Effect of Credit and Liquidity Risks on Banking Financial Stability: Empirical Validation on Commercial Banks in Burundi (2016-2023) ()
1. Introduction
Recent financial crises have caused severe tensions in the financial systems of several countries, leading to the bankruptcy of many banks, as well as low economic growth. Thus, the importance of having a stable financial system became apparent, and since then, financial stability has become one of the essential missions of central banks (Gürbüz, 2022). In the past, banks that were financially stable were able to demonstrate their robustness during various episodes of financial crises.
In the wake of the upheavals, regulatory authorities have gradually adopted policies aimed at strengthening the stability of their financial systems in order to promote sustainable economic growth. At the internal level, the risk control and monitoring process implemented in banks advocates for the prevention of any systemic risk. Despite a range of mitigation measures, credit risk and liquidity risk remain among the main risks significantly affecting the financial stability of banks, as shown by the conclusions of the studies conducted by (Bensiahmed & Azzaoui, 2023; Ayinuola & Gumel, 2023; Chai, Sadiq, Ali, Malik, & Hamid, 2022; Ghenimi, Chaibi, & Omri, 2017; Ismail & Ahmed, 2023; Hakimi, Zaghdoudi, Zaghdoudi, & Djebali, 2017).
This study aimed to understand the influence of credit and liquidity risks on the financial stability of Burundian commercial banks.
2. Literature Review
The relationship between credit risk and banking financial stability has been widely discussed in the literature. While Sang Tang My (2022) shows an indirect relationship between credit risk and financial stability, authors such as Ayinuola & Gumel (2023), Chai et al. (2022), Ismail & Ahmed (2023), Hakimi et al. (2017), Gwachha (2023), Hassan, Khan, & Paltrinieri (2019), and Rupeika-Apoga, Zaidi, Thalassinos, & Thalassinos (2018) demonstrate that credit risk is a key factor threatening the stability of a bank, or even the banking system as a whole. In fact, according to them, the higher level of credit risk, the greater the potential losses, which can affect the bank’s financial capacity and increase the likelihood of banking instability. The same authors also mention that credit risk makes them more vulnerable to economic and financial shocks.
In the same vein, a negative relationship between liquidity risk and financial stability has been demonstrated in the research conducted by Ayinuola & Gumel (2023), Hakimi et al. (2017), and Hassan et al. (2019). The conclusions of their research converge on the fact that a liquidity crisis can lead to a bank’s bankruptcy if it fails to obtain additional funds quickly. At this level, they thus recommend that banks effectively manage this risk to ensure their sustainability and the stability of the financial system as a whole.
3. Methodological Approach
The methodological approach includes sampling, data collection and processing, analysis and discussion of the results.
3.1. Population and Sample
The population concerned by this study consists of fourteen commercial banks established in Burundi, of which four banks were created from 2020 fiscal year. Given that the defined study period is from 2016 to 2023, eight commercial banks for which data were available. Throughout the entire period were selected as a sample based on a reasoned choice depending on the availability of data at the time of collection.
3.2. Data Collection Technique
The documentary technique allowed us to collect quantitative data from the targeted banks. For the period from 2019 to 2023, as the central bank ordered commercial banks to annually publish data on their websites (BRB, 2019), the data were collected from the website of each commercial bank consecutively. About the data from the period of 2016 to 2018, they were collected from the reports and financial statements filed in the archives of each bank. In addition, the consultation of books, articles, annual reports from the Bank of the Republic of Burundi, reports from the World Bank, and the International Monetary Fund greatly helped us enrich the work.
3.3. Method of Data Processing and Analysis
To successfully process the data, we used Excel and Eviews 9 software. The first software allowed us to calculate the various ratios used in our work, and Eviews 9 enabled us to perform econometric treatments based on panel data.
Before analyzing the effect of each independent variable on the financial stability of commercial banks in Burundi, and after verifying the stationarity of our variables, we first tested for multicollinearity among the independent variables using the variance-inflation factor (VIF) test to verify the independence of the explanatory variables.
Furthermore, we performed the Hausman test to choose between the fixed-effects and random-effects models. After, we estimated our equation using econometric tests on panel data, with a grouped fixed-effects model and the least squares (LS and AR) methods. The significance of the effect of these variables on banking stability was analyzed based on their probability ≤ 5%. Indeed, a probability less than or equal to 1% indicates a highly significant effect, while a probability less than or equal to 5% indicates moderate significance.
3.4. Model Specification
The model adopted for our article is inspired by the following model: Yit = β0 + βkXit + βcXt + εit used by Ayinuola & Gumel (2023) when examining the link between liquidity and credit risks and their impact on banking stability in Nigeria with a sample of 12 banks for the period from 2010 to 2021. For him, Yit: the financial stability of bank i at time t;
Xit: the internal characteristics of bank i at time t,
Xt: the external variables, εi: the error term, and βk: the regression coefficient for the Explanatory variables.
For our work, by replacing Yit, Xit, and Xt with the respective variables, that is to say:
For our work, by replacing Yit, Xit, and Xt with the respective variables, that is: ϖ Yit: the financial stability of bank i at time t.
Yit: the financial stability of bank i at time t measured by the two following ratios:
Xit: The internal characteristics of bank i at time t which are:
COVit: Coverage rate of classified receivables by provisions of bank i at time t.
Tit: Size of bank i at time t.
DIVit: Diversification of bank i at time t.
Xt: External variables:
εi: the error term
βk: regression coefficient for the explanatory variables.
Our complete model gives us the following equation:
Yit = β0 + β1CRit + β2LRit + β3COVit + β4SIZEit + β5DIVit + β6INFt + β7PIBt + εit
With: CR (Credit Risk); LR (Liquidity Risk); COV (Coverage Ratio of Classified Loans); SIZE (Bank Size); DIV (Diversification of Banking Products); INF (National Inflation); and PIB (Gross Domestic Product). Such an equation was also used by Bensiahmed & Azzaoui (2023) in their analysis of the influence of credit and liquidity risks on the financial stability of Algerian banks over the period from 2013 to 2019 with a sample of 15 banks.
4. Determination of Variables and Expected Signs
In the literature, the Z-score often reappears as a measure of the financial stability of banks. We intend to analyze its reaction to the fluctuations of the variables selected for our work, which include six bank-specific variables: credit risk, liquidity risk, economic profitability, bank size, the ratio of classified asset coverage, and the diversification of banking products, as well as two external variables: the inflation rate and GDP.
We plan to analyze its reaction following the fluctuations of the variables selected for our work, which include six specific variables related to banks, namely credit risk, liquidity risk, economic profitability, bank size, the ratio of classified loan coverage, and the diversification of banking products, as well as two external variables, namely the inflation rate and GDP. The table below presents the measurements of the selected variables in accordance with existing literature, along with the expected signs.
Table 1 below presents the measurements of the selected variables in accordance with the existing literature, as well as the expected signs.
For those that are control variables, we used those that frequently appear in the literature, namely economic profitability, bank size, loan coverage ratio, diversification of banking income, GDP, and inflation (Bensiahmed & Azzaoui, 2023; Ayinuola & Gumel, 2023; Chai et al., 2022; Ghenimi et al., 2017; Ismail & Ahmed, 2023; Sang Tang My, 2022; Hassan et al., 2019; Jameel & Siddiqui, 2023; Zaghdoudi, 2019; Armstrong & Caldwell, 2008; Crouhy, 2000; Omri, 2022; Anh & Phuong, 2021; Kaharuddin & Yusuf, 2022; Sghaier & Soyah, 2012; Abbad, Achouche, & Tadjeddine, 2015; Hakimi et al., 2017; Abbes, 2017; Tomuleasa, 2017; Khelifa & Trari Medjaoui, 2023; Gwachha, 2023; Nisar, Peng, Wang, & Ashraf, 2018; Adusei, 2015; Pham, Dao, & Nguyen, 2021; Kiemo, Olweny, Muturi, & Mwangi, 2019; Diaconu & Oanea, 2014; Athari, Irani, & Hadood, 2023). As for the expected signs, they were determined by referring to the empirical results of the economies of developing countries in general, and in particular, those whose economic context is similar to that of Burundi (Ayinuola & Gumel, 2023; Chai et al., 2022; Ismail & Ahmed, 2023; Hakimi et al., 2017; Hassan et al., 2019; Zaghdoudi, 2019; Nisar et al., 2018).
Table 1. Measurement of variables and the expected signs for independent variables.
Variables |
Expression |
Measure |
Sign Expected |
dependent variable |
Financial Stability |
Z-Score (Y) |
Z-Score = (ROA + (FP/TA)/σ(ROA) |
|
independent variables |
1. Variables of interest |
Credit risk |
CR |
Total credit/Total assets |
- |
Liquidity risk |
LR |
Total crédit/Total dépôts |
- |
2. Control Variables |
2.1. Bank-specific variables |
Coverage ratio of classified receivables |
COV |
Total Provision/Total classified loan |
+ |
Bank Size |
SIZE |
Ln (total assets) |
+ |
Diversification of Banking Products |
DIV |
Not interest income/Total Assets |
+ |
2.2. Macroeconomics Variables |
Inflation |
INF |
Indice des prix à la consommation |
- |
Gross Domestic Product |
EDP |
Economic growth rate |
+ |
Source: Authors based on the literature review.
5. Presentation, Interpretation, and Discussion of the Results
The following Table 2 presents the results of our empirical analysis.
The results of the previous table show that the VIF values range from 1.05 to 1.32. Since all these values are well below 5, we conclude that there is no multicollinearity among the independent variables.
The analysis of the results from the previous Table 3 shows that the probability of the Chi-sq. statistic for the variable Y (Z-score) is greater than 0.05 (5%). This allows us to adopt the random effects model to econometrically determine the effects of credit risk and liquidity risk on the stability of commercial banks in Burundi.
According to Table 4, the estimation results show R2 and adjusted R2 values of 0.875473 and 0.839893, respectively. This indicates that our model is generally strong, as the independent variables collectively explain more than 83% of the variation in the financial stability of the studied banks. Moreover, the probability associated with Fisher’s statistical value (0.0000) indicates that the quality of our model’s fit is better. Regarding individual significance, the results obtained can be summarized as follows.
Table 2. Results of multicollinearity test.
independent variable |
R2 |
VIF = 1/(1 − R2) |
Tolerance = 1/VIF |
CR |
0.0498 |
1.0524 |
0.9502 |
LR |
0.2475 |
1.3288 |
0.7525 |
DIV |
0.1524 |
1.1798 |
0.8476 |
COV |
0.1499 |
1.1763 |
0.8501 |
SIZE |
0.2304 |
1.2994 |
0.7696 |
PIB |
0.1976 |
1.2463 |
0.8024 |
INF |
0.1760 |
1.2136 |
0.8240 |
Source: Author, based on findings from the Eviews 9 software.
Table 3. Result of hausman test.
Correlated Random Effects—Hausman Test |
Equation: Untitled |
|
|
Test cross-section random effects |
Test Summary |
Chi-Sq. Statistic |
Chi-Sq. d.f. |
Prob. |
Cross-section random |
0.000000 |
7 |
1.0000 |
* Cross-section test variance is invalid. Hausman statistic set to zero. |
Source: Author, based on findings from the Eviews 9 software.
Table 4. Result of the Pooled square method.
Cross-section random effects test equation: |
Dependent Variable: Y |
Method: Panel Least Squares |
Date: 02/05/26 Time: 10:11 |
Sample: 2016 2023 |
Periods included: 8 |
Cross-sections included: 8 |
Total panel (balanced) observations: 64 |
Variable |
Coefficient |
Std. Error |
t-Statistic |
Prob. |
C |
15.00622 |
3.348158 |
4.481933 |
0.0000 |
CR |
3.546749 |
1.534474 |
2.311377 |
0.0251 |
DIV |
−4.421142 |
4.622439 |
−0.956452 |
0.3435 |
COV |
1.454791 |
1.187674 |
1.224908 |
0.0265 |
LR |
−0.308241 |
0.296673 |
−1.038994 |
0.3039 |
INF |
−0.015420 |
0.028687 |
−0.537522 |
0.5933 |
PIB |
10.19370 |
17.54228 |
0.581093 |
0.5638 |
SIZE |
0.501921 |
0.143836 |
3.489544 |
0.0010 |
Effects Specification |
Cross-section fixed (dummy variables) |
R-squared |
0.875473 |
Mean dependent var |
5.590547 |
Adjusted R-squared |
0.839893 |
S.D. dependent var |
4.275935 |
S.E. of regression |
1.710944 |
Akaike info criterion |
4.113655 |
Sum squared resid |
143.4391 |
Schwarz criterion |
4.619643 |
Log likelihood |
−116.6370 |
Hannan-Quinn criter |
4.312989 |
F-statistic |
24.60627 |
Durbin-Watson stat |
1.940282 |
Prob (F-statistic) |
0.000000 |
|
|
|
Source: Author, based on findings from the Eviews 9 software.
5.1. Credit Risk
According to the results of our model, credit risk has a positive and statistically significant effect on the financial stability of banks at a confidence level of 97.49%. An increase of 1% in credit risk is associated with a 3.5467% increase in the financial stability of the banks analyzed. Consequently, our hypothesis that credit risk negatively affects the financial stability of banks is rejected. Interviews conducted with executives of the commercial banks included in the study confirm this result, as they emphasized that credit risk positively influences financial stability when the credit portfolio is highly performing.
Moreover, according to the Bank of the Republic of Burundi (BRB, 2023), during the study period, the Burundian banking sector recorded a credit portfolio deterioration rate of 3.3%, which remains within an acceptable range since it is below the 5% threshold. The same source reports that, in response to this deterioration, banks set aside adequate provisions. This finding is further justified by the fact that interest income accounts for more than 59% of Net Banking Product (NBP), highlighting the importance of credit operations in bank performance. However, these results contrast with previous studies reporting a negative effect of credit risk on financial stability, such as Bensiahmed and Azzaoui (2023) for Algerian banks (My 2020), Ayinuola and Gumel (2023) for Nigerian banks, Ismail and Ahmed (2023) for Jordanian banks, Hakimi et al. (2017) for Tunisian banks, Gwachha (2023) for Nepalese banks, and Rupeika-Apoga et al. (2018) for the Latvian banking sector.
5.2. Liquidity Risk
According to the results of our model, liquidity risk (LR) has a negative but statistically insignificant influence on the financial stability of commercial banks, since the probability associated with this variable is above the 10% threshold (30.39%). Therefore, our hypothesis that liquidity risk negatively affects the financial stability of banks is confirmed, although the effect is not statistically significant. According to the same sources, when funding liquidity is low, traders are reluctant to take positions, particularly on securities with high margins. This reduces market liquidity and increases asset price volatility.
Our findings on this hypothesis are consistent with those of Ismail and Ahmed (2023) in Jordan, Adusei (2015) in Ghana, and Kaharuddin and Yusuf (2022) in Indonesia, who also found that liquidity risk does not have a significant effect on the financial stability of the banks studied. In contrast, other studies, such as Bensiahmed and Azzaoui (2023) in Algeria, Jameel and Siddiqui (2023) in Pakistan, Zaghdoudi (2019) in Tunisia, and Gwachha (2023) in Nepal, demonstrate that liquidity risk does significantly influence financial stability. Furthermore, Ayinuola and Gumel (2023) in Nigeria and Jameel and Siddiqui (2023) in Pakistan show that liquidity risk negatively impacts financial stability.
Regarding the control variables, the results indicate that COV and SIZE positively and significantly influence the financial stability of Burundian commercial banks.
For COV, comparable findings were obtained by Bensiahmed and Azzaoui (2023) for Algerian banks and Bekri and Hachicha (2019) for Tunisian banks, who argue that the stability of Tunisian banks despite rising credit risk is largely due to deposit guarantee schemes, risk-sharing mechanisms, and the use of risk derivatives. Regarding the size of the banks, our findings align with those of Khelifa and Trari Medjaoui (2023) in the Gulf region, Sang Tang My (2022) in Vietnam, Jameel and Siddiqui (2023) in Pakistan, Gwachha (2023) in Nepal, Adusei (2015) in Ghana, Pham et al. (2021) for Vietnamese listed commercial banks.
6. Conclusion
The main objective of this article was to analyze the influence of credit and liquidity risks on the financial stability of Burundian commercial banks over the period 2016-2023. To test our hypotheses that credit risk negatively influences the financial stability of Burundian commercial banks and that liquidity risk negatively influences their financial stability, we employed an econometric approach using panel data from 8 commercial banks. The results indicate that our model is globally significant and exhibits a strong fit, with the independent variables collectively explaining more than 83% of the variation in the financial stability of the banks studied. At the individual level, credit risk was found to have a positive and statistically significant effect on the financial stability of Burundian commercial banks at the 5% threshold. Liquidity risk, by contrast, had a negative but statistically insignificant effect. Regarding the control variables, the classified receivables coverage ratio (COV) and bank size (SIZE) were all found to strengthen financial stability.
Given the limited availability of research on banking financial stability in Burundi, particularly on the impact of credit and liquidity risks, this study provides a valuable reference for researchers interested in the subject. Given its managerial implications, suggestions have been made to the managers of commercial banks and the financial system regulator:
To bank managers
Effectively manage both liquidity and credit risks to ensure the stability of banks.
Comply with regulatory standards to minimize non-performing loans that may harm financial stability.
Strive to maximize bank performance, which is a key pillar of financial stability.
Continue promoting financial education to help clients adapt quickly to new banking products.
To the financial system regulator
Strengthen the regulatory framework to improve liquidity risk management.
Develop the financial market to provide banks with alternative sources of financing beyond equity and deposits.
Encourage financing in growth sectors to increase production, stimulate savings, and thereby boost bank deposits.
Perspectives for future research
While this study focused primarily on credit and liquidity risks and was limited to a sample of 8 banks, future research could broaden the analysis by including additional determinants of financial stability. Extending the scope of the study to other East African Community (EAC) countries would also allow for cross-country comparisons of the factors influencing banking stability across different financial systems.
NOTES
*PhD Student at the University of Burundi, Bujumbura, Burundi.