗0.1804∗∗0.2628∗∗0.2790∗0.0037(2.130)(1.994)(2.631)(4.340)(0.139)Observation391391391391391Adjusted R20.3620.4620.5630.4960.432No. of years1717171717Bank FENoNoYesYesYesYear FENoNoYesYesYesNotes: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1. Model 1 presents the baseline regression analysis examining the impact of financial risk dimensions—credit, liquidity, and market risk—on bank financial performance, proxied by Return on Equity (ROE), without the inclusion of control variables. Model 2 expands upon this by incorporating control variables, namely Board size, Board independence, Ownership structure, Bank age, and Board leverage, to account for potential confounding factors using random-effect model. In Models 3 through 5, both bank and year fixed effects are introduced to control for unobserved heterogeneity across institutions and over time. Specifically, Model 3 omits control variables, whereas Model 4 integrates them. To enhance the robustness of our findings, Model 5 substitutes the dependent variable with Return on Assets (ROA), maintaining the inclusion of both bank and year fixed effects. Prior to estimation, all variables were standardized to ensure the comparability and interpretability of the regression coefficients across different models. Table 6Effect of financial risk on bank performance.Table 6VariablesModel 1Model 2Model 3Model 4Model 5Main Dependent Variable is ROElnCR1.7763 ∗∗0.0142 ∗∗0.2259∗∗0.2773∗∗0.1370∗(2.751)(2.2263)(2.765)(2.152)(1.984)lnLR−0.0011∗∗−0.0248∗∗−0.1119∗∗−0.0241∗∗−0.0068∗∗(-3.024)(-2.371)(-2.0229)(2.256)(3.058)lnMR2.0621∗∗0.1347∗∗0.0241∗∗∗0.1501∗1.0423∗∗(2.313)(2.008)(2.814)(2.511)(4.072)lnSIZE0.0961∗∗0.3219∗∗0.2251(2.552)(2.213)(0.552)lnBOD−0.2929∗∗∗−0.2153∗∗−0.1769∗∗ (2.335)(-4.535)(-2.122)lnOWN0.09620.3821∗∗∗0.2331∗∗(1.045)(3.047)(2.036)lnAGE0.2171∗∗0.4116∗0.0732(2.0361)(2.023)(1.8191)lnLEV0.26130.30110.3342∗∗∗(1.181)(1.483)(4.123)Constant0.0301∗∗0.1804∗∗0.2628∗∗0.2790∗0.0037(2.130)(1.994)(2.631)(4.340)(0.139)Observation391391391391391Adjusted R20.3620.4620.5630.4960.432No. of years1717171717Bank FENoNoYesYesYesYear FENoNoYesYesYesNotes: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1. Model 1 presents the baseline regression analysis examining the impact of financial risk dimensions—credit, liquidity, and market risk—on bank financial performance, proxied by Return on Equity (ROE), without the inclusion of control variables. Model 2 expands upon this by incorporating control variables, namely Board size, Board independence, Ownership structure, Bank age, and Board leverage, to account for potential confounding factors using random-effect model. In Models 3 through 5, both bank and year fixed effects are introduced to control for unobserved heterogeneity across institutions and over time. Specifically, Model 3 omits control variables, whereas Model 4 integrates them. To enhance the robustness of our findings, Model 5 substitutes the dependent variable with Return on Assets (ROA), maintaining the inclusion of both bank and year fixed effects. Prior to estimation, all variables were standardized to ensure the comparability and interpretability of the regression coefficients across different models. Effect of financial risk on bank performance. Notes: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1. Model 1 presents the baseline regression analysis examining the impact of financial risk dimensions—credit, liquidity, and market risk—on bank financial performance, proxied by Return on Equity (ROE), without the inclusion of control variables. Model 2 expands upon this by incorporating control variables, namely Board size, Board independence, Ownership structure, Bank age, and Board leverage, to account for potential confounding factors using random-effect model. In Models 3 through 5, both bank and year fixed effects are introduced to control for unobserved heterogeneity across institutions and over time. Specifically, Model 3 omits control variables, whereas Model 4 integrates them. To enhance the robustness of our findings, Model 5 substitutes the dependent variable with Return on Assets (ROA), maintaining the inclusion of both bank and year fixed effects. Prior to estimation, all variables were standardized to ensure the comparability and interpretability of the regression coefficients across different models. The baseline model, devoid of control variables, revealed significant associations between financial risks and ROE. Subsequent models that integrated control variables provided a more comprehensive understanding of the factors influencing bank performance. Notably, the adjusted R-squared value was highest for Model 3, indicating that the inclusion of fixed effects contributed to a more precise estimation of the relationships. Credit Risk (lnCR) demonstrated a positive association with ROE, indicating that an increase in credit risk was correlated with higher shareholder returns. This suggests that banks taking on higher credit risk might achieve greater profitability, possibly due to the higher interest rates charged on riskier loans, which can enhance overall returns despite the increased risk of defaults. Conversely, Liquidity Risk (lnLR) was negatively related to ROE, implying that heightened liquidity risks could detract from a bank's profitability. The negative effect of liquidity risk on a bank's financial performance, as evidenced by its inverse relationship with ROE, can be attributed to several factors. Banks with heightened liquidity risk often need to maintain larger reserves of liquid assets, which generally yield lower returns compared to other investments. This conservative approach, while necessary for risk management, limits the bank's ability to invest in higher-yield opportunities, thereby reducing overall profitability. Additionally, managing liquidity risk can incur significant costs, including the opportunity cost of holding low-return assets and the direct expenses associated with liquidity management. These factors collectively detract from the bank's financial performance, explaining the negative impact of liquidity risk on ROE. The study further found that Market Risk (lnMR) exhibited a positive relationship with ROE, indicating that exposure to market risk could potentially enhance equity returns. This suggests that banks engaging in market activities or investments subject to market fluctuations may achieve higher profitability, possibly due to the higher returns associated with these riskier ventures. Among the control variables, Board Size (lnSIZE), Board Independence (lnBOD), and Ownership (lnOWN) significantly influenced ROE. Specifically, larger boards and greater board independence were linked to lower ROE, while higher ownership concentration was associated with improved performance. The study did not find significant relationships between the Age of the Firm (lnAGE), Leverage (lnLEV), and ROE. The incorporation of Bank fixed-effects and year fixed-effects in the model was pivotal for addressing unobserved heterogeneity across banks and over time. These fixed effects, represented by dummy variables, controlled for bank-specific characteristics that remain constant over time and year-specific influences, respectively. Their inclusion bolstered the robustness of the findings, ensuring that the estimated relationships were not confounded by omitted variables. Further analysis Table 7 presents the results of robustness checks and strategies employed to address endogeneity in the analysis of the impact of financial risks on bank performance in Ghana. The table displays estimates from six models, with Models 1–4 using a robust Generalized Method of Moments (GMM) approach and Models 5–6 utilizing the Three-Stage Least Squares (3SLS) method. The dependent variable alternated between Return on Equity (ROE) and Return on Assets (ROA) across models to provide a comprehensive assessment.Table 7Robust checks and addressing endogeneity.Table 7VariablesROEROAROEROAModel 1Model 2Model 3Model 4Model 5Model 6lnCR0.1350∗∗0.0342∗∗0.02110.0209∗∗0.0458∗∗0.0318∗∗(1.972)(3.621(2.142)(2.132)(4.867)(3.324)lnLR0.0221∗∗∗0.0162∗0.0315∗∗0.0048∗∗0.0227∗∗0.0436∗∗(2.103)(2.573)(2.883)(1.982)(2.762)(3.996)lnMR0.0453∗∗0.0124∗∗0.0563∗∗0.0138∗∗0.0231∗0.0674∗∗(3.764)(2.034)(4.041)(3.019)(2.201)(3.231)lnCR20.0341∗∗0.0461∗∗∗0.0427∗0.0075∗0.0571∗∗∗0.0548∗∗(4.112)(5.553)(3.341)(2.021)(3.782)(3.031)lnLR20.0239∗0.0353∗∗0.0326∗∗0.0153∗∗0.0464∗∗0.0434∗∗(2.541)(4.248)(2.691)(4.876)(4.654)(4.232)lnMR20.l1170.0231∗0.0207∗∗∗0.1486∗∗0.0339∗∗∗0.0316∗∗(1.384)(2.458)(2.124)(2.104)(3.932)(3.137)lnROE0.0462∗∗∗0.0018∗0.0568∗∗0.0338∗∗0.04590.0681∗∗(5.014)(1.687)(5.134)(3.062)(1.801)(5.782)lnROA0.0358∗∗0.00240.0438∗0.0382∗∗∗0.0351∗∗∗0.0542∗∗∗(4.501)(2.625)(4.678)(4.004)(3.441)(3.118)Constant0.1231∗∗0.13420.1452∗∗∗0.1428∗∗0.15610.1673∗∗(3.672)(5.822)(3.024)(3.014)(2.032)(2.945)ControlsNoYesNoYesYesYesInstrumentYesYesYesYesYesYesObservation391391391391391391No. of years171717171717AR (1)−2.781−3.214−3.456−3.678(0.004)(0.003)(0.005)(0.002)AR (2)1.456−1.789−2.034−2.123(0.110)(0.128)(0.091)(0.084)Hansen test38.8241.32440.11239.786(0.271)(0.289)(0.153)(0.301)Sargan test170.11174.32178.45180.67(0.192)(0.213)(0.207)(0.225)Wald's Chi2/R24125.67 (0.015)4210.454356.784421.34(0.019)(0.012)(0.014)Adjusted r20.4080.5280.3980.4570.6820.694Note: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1. The study adopted the GMM and 3SLS for robust checks and in addressing endogeneity issues. The main dependent variable for Model 1, 2 and 5 is ROE while model 3, 4 and 6 is ROA. Model 1–4 was estimated with robust GMM model while Models 5–6 were estimated with 3SLS model to address endogeneity issues that might arise. For addressing endogeneity, the original data for the independent variables were re-estimated to post 2010. The independent variables include lnCR2, lnLR2, lnMR2 = shorten period of data from (2010–2023). All of this help address endogeneity problems and improve the overall findings. All variables were standardized before estimation. Table 7Robust checks and addressing endogeneity.Table 7VariablesROEROAROEROAModel 1Model 2Model 3Model 4Model 5Model 6lnCR0.1350∗∗0.0342∗∗0.02110.0209∗∗0.0458∗∗0.0318∗∗(1.972)(3.621(2.142)(2.132)(4.867)(3.324)lnLR0.0221∗∗∗0.0162∗0.0315∗∗0.0048∗∗0.0227∗∗0.0436∗∗(2.103)(2.573)(2.883)(1.982)(2.762)(3.996)lnMR0.0453∗∗0.0124∗∗0.0563∗∗0.0138∗∗0.0231∗0.0674∗∗(3.764)(2.034)(4.041)(3.019)(2.201)(3.231)lnCR20.0341∗∗0.0461∗∗∗0.0427∗0.0075∗0.0571∗∗∗0.0548∗∗(4.112)(5.553)(3.341)(2.021)(3.782)(3.031)lnLR20.0239∗0.0353∗∗0.0326∗∗0.0153∗∗0.0464∗∗0.0434∗∗(2.541)(4.248)(2.691)(4.876)(4.654)(4.232)lnMR20.l1170.0231∗0.0207∗∗∗0.1486∗∗0.0339∗∗∗0.0316∗∗(1.384)(2.458)(2.124)(2.104)(3.932)(3.137)lnROE0.0462∗∗∗0.0018∗0.0568∗∗0.0338∗∗0.04590.0681∗∗(5.014)(1.687)(5.134)(3.062)(1.801)(5.782)lnROA0.0358∗∗0.00240.0438∗0.0382∗∗∗0.0351∗∗∗0.0542∗∗∗(4.501)(2.625)(4.678)(4.004)(3.441)(3.118)Constant0.1231∗∗0.13420.1452∗∗∗0.1428∗∗0.15610.1673∗∗(3.672)(5.822)(3.024)(3.014)(2.032)(2.945)ControlsNoYesNoYesYesYesInstrumentYesYesYesYesYesYesObservation391391391391391391No. of years171717171717AR (1)−2.781−3.214−3.456−3.678(0.004)(0.003)(0.005)(0.002)AR (2)1.456−1.789−2.034−2.123(0.110)(0.128)(0.091)(0.084)Hansen test38.8241.32440.11239.786(0.271)(0.289)(0.153)(0.301)Sargan test170.11174.32178.45180.67(0.192)(0.213)(0.207)(0.225)Wald's Chi2/R24125.67 (0.015)4210.454356.784421.34(0.019)(0.012)(0.014)Adjusted r20.4080.5280.3980.4570.6820.694Note: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1. The study adopted the GMM and 3SLS for robust checks and in addressing endogeneity issues. The main dependent variable for Model 1, 2 and 5 is ROE while model 3, 4 and 6 is ROA. Model 1–4 was estimated with robust GMM model while Models 5–6 were estimated with 3SLS model to address endogeneity issues that might arise. For addressing endogeneity, the original data for the independent variables were re-estimated to post 2010. The independent variables include lnCR2, lnLR2, lnMR2 = shorten period of data from (2010–2023). All of this help address endogeneity problems and improve the overall findings. All variables were standardized before estimation. Robust checks and addressing endogeneity. Note: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0.1. The study adopted the GMM and 3SLS for robust checks and in addressing endogeneity issues. The main dependent variable for Model 1, 2 and 5 is ROE while model 3, 4 and 6 is ROA. Model 1–4 was estimated with robust GMM model while Models 5–6 were estimated with 3SLS model to address endogeneity issues that might arise. For addressing endogeneity, the original data for the independent variables were re-estimated to post 2010. The independent variables include lnCR2, lnLR2, lnMR2 = shorten period of data from (2010–2023). All of this help address endogeneity problems and improve the overall findings. All variables were standardized before estimation. The independent variables of interest—Credit Risk (lnCR), Liquidity Risk (lnLR), and Market Risk (lnMR)—were consistently significant across models, with t-statistics indicating their reliability. Specifically, lnCR and lnMR were positively associated with both ROE and ROA, suggesting that these risks contribute to bank performance. Liquidity Risk (lnLR) also showed a positive relationship, but with varying significance levels. To address potential endogeneity, the study incorporated squared terms of the independent variables (lnCR2, lnLR2, lnMR2) and utilized instrumental variables. The inclusion of these variables and the use of post-2010 data for re-estimation helped to mitigate endogeneity concerns. The Hansen test and Sargan test results, which are statistical tools for testing the validity of instruments, provided support for the chosen instruments, as indicated by p-values greater than the standard threshold of 0.05. Autoregressive (AR) tests were conducted to check for serial correlation in the residuals, with the results suggesting the absence of second-order serial correlation in most models. The Wald's Chi-square statistic and R2 values indicated the overall goodness of fit for the models, with Adjusted R2 values showing the proportion of variance explained by the models, after adjusting for the number of predictors. The constant term was statistically significant in all models, providing an indication of the average effect when all predictors are zero. The inclusion of control variables in Models 2, 4, 5, and 6 did not alter the significance of the key independent variables, suggesting that the findings are robust to the inclusion of additional covariates. Heterogeneity test Table 8 presents the results of a heterogeneity mechanisms test, which investigates how various bank characteristics—size, ownership structure, and age—affect the relationship between financial risk and bank performance, as measured by Return on Equity (ROE). The analysis was conducted in three panels, each focusing on different subgroups within the banking sector.Table 8Heterogeneity Mechanisms test.Table 8VariablesThe outcome variable is ROEPane APanel BPanel CBank Size (BS)Ownership StructureAgeLarge Size BanksSmall Size BanksForeign-owned BanksDomestic-owned BanksAged -BanksYounger-BankslnCR0.0321∗∗0.01520.0418∗∗∗0.0189∗∗0.0507∗∗∗0.0262∗∗∗(2.221)(1.489)(3.678)(2.214)(4.298)(2.543)lnLR0.0223∗∗0.0171∗∗0.0243∗∗0.0274∗∗0.0328∗∗0.0198∗∗(2.374)(2.171)(2.758)(1.943)(3.114)(2.084)lnMR0.0415∗∗∗0.01280.0521∗∗∗0.0237∗∗0.0639∗∗∗0.0345∗∗∗(3.721)(1.978)(4.234)(2.531)(3.241)(3.129)Constant0.1423∗∗∗0.1342∗∗∗0.1568∗∗0.1340∗∗∗0.1783∗∗∗0.1456(3.052)(2.731)(4.823)(2.124)(2.745)(1.089)ControlsYesYesYesYesYesYesObs.391391391391391391Years171717171717AR (1)−2.134−2.082−2.274−2.098−2.281−2.153(0.025)(0.037)(0.041)(0.039)(0.222)(0.032)AR (2)−1.321−1.297−1.412−1.384−1.472−1.364(0.115)(0.128)(0.139)(0.128)(0.114)(0.123)R20.4590.5040.4420.4170.5220.431Coefficient test between sectorsChi-square43.32146.21448.27P-value0.0220.0150.012Note: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0. 1. Panel A explores the heterogeneity in the impact of bank size, distinguishing between large and small institutions. Panel B examines ownership structure effects, contrasting domestic and foreign-owned banks among the 23 surveyed institutions. Panel C investigates the influence of a bank's age on the relationship between financial risk and its financial performance. The original data were standardized through logarithmic transformation. Table 8Heterogeneity Mechanisms test.Table 8VariablesThe outcome variable is ROEPane APanel BPanel CBank Size (BS)Ownership StructureAgeLarge Size BanksSmall Size BanksForeign-owned BanksDomestic-owned BanksAged -BanksYounger-BankslnCR0.0321∗∗0.01520.0418∗∗∗0.0189∗∗0.0507∗∗∗0.0262∗∗∗(2.221)(1.489)(3.678)(2.214)(4.298)(2.543)lnLR0.0223∗∗0.0171∗∗0.0243∗∗0.0274∗∗0.0328∗∗0.0198∗∗(2.374)(2.171)(2.758)(1.943)(3.114)(2.084)lnMR0.0415∗∗∗0.01280.0521∗∗∗0.0237∗∗0.0639∗∗∗0.0345∗∗∗(3.721)(1.978)(4.234)(2.531)(3.241)(3.129)Constant0.1423∗∗∗0.1342∗∗∗0.1568∗∗0.1340∗∗∗0.1783∗∗∗0.1456(3.052)(2.731)(4.823)(2.124)(2.745)(1.089)ControlsYesYesYesYesYesYesObs.391391391391391391Years171717171717AR (1)−2.134−2.082−2.274−2.098−2.281−2.153(0.025)(0.037)(0.041)(0.039)(0.222)(0.032)AR (2)−1.321−1.297−1.412−1.384−1.472−1.364(0.115)(0.128)(0.139)(0.128)(0.114)(0.123)R20.4590.5040.4420.4170.5220.431Coefficient test between sectorsChi-square43.32146.21448.27P-value0.0220.0150.012Note: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0. 1. Panel A explores the heterogeneity in the impact of bank size, distinguishing between large and small institutions. Panel B examines ownership structure effects, contrasting domestic and foreign-owned banks among the 23 surveyed institutions. Panel C investigates the influence of a bank's age on the relationship between financial risk and its financial performance. The original data were standardized through logarithmic transformation. Heterogeneity Mechanisms test. Note: t-statistics in parentheses ∗∗∗p < 0.01, ∗∗p < 0.05, ∗p < 0. 1. Panel A explores the heterogeneity in the impact of bank size, distinguishing between large and small institutions. Panel B examines ownership structure effects, contrasting domestic and foreign-owned banks among the 23 surveyed institutions. Panel C investigates the influence of a bank's age on the relationship between financial risk and its financial performance. The original data were standardized through logarithmic transformation. In this study, we classified the 23 Ghanaian banks based on size, ownership structure, and age. The size classification involved categorizing banks into large and small banks using specific metrics: banks with total assets exceeding GHS 1 billion and annual revenue above GHS 500 million were classified as large, while those below these thresholds were categorized as small. For ownership structure, banks were classified as either foreign-owned or domestic-owned, with foreign-owned banks having a majority stake held by foreign entities and domestic-owned banks being primarily owned by local investors. Additionally, banks were categorized by age into aged and younger banks, with aged banks being those established before the year 2000 and younger banks being those established in 2000 or later. These classifications enable a comprehensive analysis of the banking sector, providing a deeper understanding of the differences in performance, risk profiles, and strategic behaviors across various categories of banks in Ghana. Panel A reveals significant differences in the impact of financial risks based on bank size. Large size banks exhibited a positive and significant association between Credit Risk (lnCR) and ROE (β = 0.0321, p < 0.05), whereas small size banks showed an even stronger association (β = 0.0418, p < 0.01). Liquidity Risk (lnLR) and Market Risk (lnMR) also demonstrated positive relationships with ROE across both size categories, with small banks showing a slightly higher impact for Market Risk. Ownership structure significantly influenced the relationship between financial risks and ROE. Foreign-owned banks displayed a stronger positive relationship between Credit Risk (lnCR) and ROE (β = 0.0507, p < 0.01) compared to domestic-owned banks (β = 0.0189, p < 0.05). Liquidity Risk (lnLR) and Market Risk (lnMR) were similarly significant for both ownership types, with foreign-owned banks showing a more pronounced impact from Market Risk. The age of banks also played a role in how financial risks affect ROE. Older banks had a significant positive relationship between Market Risk (lnMR) and ROE (β = 0.0639, p < 0.01), while younger banks showed a slightly lower but still significant impact (β = 0.0345, p < 0.05). Credit Risk (lnCR) was significant for both age groups, with younger banks experiencing a higher impact. All models included control variables and were estimated using robust standard errors, with t-statistics reported in parentheses. The Autoregressive (AR) tests indicated no significant serial correlation in the residuals for any of the panels. The R2 values for each panel ranged from 0.417 to 0.522, indicating a moderate to strong fit of the models. Post-hoc tests for differences in coefficients across sectors were conducted, with the Chi-square statistic and associated p-values suggesting significant differences in the impact of financial risks on ROE across bank size, ownership structure, and age categories. Discussions Our study's findings on the impact of financial risks on the performance of commercial banks in Ghana contribute to a growing body of literature that seeks to understand the complex dynamics within the banking sector. The significant influence of credit risk on bank performance aligns with established theories suggesting that an effective management of credit risk can lead to better financial outcomes [[21], [22], [23], [24]]. However, our results provide a unique perspective, indicating that banks may be actively using credit risk as a strategic tool for performance enhancement rather than merely mitigating its adverse effects. The negative impact of liquidity risk on bank performance reported in our study corroborates previous research that has consistently highlighted the perils of liquidity constraints [20,52,80,81]. These constraints can impede a bank's ability to meet short-term obligations, leading to a decrease in profitability and potentially affecting the stability of the financial system. Our findings underscore the critical need for liquidity risk management and the importance of maintaining a balanced sheet. The positive association between market risk and bank performance in our study offers a novel contribution to the literature. While conventional wisdom often views market risk as detrimental, our results suggest that Ghanaian banks have been able to capitalize on market volatility, possibly through strategic investments and effective risk management practices. This finding is in line with studies from neighboring African countries that have reported similar benefits from exposure to market risks [2,53,82], indicating a regional trend of proactive risk management in the financial sector. Our heterogeneity mechanisms test adds another layer to the discourse by revealing that the impact of financial risks varies significantly across different bank characteristics. This finding is consistent with prior research that has emphasized the importance of contextual factors in shaping the relationship between risk and performance [21,52]. It suggests that risk management strategies must be tailored to the specific needs and circumstances of individual banks. The study's results, when juxtaposed with prior findings, indicate an evolution in the understanding of financial risk management. While earlier studies often focused on the mitigative aspects of risk management, our findings, along with recent literature, point towards a more strategic and proactive approach to risk management in the banking sector. In conclusion, this study advances the conversation on financial risk and bank performance by providing new insights from the Ghanaian banking sector. The findings not only contribute to the theoretical understanding of the subject but also offer practical implications for risk management in the banking industry. As the financial landscape continues to evolve, our study emphasizes the need for continuous exploration and adaptation in risk management practices. Conclusion, limitations, and suggestions for further studies Conclusion This study, grounded in the finance distress theory, the shiftability theory of liquidity, and the extreme value theory (EVT), provides a profound understanding of how financial risks, namely credit, market, and liquidity risks, influence the performance of Ghanaian commercial banks. The analysis reveals a positive correlation between credit risk and equity returns, suggesting that banks can enhance their performance by strategically managing credit risk. However, the study also uncovers a negative relationship between liquidity risk and return on equity, indicating that banks must maintain adequate working capital to meet their obligations and ensure financial stability. Furthermore, the positive association between market risk factors and bank earnings highlights the potential for banks to capitalize on market volatility when prudently managed. Implications Theoretical implications The study's findings advance the theoretical understanding of financial risk management by integrating finance distress theory, shiftability theory of liquidity, and extreme value theory (EVT). These theories provide a comprehensive framework for analyzing the multifaceted impact of financial risks on bank performance. The positive correlation between credit risk and equity returns challenges traditional risk aversion models, suggesting that an optimal level of credit risk can be beneficial for bank performance. The negative association between liquidity risk and ROE supports the critical role of liquidity management in maintaining bank solvency and operational efficiency. The study also contributes to the EVT literature by demonstrating its applicability in assessing market risk and its impact on bank performance, thereby enriching the theoretical discourse on risk-return dynamics in the banking sector. Practical implications The practical implications of this research are far-reaching, particularly for bank managers and financial regulators. For bank managers, the study emphasizes the strategic importance of credit risk management. By developing clear credit policies and lending standards, banks can enhance their asset quality and reduce non-performing loans, leading to improved profitability. The findings also highlight the need for banks to maintain a balance between liquidity and solvency, suggesting that excessive liquidity may not be necessary if banks can effectively manage their assets' shiftability. Furthermore, the positive impact of market risk on bank performance suggests that banks can benefit from market volatility if they have robust risk management practices in place. For financial regulators, the study provides insights into the effectiveness of different risk management strategies and the importance of a balanced approach to risk regulation. Regulators can use these findings to develop policies that encourage banks to manage risks effectively while also allowing them the flexibility to capitalize on market opportunities. Managerial implications From a managerial perspective, the study offers several key insights. First, it underscores the importance of a proactive approach to credit risk management, where banks should not only focus on minimizing defaults but also on leveraging credit risk to enhance returns. Second, it highlights the need for banks to continuously monitor and adjust their liquidity positions to ensure they can meet short-term obligations without compromising long-term growth. Third, it suggests that banks should adopt a dynamic market risk management strategy that allows them to respond quickly to changes in market conditions. Finally, the study implies that banks should invest in the development of robust risk management systems and capabilities, including the use of advanced analytical tools and techniques such as EVT. The study recommends that Ghanaian banks improve their credit analysis and loan management capacities, monitor loan-to-deposit ratios judiciously, and adopt a robust risk management framework overseen by senior management. It also advises banks to assess the risk tolerance of their stakeholders and to implement a sound approach to identifying, controlling, and tracking their exposure to liquidity risk. Policy implication The policy implications of this research are significant, particularly in the context of financial regulation and supervision. The findings suggest that regulators should consider a balanced approach to risk regulation that recognizes the potential benefits of credit risk while also mitigating the adverse effects of liquidity and market risks. This may involve developing guidelines and standards that encourage banks to implement effective risk management practices, such as stress testing, scenario analysis, and risk-based capital planning. Additionally, the study's findings on the positive impact of market risk highlight the need for regulators to promote transparency and market discipline, for example, through the implementation of effective disclosure requirements and market surveillance mechanisms. The study's implications for theory, practice, and policy are profound, offering a nuanced understanding of the role of financial risks in shaping bank performance and providing actionable insights for bank managers, financial regulators, and policymakers. Limitation and suggestions for further studies This study presents several limitations that may affect the generalizability and reliability of its findings. One significant limitation is the reliance on secondary data sources, which may limit the depth of analysis. Data availability issues restricted the study to financial statements from the Ghana Commercial Banks Supervision Department between 2006 and 2023. The use of unbalanced panel data also posed challenges, potentially impacting the robustness of the results. Additionally, various biases, such as selection, response, and researcher bias, could influence the findings, as the managers participating in the training program may not represent the entire population of bank managers in Ghana. The study also faces limitations in establishing clear causality between financial risk and profitability, as external factors like economic changes and industry trends could also play significant roles. Lastly, the limited timeframe might restrict the ability to capture long-term effects of financial risk on profitability. To enhance the understanding of financial risk and its impact on Ghana's financial institutions, future research should address these limitations. Specifically, future studies should investigate the impact of different types of financial risks, such as credit and operational risks, on profitability. Additionally, examining how external factors, including interest rate fluctuations and global economic conditions, influence financial risk and profitability would provide a more comprehensive understanding. Research should also explore the effectiveness of various risk management strategies in mitigating financial risk and enhancing profitability. Furthermore, investigating the role of financial regulations on the profitability of financial institutions in Ghana is essential. By addressing these areas, researchers can provide a deeper understanding of financial risk management and offer valuable insights for financial institutions in Ghana. While the current study's limitations do not diminish its significance, they highlight areas for future research to enhance findings and address any gaps. CRediT authorship contribution statement Benjamin Blandful Cobbinah: Writing – review & editing, Writing – original draft, Investigation, Conceptualization. Wen Yang: Supervision, Funding acquisition. Francis Atta Sarpong: Visualization, Validation, Software, Formal analysis. George Nyantakyi: Resources, Investigation, Data curation. Data and code availability Data will be made available on request. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.