The influence of solvency and liquidity ratios on profitability of Tunisian banks: the moderating effect of asset quality

Abstract

Purpose

The objective of this research is to determine the influence of solvency and liquidity on the profitability [return on assets (ROA)] of Tunisian banks from Q2-2020 to Q3-2022 by considering asset quality as a moderating variable.

Design/methodology/approach

This study uses data on liquidity, solvency, ROA and asset quality for 12 banks. It also considers bank size, gross domestic product (GDP) growth and inflation as control variables. The methodology is based on panel data with generalized least squares (GLS) estimation to assess the moderate influence of the asset quality on solvency, liquidity and ROA. Also, the generalized method of moments (GMM) estimation is used as a robustness test.

Findings

The results of the GLS model estimation indicated a negatively significant moderating correlation between the liquidity and the solvency. The data from the GMM model indicate that the liquidity variable predicted by the liquidity has a positively significant influence on a bank's ROA as well as for the solvency variable, which is predicted by the capital capacity. Therefore, we conclude that these two variables had a positively significant impact on the ROA.

Research limitations/implications

The studies have many implications for banks and their management in addition to the industry regulators. The results of this study will enable political decision-makers to determine the banks' profits based on their liquidity and solvency.

Originality/value

This analysis provides financial explanations and recommendations for stakeholders in Tunisian banks. Furthermore, these banks must also be able to maintain their liquidity and solvency to ensure their profits in times of COVID-19.

Citation

Ben Abdallah, M. and Bahloul, S. (2024), "The influence of solvency and liquidity ratios on profitability of Tunisian banks: the moderating effect of asset quality", African Journal of Economic and Management Studies, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/AJEMS-02-2024-0136

1. Introduction

Progress in the international economy demands good business administration. Banks must perform and develop in all domains in order to keep up competitively. In the case of Tunisia, business sectors such as service, trading and manufacturing companies are characterized by very tight competition. After 2010, the situation of the Tunisian economy was less favorable. This was due to the consequences of the 2008 global crisis. Indeed, a number of firms have run into financial trouble because they haven't paid their financial debt.

This involves risk, as highlighted by Kyule (2015) and Rafique et al. (2020), who state that risk is a constitutive element relating to investment. The financial risks faced by a business include liquidity, solvency and profitability risks, but businesses that adopt good business planning and performance management are more likely to achieve their strategic and operational objectives (Savchenko and Derbeneva, 2020). Consequently, some researchers such as Muthoni (2015) evaluated the negative influence of liquidity and solvency risk on performance [return on assets (ROA) and return on equity (ROE)]. In this context, Kepramareni et al. (2022) explored this effect for 23 Perkreditan Rakyat banks on Denpasar city from 2018 to 2020. They show that banks with a high level of capital do not significantly influence the bank's financial performance. Banks' lack of optimal use of their capital can explain the absence of this influence. In addition, the higher a bank's liquidity, the greater the credit it can extend. A high or low loan-to-deposit did not affect financial performance. This could be due to credit being disbursed at less than the maximum, while funds collected are high, which can lead to losses.

Liquidity, solvency, asset quality and ROA are the four fundamental indices of financial performance. Firstly, bank liquidity refers to the bank's ability to meet its short-term obligations within a given timeframe and reveals the bank's willingness to cover its current debts, management's efficiency in managing working capital and developments in the field of finance. Secondly, bank solvency indicates the bank's capacity to cover its long-term debts within a given timeframe. Shareholders, bondholders and long-term creditors, such as financial institutions, are all concerned by the bank's solvency. Therefore, solvency is also studied to determine the bank's capital structure. Thirdly, asset quality is an important factor in the financial reliability and health of the banking sector. The main determinant of this factor is the value of the loan portfolio and the banks' control of credit management. The actual credit risk expresses asset quality ratings, which are related to the investment portfolios of loans. In our study, we are only concerned with ROA, which indicates the efficiency of the bank's operations in converting assets into income.

Various analyses, such as those by Abdilahi and Davis (2022), Samara et al. (2022) and Kumalo (2023), have been carried out to research the impact of liquidity, solvency and asset quality on bank performance in the African context. However, to our knowledge, there is a lack of research on this impact in the Tunisian context. There is thus a lacuna in the existing literature that this research aims to address.

Our research objective is to analyze the impact of liquidity and solvency on the profitability of 12 Tunisian banks from 2020 to 2022, taking into account the moderating effect of asset quality.

The outcomes of this paper will provide a variety of contributions to the theoretical and practical aspects of finance. They will help educate policymakers, particularly in the Tunisian Central Bank and the Treasury, regarding how liquidity and solvency influence bank profitability. In addition, financial advisors will be able to employ the research to guide their customers on financial ratios.

The research is structured around six sections. Section 1 introduces the study. Section 2 xoutlines the literature review and hypothesis formulation. Section 3 shows the research methodology. Section 4 details the data analysis results. Section 5 deals with the robustness tests. Finally, Section 6 is the conclusion.

2. Literature review and hypothesis formulation

2.1 Theoretical analysis

2.1.1 Baumol model theory

According to Muthoni (2015), Baumol's liquidity management model allows companies to determine the maximum level of liquidity to hold in situations of confidence (Baumol, 1952). It is founded on compensating the liquidity created by having cash with the loss of interest caused by maintaining the asset value in the form of non-interest-earning current accounts. This approach is used to define the cash flow target. This model involves payments and assumes that the firm holds cash until the end of a given period. It also supposes that the interest cost of liquidity is determined and held in the form of short-period investments. Once the inputs and outputs have been established, the company is in a position to define the average cash balance.

2.1.2 The Miller–Orr model

Miller and Orr (1966) developed this theory, and it is applied to determine a company's target cash balance. This model means that treasury flows are unreliable, and it corrects the shortcomings of the previous model, which does not enable treasury flows to change (Muthoni, 2015). To compensate for these shortcomings, this model takes into account changes in cash flow on a daily basis.

2.1.3 Liquidity preference theory

In macroeconomic theory, this theory concerns the need for monetary policy, which is a liquidity policy (Kipngetich, 2019). Keynes (1936) developed the concept of liquidity to attribute the determination of the interest rate to the offer and the demand and listed three criteria justifying the demand for and preference for liquidity: the transaction criterion, which companies and individuals use to conduct their daily operations; the cautionary criterion, which implies keeping liquidity to respond to unexpected emergencies, and the speculation criterion, which gives a company the ability to take advantage of special opportunities that will be favorable.

2.1.4 Shiftability theory

In accordance with Kyule (2015), Maina (2017), Muthike (2017), Kipngetich (2019) and Muriuki (2022), Moulton (1918) developed this theory in the United States of America. It is based on guaranteed liquidity and assets that can be shifted or sold to other investors or creditors to receive cash without loss. This theory therefore asserts that the transferability, negotiability or assign ability of a bank's assets is means of guaranteeing liquidity. This theory contains elements to which banks can take on reliable assets that can be placed with other banks. For example, shareholders, bonds, cash bills and exchange invoices are considered to be liquid.

2.2 Previous empirical research

A great deal of work has been done on the influence of financial ratios on the performance of banks. Salih et al. (2019) conducted empirical research examining data on financial ratios underpinning the performance of 46 conventional and ethical banks in the Gulf Cooperation Council (GCC) from 2006 to 2012. Using cross-sectional time series data, the ROA firstly provides a significant estimate of how an Islamic bank's management is more efficient than that of a conventional one in using assets to create profits. Secondly, the loan-to-deposit ratio is a very useful variable for assessing the bank's ability to meet its financial obligations. This ratio is more significant in conventional banks than in Islamic banks. Thirdly, the solvency is an essential parameter as it assesses the bank's willingness to meet its debts and financial obligations. This ratio is more significant in conventional banks than in Islamic banks.

Zaidan et al. (2022) showed that the measurements of financial ratios are essential to guarantee the performance of Jordanian banks. Furthermore, the model suggested in this study is the theory of 3P showed the link between these ratios, using net income as a moderating variable. The results validated the hypotheses put forward, namely that bank profitability is considerably affected by liquidity and solvency.

In South African context, Kumalo (2023) examined the effects of liquidity and solvency on 13 banks’ performance over the period 2012–2022. Using ordinary least squares (OLS) and generalized method of moments (GMM) estimations, results showed that liquidity and solvency have a positive and strong relationship with profitability and when liquidity increases profitability also increases. This finding is coherent with Marbun and Mulan (2020).

Asset quality is based on the quality of loan portfolios. It is fundamentally established through investment in assets. The influence of asset quality on the performance of banks has been the subject of numerous studies. Samara et al. (2022) analyzed and tested bank compliance, asset quality and liquidity to profitability of 43 banks in Indonesia from 2018 to 2020. Quantitative causative research results show the effect of bank compliance, asset quality and liquidity on profitability is simultaneously able to justify the variable of profitability.

In three major sub-Saharan African countries, namely Kenya, Nigeria and South Africa, Abdilahi and Davis (2022) sought to identify the determinants of bank profitability in 240 banks during 1990–2019. Applying a random effects model, their results reveal that liquidity and capital adequacy ratio (CAR) have a significantly positive effect on bank ROA, but asset quality has a significant negative effect on ROA.

2.2.1 Influence of liquidity on profitability in banks

Thi Doan and Buia (2021) examined the role of liquidity on the ROA of 26 banks from 2013 to 2018. GMM estimation is used to assess the significantly positive effect of liquidity on the ROA of Vietnamese banks. This is incoherent with Khan et al. (2023).

Islam et al. (2022) examined how liquidity favorably or unfavorably influences the ROA of 5 banks in Bangladesh over 10 years (2011–2020). The OLS model was used to investigate the relationship between these variables. This analysis reveals that an increase in liquidity would result in an improvement in ROA. This is in line with Paul et al. (2020).

Pandey and Budhthoki (2020) seek to analyze the impact of liquidity on the ROA of 27 Nepalese banks from 2008–2009 to 2017–2018. Using regression models, the results of this study indicate a negative relationship between liquidity and ROA, meaning that the higher the liquidity, the lower the ROA. This is in line with Pradhan and Shrestha (2018) and Sah and Pradhan (2023).

In addition, during the COVID-19 in Indonesia, Kamal (2023) aimed at verifying the influence on ROA. Using GLS and GMM system, the results reveal that the rapport between these variables is negative. This investigation complies with Poniman's study (2022).

However, in the Tunisian context, Ben Moussa et al. (2022) analyzed the relationship between liquidity and ROA of 18 banks in this context from 2000 to 2017. Based on the statics of the panel, they found that the liquidity has a significant and positive impact on ROA. This is because a good liquidity minimizes the risk of default, which can lower the cost of funding and therefore increase ROA. Similarly, for the African context, Waweru and William (2023) examined the impact of liquidity on the financial performance of 39 banks in Kenya from 2017 to 2021. The regression analysis showed that liquidity has a positive and significant effect on financial performance. These results are expected to shed further light on the critical role played by liquidity requirements in the financial stability and resilience of the banking sector in Kenya.

Indeed, some empirical studies seem to show a favorable influence, while other studies have shown that this influence is unfavorable. This suggests that previous studies on this influence are still very incomplete as they have not provided concrete results on this relationship. Indeed, we will suggest the following hypothesis:

H1.

Liquidity has a positive impact on profitability for banks in Tunisia.

2.2.2 Influence of solvency on profitability in banks

In accordance with Pandey and Budhthoki (2020), solvency has a positively influence on ROA, meaning that the higher the solvency, the greater the ROA. In addition, Qadri et al. (2023) analyzed the relationship between COVID-19 and the performance of the 34 banks in South Asian region between 2016 and 2021. A Wilcox rank test was performed to establish this relationship. By applying solvency as a measure of overall performance, the bank was found to perform better in this period.

Likewise, Saiz-Sepúlveda et al. (2024) examined the solvency and ROA of the Spanish banking sector from 2011 to 2021. Based on a descriptive and explanatory methodology, the conclusions indicate a significant link between solvency and ROA. However, Tazriah and Pratiwi (2023) attempted to measure the financial performance of PT Bank Mandiri over a 10-year sample, more precisely from 2012 to 2021. Using the one-sample t-test method, the results show a negatively significant relationship between solvency and ROA. Therefore, in coherence with the banking standard defined by Cashmere in 2016, a ratio below 100% is classified as favorable, while a ratio above 100% is classified as unfavorable. This research is consistent with Khan et al. (2021).

In the African context, more precisely in Morocco and during the period from 2011 to 2020, Aayale et al. (2022) showed the link between indicators of solvency, liquidity, asset quality and profitability. Based on regression, they found that these indicators have an unfavorable correlation with the ROA of these banks.

In the case of Tunisia, Ben Moussa et al. (2022) assessed the impact of CAR on the profitability of a sample of 11 Tunisian banks from 2000 to 2018. According to the panel data, the results show that this ratio has a significant positive effect on ROA. This means that a high volume of assets reduces solvency, which in turn increases profitability.

Indeed, several studies seem to show that this influence is positive, whereas other studies have shown that this influence is negative. This suggests that previous studies on this influence are still very incomplete because they have not provided concrete results on this relationship. Indeed, we will suggest the following hypothesis:

H2.

Solvency has a positive impact on profitability for banks in Tunisia.

2.2.3 Influence of asset quality on profitability in banks

Widyatmoko and Risman (2024) seek to examine the impact of digital finance, CAR, efficiency and asset quality on seven Indonesian bank's ROA from 2015 to 2022. Using the fixed effects model, the findings of test revealed that the asset quality variable has no effect on ROA. Additionally, the moderator variable test suggests that also this ratio has no significant but positive influence on ROA. This research is incompatible with Chakraborty (2023).

However, in the African banking sector, Gathara et al. (2023) examined the influence of asset quality on the value of 43 banks in Kenya between 2014 and 2021. Using regression technique, the researchers found that asset quality has a positive influence on the value of banks in Kenya. This indicates that banks with higher asset quality are considered more attractive by investors in African markets.

Likewise, Linda et al. (2023) determined the significantly positive impact of asset quality on the 12 bank's ROA in Kenya for the financial years 2017–2021. Using the descriptive methods and inferential statistics, the analysis suggests that banks should consider investing in their assets because it improves profitability.

Similarly, Ikpesu and Oke (2022) analyzed the relationship between CAR and asset quality on the performance of 12 banks in Nigeria from 2010 to 2019. Using the GMM estimation, the findings revealed that CAR and asset quality positively influences performance. This means that CAR and asset quality reinforce and promote performance.

However, Jean (2022) assessed this effect of Congo banks from 2010 to 2020. After the Statistical Package for the Social Sciences (SPSS) analysis, this finding indicates that asset quality has a negative influence on ROA, which declines in monetary units for each unit increase in non-performing loans. This result is also due to the primary reason for the vulnerability of the banking sector.

In the case of Tunisia, to our knowledge, there are unfortunately no empirical or theoretical studies to date on the impact of quality of the loan portfolio on profitability and its moderating role. For this reason, we will consider this impact as a contribution compared to previous studies. Indeed, we will suggest the following hypothesis:

H3.

Asset quality has a positive impact on profitability for banks in Tunisia.

3. Research methodology

The sample of this study uses the financial ratios of the Tunisian banks named [Tunisian Banking Company (STB), Housing Bank (BH), National Agricultural Bank (BNA), Amen Bank, Banking union for commerce and industry (UBCI), International banking union (UIB), Bank of Tunisia (TB), Arab International Bank of Tunisia (BIAT), Attijari Bank, Arab Tunisian Bank (ATB), Wifak Bank and Bank of Tunisia and the Emirates (BTE)] from financial reports. For this investigation, we collected 12 conventional Tunisian banks covering the period from Q2-2020 to Q3-2022. To analyze our investigation, we adopt the equation as follows:

(1) ROAit=α0+α1Solvencyit+α2Liquidityit +α3Asset Quality+Asset Quality×(α4Solvencyit+α5Liquidityit )+α6SIZEit+α7GDPit+α8INFLit+εit

with:

• (1)

  ROAit: indicates bank's profitability;

• (2)

  Solvencyit and  Liquidityit : indicate financial ratios;

• (3)

  Asset quality_it: indicates moderator financial ratios and

• (4)

  SIZEit, GDPit, and INFLit: indicates parameters variables.

3.1 Dependent variable

Many studies such as Aslam et al. (2022), Rohman et al. (2022) and Kamal (2023) used ROA, ROE and net interest margin (NIM) as performance measures. In our study, we are only interested in ROA as a dependent variable, because it is considered an indication of profitability through optimal use of resources. It is obtained by relating net profit to total assets (Yuanita, 2019; Monsia et al., 2021; Safarda et al., 2023).

3.2 Independent variables

According to David and Osemwegie (2016) and Safarda et al. (2023), solvency is defined as a bank's capacity to pay its long-term obligations. The higher is this ratio, the less solvent is the bank. The second variable is the liquidity. It defines the bank's willingness to pay its short-term debts and obligations on a specific date. The loan-to-deposit ratio is an important measure of financial liquidity, enabling banks to determine their financial health. If a bank is solvent in the short term, the liquidity must be greater than 1. The moderating variable asset quality was measured by the quality of the loan portfolio. It demonstrates the extent to which a bank manages its financial decisions (particularly with regard to its loan portfolio) in line with its debts (Kalanidis, 2016).

3.3 Control variables

Size of bank, which was calculated as total assets using a natural logarithm (Charmler et al., 2018; Kamal, 2023) and macroeconomic variables (GDP and INFL) (Titko et al., 2015; Nielsen and Raswant, 2018; Phan et al., 2020; Wahyudi, 2020; Soesetio et al., 2022; Kamal, 2023). GDP has been employed in most empirical work to evaluate the effectiveness of the country's economic situation on banking activity and performance. Inflation (INFL) is the general rise in prices and its decline in the purchasing power of economic players. Appendix, Table A1 below summarizes all the variables.

4. Data analysis results

4.1 Descriptive statistics

Table 1 presents the outcomes of the descriptive statistical analysis, which was showed that the average of return on assets during this period is 0.0359, with Min of 0.0002 and Max of 0.1087. This means that these 12 banks were affected by a slowdown in ROA growth.

The mean for solvency is 0.0040, liquidity is 0.2014 and asset quality is 0.0450. For the control variables, the average of bank size is 7.3489, the mean of GDP is 4.0339 and INFL is 6.4690. As far as the minimum and maximum levels of these parameters are concerned, solvency is 0.0001 and 0.1020, liquidity is 0.0041 and 0.6093 and asset quality is 0.0035 and 0.4470. Meanwhile, SIZE is 6.7897 and 7.2941, GDP is 3.9829 and 4.1025 and INFL is 5.64 and 8.31%.

Like Ben Abdallah and Bahloul (2022) and Ben Abdallah and Bahloul (2024), we use correlation to check multicollinearity. From Table 2, we found that there is no multicollinearity problem.

Solvency and liquidity are positively and significantly correlated with ROA. This suggests that an improvement in these ratios may lead to a higher ROA. The positive relationship between liquidity and ROA is justified by the amount of loans issued to clients' increases in compliance with the profits obtained (Djebali and Zaghdoudi, 2020). While the positive correlation between solvency and ROA is explained by the fact that the banks' management attitude complies with the Tunisian Central Bank's regulations. When this ratio is good, the bank performs well and protects itself against both credit and other risks. These results are inconsistent with previous research by Tabash and Hassan (2017). However, these findings are compatible with Agustin (2018), Ardichy (2022), Hanifa et al. (2022), Afrilien (2023), Hakim (2023) and Murtiningrum and Wahyuningsih (2024).

Table 2 shows a positive association between ROA and asset quality with a coefficient of 0.130. This finding is coherent with Shittu and Abdulkadir (2023). The reduction in provisions for asset quality, measured by the quality of the loan portfolio, enhances banks' profitability.

4.2 Panel data tests

4.2.1 Hausman, heteroscedasticity and auto-correlation tests

Firstly, before concretizing Hausman's test, we carried out some specific tests. Secondly, to control heteroscedasticity, we used the Breusch–Pagan test. Finally, we perform Wooldridge test to confirm the auto-correlation test. These results are shown in Table 3.

Table 3 found that Hausman's p-value >1%. Therefore, the results for random effects panel data are retained. For the second test, a p-value of less than 5% indicates a random-effects heteroscedasticity problem. The p-value of the third test is less than 5%, so there's an auto-correlation problem.

Therefore, we need to use an estimate for the random effect in order to obtain standard errors for both models. Consequently, the availability of tests 2 and 3 took the GLS regression model into consideration.

4.2.2 Empirical results

Table 4 shows that the solvency and liquidity have an impact positively significant on the Tunisian banks ROA. Moreover, our finding is consistent with hypotheses H1 and H2. As the source of the solvency has been increased, ROA that will be generated will also increase. These findings are in accordance with Abbas et al. (2019), Kuncoro and Anwar (2022), Agustin (2018), Afrilien (2023) and Murtiningrum and Wahyuningsih (2024), while the conclusions of this study disagree with Ardichy (2022) and Hakim (2023). Therefore, we can conclude that this impact justified the need for management to always maintain a fixed solvency level in order to be compatible with Indonesian banks' regulations.

Indeed, according to research by Poniman (2022) and Kamal (2023), liquidity does not have any influence on the ROA of Indonesian. However, the conclusions of this investigation are inconsistent with Wangsawidjaja (2012), Ibrahim (2017), Hanifa et al. (2022), Ardichy (2022), Hakim (2023) and Sah and Pradhan (2023), who have shown that there is a positively significant effect between liquidity and ROA. This explained that the number of loans issued to clients increases ROA. In this context, we can conclude that the level of our banks' liquid assets improves their profitability. This finding runs counter to financial theory, which tends to establish a link between risk and return.

Also, the asset quality has a positively significant association with the ROA. Moreover, our finding is consistent with hypotheses H3. This finding is similar with Linda et al. (2023), Cheruiyot (2016) and Ogbulu and Eze (2016). Therefore, these investigations suggest that banks should consider investing in assets as they enhance financial performance. Other studies such as Widyatmoko and Risman (2024) and Shittu and Abdulkadir (2023) show in their research that the effect between asset quality and ROA is positive but insignificant. However, our conclusions disagree with Jean (2022), who found that there is no impact between asset quality and ROA because of the vulnerability of the banks.

The findings of the analysis are in line with the theory of real options because, on the one hand, this theory is useful and, on the other hand, it encourages a more positive perception of liquidity and solvency risk, which means that the majority of these practices reveal a strong aversion to risk on the part of operators and find it difficult to estimate the weighted average cost of capital associated with a project. Real options must therefore be applied appropriately, as they enable risks to be assessed and strategies to be created to mitigate these risks.

The notion of real option is applied to various fields such as oil exploitation projects. To this end, some researchers have studied the unfavorable effect of oil prices on performance. For example, Lee and Lee (2019) asserted this effect for Chinese banks from 2000 to 2014. However, these researchers claim that these negative impacts are eliminated by economic and political stability. ROA is adversely affected by the cost of oil may be linked to a reduction in its balance sheet operations. On the one hand, a higher petrol price has a negative impact on economic activity and on borrowers' capacity to meet their obligations, which can weaken banks' balance sheets. On the other hand, the fall in household demand and investment activity has a negative impact on banks' fee income, leading to a reduction in off-balance sheet activities. The negative influence of oil price volatility on corporate profitability is less visible for companies with a higher market value. In other words, large listed companies focus on share issues to make up financing shortfalls when the effect of oil price volatility causes their profits to fall. This is incompatible with Alyousfi et al. (2020).

Table 4 indicates an unfavorable effect between asset quality, liquidity and ROA, with a value of −0.654. This means that when liquidity improves by one point, profitability falls by 65.4%, even if the improvement is significant. The negative value suggests that the most aggressive banks are engaged in liquidity operations and are therefore unprofitable. Nevertheless, in countries such as Tunisia, where people tend to default on their loans, greater liquidity can be inefficient because it raises the cost of collecting cash.

Table 4 also reveals that asset quality harms the association between the asset quality, solvency and ROA is negatively significant with a value of −0.737. This means that funded banks are in a stronger situation to withstand any adverse shocks. In Tunisia, the higher cost of capital implies that it is costly to hold more capital efficiency.

For the control parameters, the results revealed a significantly positive link between the size of the bank and ROA. This is because big banks tend to manage risk more effectively and efficiently (Charmler et al., 2018). GDP has a positive connection with ROA; the higher is Tunisia's GDP, the higher is the ROA of public banks. This finding is inconsistent with Soesetio et al. (2022) and is consistent with Phan et al. (2020). However, these results showed a negatively significant relationship between the INFL and ROA. This finding is consistent with Wahyudi (2020), Soesetio et al. (2022) and Kamal (2023) and is inconsistent with Phan et al. (2020).

5. Robustness test

The GMM is also more effective for checking potential endogeneity problems and confirming the robustness of our main results (Blundell and Bond, 1998; Bond and Hoeffler, 2001). In fact, the Hansen test and the Arellano–Bond test for AR (2) in first difference are adequate. We designed the following regression model.

(2)ROAit=β0+β 1  ROAit−1+β 2 Solvencyit+β 3 Liquidityit+β 4 SIZEit+β 5 GDPit+β 6 INFLit+εit

The robustness of the GMM study presented in Table 5 validates the results presented in Table 4.

The impact of the liquidity is positively significant on ROA. This is consistent with Thi Doan and Bui (2021). As the increase in liquidity translates into an increase in loan income and a decrease in the cost of mobilizing deposits, the bank's ROA will rise. Nevertheless, liquidity must be kept at a reasonable level. Banks concentrate their lending in lower-risk segments to limit liquidity risk and improve profitability.

CAR has a significantly positive effect on ROA. The higher the CAR, the greater the bank's ability to manage credit risk or its profitability. This is consistent with Ramadhanti et al. (2019), which show that CAR is a ratio to support risk-controlling assets. In addition, the high level of CAR can promote public confidence in domestic banks, as the guarantee of funds for the community is increasingly strong.

6. Conclusion

Financial ratios are essential in banks, as they provide a means of analyzing profitability. Furthermore, the financial ratios provided by this analysis can be exploited to identify developments and trends and to improve the quality of information for banks. In addition, this investigation can be used by Tunisian banks to understand the economic climate in Tunisia during the last health crisis.

In summary, solvency, liquidity and the moderator variable had a significantly and positively influence on the ROA. This means that these two ratios significantly increased the profitability of our banks. The GLS and GMM regression show the same conclusions, confirming that the link between solvency, liquidity and ROA is positively significant.

This paper's implication might be helpful to the Tunisian Government as well as investors in highlighting the crucial role of the financial ratios in monitoring and assessing firm profitability, particularly in the case of financial difficulties. In addition, an effective liquidity and solvency management policy helps to guarantee a bank's capacity to face up to its treasury and capital adequacy, which are random elements because they are affected by external factors and by the behavior of other actors. In our future research, we will exploit other sectors of activity and other variables that are not considered in this research in order to prove their effect on the company's profitability and the profit of managers, investors and decision-makers.

Table 2

Correlation matrix

	ROA	Solvency	Liquidity	Asset quality	SIZE	GDP	Inflation
Panel A: correlation matrix
ROA	1
Solvency	0.248**	1
Liquidity	0.293***	−0.079	1
Asset quality	0.130	0.083	0.146	1
SIZE	0.512***	0.531***	−0.178*	0.524	1
GDP	0.060	0.075	−0.094	0.345	0.140	1
Inflation	−0.042	0.107	−0.180	0.514	0.120	0.736***	1
Panel B: VIF
		1.34	1.09	1.33	1.50	3.51	3.51

Note(s): ***, ** and * indicate the significance level of 1, 5 and 10% respectively

Source(s): Created by authors

Table 3

Results of the tests

	p-value
Hausman	0.9645
Heteroscedasticity	0.0008***
Auto-correlation	0.0000**

Note(s): ** and *** indicate the significance level of 5 and 1%, respectively

Source(s): Created by authors

Table 4

The link between solvency, liquidity and ROA in Tunisian banks: moderating impact of asset quality

Variables	ROA
	Coefficients	p>/Z/
Solvency	0.020	0.049**
Liquidity	0.008	0.020**
Asset quality	0.870	0.050**
Asset*liquidity	−0.654	0.000*
Asset*solvency	−0.737	0.010**
SIZE	0.030	0.030**
GDP	0.009	0.651
Inflation	−0.840	0.097***
Constant	−0.170	0.119
Wald χ2	10.12	0.078***

Note(s): *, ** and *** to indicate the significance level of 10%, 5 and 1%, respectively

Source(s): Created by authors

Table 5

Impact of the solvency and liquidity of banks in Tunisia on ROA via GMM system

Variables	ROA
	Coefficients	p>/Z/
ROAt−1	0.150	0.052**
Solvency	1.715	0.010***
Liquidity	0.041	0.001***
SIZE	0.120	0.000***
GDP	0.161	0.728
Inflation	−0.870	0.031**
Constant	−0.227	0.096***
Wald χ2	591.14	0.000***
AR(1)	−1.28	0.200
AR(2)	0.37	0.713
Hansen test	6.67	0.573
Sargan test	203.93	0.000***

Note(s): ***Significant at 1%

Source(s): Created by authors

Table A1

Variable measurements

Variables	Abbreviation	Formulas
Dependent variable
Return on assets	ROA	Net income/total assets
Independent variables
Solvency ratio	SOL	Capital adequacy
Liquidity ratio	LIQ	Loan/deposit
Moderator variable
Asset quality	AQ	Loan portfolio
Control variables
Bank SIZE	SIZE	Log TA
GDP growth	GDP	Log real GDP per capita
Inflation rate	INFL	[(CPIt − CPIt-1)/CPIt − 1] × 100

Source(s): Created by authors

Appendix

Table 1

Descriptive statistics

	Mean	Std	Min	Max
ROA	0.0359	0.0458	0.0002	0.1087
Solvency	0.0040	0.0030	0.0001	0.1020
Liquidity	0.2014	0.0989	0.0041	0.6093
Asset quality	0.0450	0.0780	0.0035	0.4470
SIZE	7.3489	0.1866	6.7897	7.2941
GDP	4.0339	0.0363	3.9829	4.1025
Inflation	6.4690	1.2106	5.64	8.31

Source(s): Created by authors