OVERVIEW OF RESEARCH
Introduction
Banks serve as crucial financial intermediaries that facilitate a country's development by channeling funds within the economy through various financial services, including accepting deposits and providing business loans When loans are approved, banks earn profits through interest rates and service fees; however, they also face credit risk due to the potential for borrowers to default, resulting in non-performing loans (NPLs) This risk is often influenced by fluctuations in the macroeconomic environment, which can significantly affect the revenues and operational activities of borrowers.
This paper will explore the impact of economic determinants on bank credit risk The chapter will cover the background of the issue, outline the problem statements, and define the research objectives and questions Additionally, it will highlight the significance of the research and provide an overview of the chapter's layout.
The financial sector in developing countries has experienced remarkable growth due to economic expansion and financial liberalization, coupled with advancements in technology and management practices that enhance decision-making in banks However, the banking systems were significantly impacted by major economic recessions in 1997 and 2007, leading to a decline in the quality of bank assets and a sharp rise in non-performing loans (NPLs), closely linked to the economic cycle.
When borrowers fail to meet their loan obligations, it creates credit risk for banks, representing a major concern among various risks faced by commercial banks Credit risk comprises two key components: systematic and unsystematic credit risk (Castro, 2013) Developing an effective credit risk management policy is challenging due to the unpredictable nature of economic conditions.
The interplay between the business cycle and credit risk significantly influences banking-specific factors and poses various risks within the banking industry This relationship has sparked considerable concern among researchers and policymakers, highlighting the need to comprehend how credit risk fluctuates with economic cycles to maintain the stability of the banking system.
The recent financial crisis was triggered by the collapse of Lehman Brothers, the fourth-largest investment bank in the U.S., primarily due to the subprime mortgage crisis that resulted in numerous loan defaults and severe bank illiquidity This loss of confidence led to massive withdrawals by depositors, leaving banks unable to operate effectively and contributing to the bankruptcy of Washington Mutual Additionally, Lehman Brothers' global operations exposed banks worldwide to significant credit risk.
Banks traditionally provide loans, but this practice introduces credit risk due to borrowers' inability to repay According to Castro (2013), an increase in bad loans can lead to liquidity and insolvency issues, signaling a potential banking crisis When banks face illiquidity and insolvency, they struggle to meet their obligations, risking shutdowns and resulting in losses for both banks and borrowers, ultimately impacting the economy Therefore, raising awareness of credit risk is essential to identify its causes and prevent banks from facing liquidity and insolvency challenges.
To effectively manage credit risk, banks must comprehend the underlying factors contributing to it According to Garr (2013), the unpredictable nature of the macroeconomic environment, combined with various microeconomic influences, complicates credit risk management for banks A deficiency in knowledge and experience in this area can expose banks to heightened risks.
Ratnovski (2013) argues that credit risk management can become a burden for banks, consuming valuable resources and time Managers must invest significant effort in acquiring knowledge and experience, which can lead to increased administrative costs, especially when low returns on highly liquid assets fail to offset these expenses Implementing an effective credit risk program requires substantial time and resources, including capital and labor, to safeguard against sudden credit risk threats If credit risk policies are not aligned with the actual factors influencing credit risk, banks may not only waste money and time on ineffective programs but also face a considerable increase in credit risk issues.
Researchers and policymakers are increasingly focused on identifying the factors contributing to bank credit risk Understanding these factors is essential for developing effective credit risk management strategies aimed at minimizing the likelihood of credit risk occurrences.
This study analyzes the impact of macroeconomic factors on the non-performing loans (NPLs) ratio in five ASEAN countries: Indonesia, Malaysia, the Philippines, Thailand, and Vietnam, over a 13-year period.
From 2002 to 2015, the development rates in the area remained consistent; however, due to the unavailability of non-performing loans (NPLs) data at the country level, this study will analyze the NPL ratios of individual commercial banks To avoid bias and ensure model consistency, additional bank-specific factors will be incorporated Data has been gathered from 162 commercial banks, with macroeconomic determinants sourced from the World Bank and bank-specific data obtained from Fitch's International Bank Database The primary objectives of this paper are outlined accordingly.
- To examine the impacts of macroeconomic determinants to the NPLs ratio of the commercial banks in the five countries of ASEAN.
- To study the nature of the commercial banks’ specific factors toward the NPLs in the five countries of ASEAN.
- To find an appropriated method to measure the relationship between macroeconomic factors and the NPLs ratio
- To ensure the consistent of the chosen method through the application of robustness check and additional analytical tests.
- Give recommendation to policy makers.
The questions of this paper will be raised to match with the objectives above, these are as follows:
- Which is the macroeconomic factor that significantly effects the NPLs ratio in the commercial banks of the five ASEAN countries?
- How do banks’ management in these countries affect their NPLs?
This paper will examine the impacts of five macroeconomic factors to the NPLs rate, thus the five hypotheses are as follows:
H1: Gross Domestic Product (GDP) has a significant negative relationship with bank credit risk in the five studied ASEAN countries.
H2: Interest rate has a significant positive effect on bank credit risk in the five studied ASEAN countries.
H3: Inflation rate has a significant impact on bank credit risk in the five studied ASEAN countries.
H4: Exchange rate appreciation has a significant relationship with bank credit risk in the five studied ASEAN countries.
H5: Unemployment rate has a significantly positive impact on bank credit risk in the five studied ASEAN countries.
This study investigates the determinants of bank credit risk, focusing on 11 factors that include five key macroeconomic indicators and six bank-specific variables It is the first research to analyze the effects of these determinants on non-performing loans (NPLs) in commercial banks across five ASEAN countries—Indonesia, Malaysia, the Philippines, Thailand, and Vietnam—over the period from 2002 to 2015 Utilizing a comprehensive methodological design, the research employs dynamic panel data econometric techniques to ensure robust cross-validation of results, supplemented by additional analyses and robustness tests.
This research aims to enhance the understanding of key credit risk factors in commercial banks across the studied countries It will provide valuable insights into the causes of bank credit risk and assess bank performance regarding non-performing loans (NPLs) As highlighted by Demirguc-Kunt and Detragiache (1998), countries with high inflation, unemployment, and interest rates tend to experience increased bank credit risk, making banking crises more likely Consequently, this study will elucidate the relationship between economic developments and credit risk, offering insights into how banks operate within varying economic conditions.
Understanding the economic and bank-specific factors that influence bank performance enables investors and depositors to make informed decisions regarding their funds and avoid poor investments This knowledge also aids bank managers in developing effective loan and credit risk management policies by identifying the determinants that impact credit risk For instance, factors such as rising inflation rates, interest rates, or domestic currency appreciation can guide banks in monitoring and controlling risk exposures more accurately Ultimately, a robust credit risk management strategy enhances capital allocation, improves banking performance, and increases operating efficiency and profitability.
LITERATURE REVIEWS
Theoretical reviews
Credit risk refers to the potential loss lenders face when borrowers are unable to repay loans, either partially or fully Recent years have seen banks worldwide suffer significant losses and capital reductions due to a decline in asset quality, which has heightened their vulnerability to economic crises and limited their lending capabilities This situation has direct and indirect repercussions on financial stability and economic activities, underscoring the importance of credit risk analysis Such analysis is essential not only for maintaining a stable banking system conducive to economic growth but also for alerting regulatory authorities to avert future crises According to Castro (2013), credit risk can be categorized into systematic and unsystematic risk Systematic credit risk is influenced by macroeconomic factors, shifts in economic policies, and political changes, while unsystematic credit risk is determined by individual-specific factors like personality and financial solvency, as well as company-specific factors such as management quality and financial health.
The interplay between the economy and the financial system is explored through various theories, particularly in the context of business cycles Messai and Jouini (2013) highlight the connection between macroeconomic factors and loan quality, linking financial vulnerability and banking performance to business cycle movements Their work builds on Williamson's (1987) theoretical models, which focus on credit risk and the influence of business cycles on a country's financial sector Furthermore, Messai and Jouini (2013) provide a theoretical review of how different phases of the business cycle affect banking performance, examining the relationship between macroeconomic indicators—such as GDP growth, real interest rates, inflation rates, exchange rates, and unemployment rates—and the quality of loans.
During periods of economic expansion, the incidence of bad loans remains low, as borrowers typically feel secure in their financial stability and ability to meet repayment deadlines This confidence often leads lenders to relax credit standards and take on additional risks, as they are less concerned about potential defaults.
Research indicates that improved repayment capabilities of borrowers can lower credit risk for lenders (Salas and Saurina, 2002) However, during economic downturns, studies by Jiménez and Saurina (2006) and Bohachova (2008) reveal that banks face heightened vulnerabilities related to adverse selection and moral hazard among creditors, ultimately leading to an increased risk in loan portfolios.
Higher interest rates, primarily driven by monetary policy, significantly contribute to the debt burden due to increased interest payments, resulting in a higher rate of non-performing loans (NPLs) As interest rates rise, borrowers may face adverse selection issues, leading them to pursue riskier projects for higher returns instead of safer investments This scenario increases the likelihood of credit risk on banks' balance sheets, as they benefit from higher returns on new and floating-rate loans while borrowers struggle with elevated payments Additionally, banks mitigate risks by diversifying their financial roles, lending to numerous borrowers, and borrowing from many depositors In countries with interest rate liberalization, the costs of funds rise, prompting banks to charge higher rates to high-risk borrowers, ultimately increasing their overall risk exposure.
During economic downturns, the return on bank assets declines more sharply than the interest paid to depositors, leading to reduced profits or potential losses for banks Since banks primarily hold long-term fixed interest rate loans, they struggle to quickly adjust their return on assets Consequently, to manage their liability payments, banks often increase short-term lending interest rates.
In 1996, it was observed that borrowers facing significant debt burdens increase the risk for banks, resulting in a larger risky loan portfolio Consequently, banks require a higher net interest margin to offset the elevated risk of default, as noted by Ahmad and Ariff (2007) This situation can lead to systemic issues within the banking sector.
Inflation, driven by limited money supply growth and nominal depreciation of the domestic currency, significantly impacts both banks' lending decisions and borrowers' behaviors As inflation rises unpredictably, the prices of goods and services increase, leading to greater volatility in firms' profits and heightened debt obligations (Peyavali, 2015) This elevated inflation negatively affects the real rates of return on bank assets and the incomes of existing borrowers, ultimately deteriorating the quality of previously extended loans and resulting in credit rationing (Bohachova, 2008) Furthermore, in a variable loan rate environment, high inflation prompts adverse selection among borrowers, as banks may adjust lending rates to maintain stable real returns or governments may implement monetary policies to combat inflation (Nkusu, 2011).
Disinflation impacts loan quality by creating high real interest rates in a previously high-inflation economy This situation leads to decreased earnings for borrowers and encourages risk-taking behavior, akin to the effects of rising nominal interest rates (Mishkin, 1996).
Exchange rate, which indicates the value of domestic currency in terms of another, is also one of macroeconomic sources of economic instability as well as bank risk
Most banking panics in the United States are triggered by rising short-term lending interest rates, which heighten exposure to credit risk When domestic currency depreciates, borrowers with loans in foreign currencies face increased debts, leading to higher rates of loan defaults This situation particularly affects importers, as the cost of foreign goods rises, necessitating more domestic currency to purchase the same amount of imports, ultimately reducing firms' profitability and their ability to service loans Additionally, banks may experience increased volatility and credit risk if their liabilities in foreign currencies exceed their foreign exchange assets Significant depreciation of the domestic currency can also prompt disintermediation, as depositors withdraw funds to invest in higher-yielding hard currency assets, resulting in capital shortages and further escalating bank credit risks.
Unemployment serves as a significant indicator of banking credit risk, particularly during economic downturns As the unemployment rate rises, households and individuals experience reduced incomes, leading to deteriorating cash flow and increased likelihood of loan defaults Similarly, in the corporate sector, a decline in production due to decreased consumer demand results in revenue losses and weakened liquidity, further heightening credit risk for banks (Castro, 2013).
Specifically, the relation between unemployment and NPLs are proposed by Lawrence
In their 1995 theoretical model on life-cycle consumption, Rinaldi and Sanchis-Arellano highlighted that an increase in unemployment leads to reduced income for borrowers, thereby elevating the likelihood of loan defaults due to diminished debt-servicing capacity To mitigate risk and safeguard bank capital, lenders tend to offer higher interest rates to clients deemed higher risk Their extended research in 2006 further emphasized that the probability of non-performing loans (NPLs) is influenced by the unemployment rate, which affects borrowers' current and future income expectations, as well as the lending rates set by banks Additionally, the model indicates that the volume of loans, investment amounts, and time preference rates also play significant roles in default probabilities.
During economic recessions, rising unemployment and declining corporate earnings can lead to an increase in non-performing loans (NPLs) and bank losses, as highlighted by Berge and Boye (2007) Higher unemployment exacerbates borrowers' debt-servicing challenges, while banks must adjust their loan provisions based on anticipated borrower income and expenses As these economic factors deviate from expectations, borrowers' ability to service debt diminishes, resulting in heightened credit risk for banks.
Empirical reviews
Gross Domestic Product (GDP) represents the total monetary value of all finished goods and services produced within a country's borders during a specific time frame This article utilizes the annual growth rate of real GDP at constant prices as a key indicator of economic activities and business cycles, which can significantly affect the banking system and associated bank risks Numerous studies indicate a notable negative relationship between GDP growth and non-performing loans (NPLs) For instance, Shu (2002) conducted stress testing on Hong Kong's banking sector to assess the fluctuations in loan quality.
Changes in macroeconomic determinants significantly impact borrowers' loan repayment abilities and banks' portfolio positions, which are identified as key risk factors The author concludes that economic growth correlates with increased corporate profitability, leading to reduced default rates and lower bank risk exposure, thereby facilitating more lending opportunities Utilizing Merton’s methodology, Jakubik (2007) reveals that declines in real GDP growth negatively affect banks’ loan portfolio quality due to shifts in corporate earnings, wage growth, and rising unemployment, ultimately increasing bank credit risk Similarly, Zribi and Boujelbène (2011) analyze a panel of ten Tunisian commercial banks from 1995 to 2008, finding that GDP growth adversely impacts bank credit risk.
Louzis et al (2012) utilize a dynamic panel approach to analyze various loan types, including consumer loans, business loans, and mortgages, within the Greek banking system from 2003 to 2009 Their findings indicate that borrowers' ability to repay loans is influenced by the economic cycle; specifically, non-performing loans (NPLs) tend to increase during economic downturns or periods of low GDP growth Conversely, during economic expansions, borrowers typically have adequate income to meet their repayment obligations Consequently, there is a negative correlation between NPLs and the economic cycle, with NPLs rising during times of economic slowdown, which adversely affects the quality of all loan types.
A study by Castro (2013) utilizing dynamic panel data from 1997q1 to 2011q3 highlights the significant relationship between GDP growth and bank credit risk in Greece, Ireland, Portugal, Spain, and Italy during the recent financial crisis The findings indicate that higher GDP growth correlates with a reduction in non-performing loans (NPLs), as increased income for borrowers leads to improved cash flows, enhancing bank profitability and decreasing bad debts Similar conclusions are supported by Nkusu (2011) in a study of 26 advanced countries from 1998 to 2009, and by Messai and Jouini (2013) in their analysis of Italy.
Greece and Spain for the period of 2004-2008; or Klein (2013) in case of Central,Eastern and
South-Eastern Europe (CESEE) in the period of 1998–2011; or Chaibi and Ftiti (2015) in case of comparison between French and German economy.
Several researchers have found no significant relationship between GDP growth and bank credit risk For instance, Poudel (2013) studied 31 Nepalese commercial banks from 2001 to 2011 and found no correlation between GDP and non-performing loans (NPLs) This can be attributed to banks exercising caution during economic downturns, where they rigorously assess borrowers' creditworthiness before issuing new loans Additionally, banks proactively categorize clients to manage NPLs and credit risk, resulting in reduced credit volume during periods of low GDP growth Similar findings have been reported by Kalirai & Scheicher (2002) in the Austrian banking system, Fofack (2005) in Sub-Saharan African banks, and Aver (2008) in Slovenia.
Interest rates play a crucial role in understanding the relationship between interest rates and credit risk, as they directly impact borrowers' debt burdens This study focuses on the real interest rate due to data availability, anticipating a positive correlation Furthermore, various interest rates are often interconnected; for instance, an increase in the interbank rate typically results in a rise in the monetary policy interest rate, which subsequently drives up money market rates and the yields on long-term fixed-income securities (Bohachova, 2008).
Fofack (2005) identifies a positive correlation between real interest rates and credit risk in Sub-Saharan Africa, indicating that higher interest rates elevate borrowing costs for borrowers and reduce commercial banks' profits due to increased deposit costs, ultimately leading to a rise in default rates Similarly, Jiménez and Saurina (2006) employed the Generalized Method of Moments (GMM) estimator for dynamic panel models to explore the effects of real interest rates on loan losses.
A study examining Spanish commercial and savings banks from December 1984 to December 2002 revealed a significant positive correlation between interest rates and loan losses Similarly, research by Curak et al (2013) on banking systems in Southeastern Europe indicates that higher real interest rates increase the likelihood of non-performing loans (NPLs) for variable-rate loans, as rising rates impose greater financial burdens on borrowers, making it more challenging for them to meet their payment obligations.
Research by Castro (2013) in GIPSI countries indicates that long-term interest rates serve as the most suitable benchmark for analyzing credit risk, as banks primarily offer long-term loans The study reveals a significant positive correlation between interest rates and credit risk Furthermore, robustness checks replacing long-term interest rates with real interest rates and interest rate spreads yield consistent results Notably, long-term interest rates are crucial for assessing credit risk, especially when loan interest rates fluctuate Higher interest rates increase obligations for both corporate and individual borrowers, thereby elevating banks' credit risk Similar findings are echoed in Quagliariello's (2007) research, which correlates long-term interest rates, represented by ten-year Italian Treasury bonds, with credit risk indicators like loan loss provisions Additionally, the findings of Solarin et al support these conclusions.
A study by Asari et al (2011) utilizing the Auto Regressive Distributed Lag (ARDL) approach reveals a significant positive long-term impact of interest rates on non-performing loans (NPLs) in Malaysian Islamic banks Additionally, their research on commercial banks from 2006 to 2010, employing the vector error correction model, demonstrates a strong long-run relationship between interest rates and NPLs, while indicating that interest rates do not significantly affect NPLs in the short run.
Ali and Daly (2010) conducted a study utilizing credit risk logit regression to analyze the banking systems in the USA and Australia over a 14-year period from the first quarter of 1995 to the second quarter of 2009 Their findings revealed no significant relationship between nominal interest rates or short-term interest rates (6-month) and credit risk in either country.
Inflation reduces the purchasing power of currency as the general price level of goods and services rises, impacting both interest rates and the banking sector's efficiency The relationship between inflation and non-performing loans (NPLs) can be complex; while higher inflation may decrease the real value of loans, making repayment easier for debtors, it can also diminish profitability and increase capital costs, ultimately weakening borrowers' repayment capacity (Curak et al., 2013).
Numerous studies indicate that inflation significantly impacts banks' credit risk Research by Demirguc-Kunt and Detragiache (1998) demonstrates that high inflation exacerbates risk issues within the banking sector, particularly through its association with elevated nominal interest rates, which complicate banks' ability to manage maturity transformation Furthermore, when restrictive monetary policies are enacted to combat inflation and stabilize the banking sector, they often lead to increased real interest rates, heightening the risk of banking crises In a study focused on India, Thiagarajan et al (2011) found a positive correlation between current inflation and credit risk in public sector banks, while no significant relationship was observed in private sector banks Additionally, the research by Badar & Javid further explores these dynamics.
A study conducted by Curak et al (2013) highlights a significant positive correlation between inflation and bank credit risk, analyzing the effects of macroeconomic factors on non-performing loans (NPLs) across 36 Pakistani commercial banks from 2002 to 2011 The research indicates that rising inflation negatively impacts the profitability of banks and heightens credit risk When the Consumer Price Index (CPI) increases, contractionary fiscal policies lead to higher interest rates, which elevate borrowing costs and diminish borrowers' real income, ultimately impairing their ability to repay loans This results in an increase in NPLs and a decline in bank profitability Additionally, a similar study involving 69 banks across 10 Southeastern European countries from 2003 to 2010 confirms that inflation significantly influences credit risk, as currency instability and variable interest rates reduce the real value of incomes, making it challenging for debtors to meet their loan obligations.
In the opposite direction, there are researchers argue that inflation has a negative relationship with credit risk Using bank-level data for 80 countries in the year 1988-
Research by Demirguc-Kunt & Huizinga (1999) indicates a negative correlation between banks' credit risk and inflation, suggesting that banks may benefit from higher income through bank float during inflationary periods This is due to delays in crediting customer accounts, allowing banks to maintain a lower cost than their net interest margin, thus enhancing profitability Similarly, Poudel (2013) found a negative relationship between inflation and credit risk in Nepalese banks from 2001 to 2011, noting that rising interest rates during high inflation lead banks to limit loan volumes and focus on higher quality loans in secure sectors This careful assessment of borrower quality further reduces credit risk, aligning with findings from Shu (2002) and Zribi.
& Boujelbene (2011), Vogiazas & Nikolaidou (2011) and Washington (2014), which also conclude negative relationship between inflation and bank credit risk.
Conclusion
Research indicates that macroeconomic factors significantly influence loan quality, although findings vary based on the banking systems and institutional contexts of different countries This paper employs a dynamic panel data approach to analyze commercial banks across five ASEAN member states, focusing on both macroeconomic determinants and bank-specific factors.
Page | 22 control variables are investigated and assumed to contribute a various important roles on the bank risk of
Page | 23 default Thus, these factors will take a substantial part in the explanation of the credit risk in this study.
Research Hypothesis
Previous research indicates that the economic environment plays a crucial role in influencing credit risk behavior This study aims to analyze the impact of five independent macroeconomic variables on bank credit risk across five ASEAN member countries The following hypotheses will be tested:
GDP growth is a critical indicator of the economy's cyclical position and has a significant impact on bank credit risk Research shows that advanced economies with higher GDP are less prone to banking crises (Demirgỹỗ-Kunt & Detagiache, 1998) A study by Jakubik (2007) reveals that a decline in GDP increases credit risk probabilities due to rising unemployment and deteriorating company profitability and asset values During economic expansions, borrowers typically generate sufficient income to manage their debts, resulting in lower non-performing loans (NPLs) (Chaibi & Ftiti, 2015) However, prolonged periods of economic growth can lead to lending to lower-quality borrowers, which often results in increased NPLs when a recession occurs (Castro, 2013).
Gross Domestic Product (GDP) serves as a fundamental measure of the economy's cyclical position A low GDP growth rate adversely impacts corporate earnings, wage growth, and can lead to increased unemployment or asset prices, ultimately resulting in a decline in loan portfolio quality Consequently, there exists a negative relationship between GDP and bank credit risk, supporting the first hypothesis.
H1: Gross Domestic Product (GDP) has a significant negative relationship with bank credit risk in the five studied ASEAN countries.
Interest rate is a strong determinant of credit risk because it influences the debt burden of borrowers, there have positive relationship between interest rate and credit risk A
The recent 24% rise in interest rates has negatively impacted the quality of bank loan portfolios, as it raises financing costs for both corporations and households This increase in costs diminishes the market value of assets and heightens the debt burden for borrowers, whether corporate or individual (Jakubik, 2007; Chaibi).
& Ftiti, 2015) Consequently, borrowers’ ability of pay back loan is weaken that leads to a higher rate of NPLs (Nkusu, 2011; Louzis et al., 2012; Castro, 2013).
According to asymmetric information theories, higher interest rates can worsen the issue of "adverse selection," leading to a higher likelihood of selecting borrowers with poor project outcomes (Mishkin, 1996) The instability associated with elevated interest rates discourages borrowers from pursuing low-risk projects, resulting in a shift in the risk profile of loan applicants toward higher-risk individuals Furthermore, increasing interest rates can alter borrowers' incentives, prompting them to engage in riskier projects (Stiglitz & Weiss, 1981) Consequently, in an environment characterized by information asymmetries, a rise in interest rates is likely to elevate the credit risk reflected on banks' balance sheets.
H2: Interest rate has a significant positive effect on bank credit risk in the five studied ASEAN countries.
Inflation significantly impacts the real cost of loans, influencing borrowers' ability to service debt and the overall performance of the banking sector While the relationship between inflation and non-performing loans (NPL) remains ambiguous, higher inflation can reduce the real value of outstanding loans, thereby enhancing borrowers' repayment capacity Consequently, this creates a negative correlation between inflation and bank credit risk Numerous studies support this connection, including research by Shu (2002) on Hong Kong banks, Zribi & Boujelbene (2011) in the Tunisian banking system, and Vogiazas & Nikolaidou (2011) in Romania.
(2013) in the case of Slovenian banking system and Poudel (2013) in Nepal.
However, inflation can also limit debtors’ ability to service debt expressed in both nominal and floating rate loans (Curak et al., 2013) by diminishing real income when
Rising inflation leads to sticky wages, which diminishes cash flow for loan repayments and reduces the value of bank assets, consequently affecting credit rationing To sustain real returns and profitability, banks adjust their lending rates, increasing the financial burden on both new and existing borrowers with variable rate loans Researchers, including Demirguc-Kunt and Detragiache (1998), Rinaldi & Sanchis-Arellano (2006), and Thiagarajan et al (2011), conclude that there is a positive correlation between inflation and non-performing loans (NPLs).
& Javid; 2013; Curak et al., 2013) In order to clearly find out the relationship between inflation rate and NPLs, it will be tested by the following hypotheses:
H3: Inflation rate has a significant impact on bank credit risk in the five studied ASEAN countries.
The identification of bank non-performing loans (NPLs) related to foreign exchange rates hinges on the relationship between currency fluctuations and a bank’s exposure to foreign exchange According to Lindgren et al (1996), the overall credit risk for banks concerning exchange rates is influenced by the net loan exposure of borrowers involved in import and export activities When the domestic currency depreciates, banks may increase loans to importers to mitigate credit risk in the exporting sector Bochahova (2008) highlights that excessive exchange rate volatility can trigger banking crises by undermining both economic and financial stability Specifically, a depreciated domestic currency can elevate credit risk for banks with a high liabilities-to-assets ratio in foreign currency Additionally, significant currency depreciation may lead to disintermediation, as depositors withdraw funds to invest in higher-return “hard currency assets,” resulting in capital shortages for banks and an escalation in credit risks.
Countries at different stages of economic development must enhance their financial market efficiency to foster growth and competitiveness In developing nations, banking systems often allocate a significant portion of loans to export-import activities, particularly within the manufacturing sector, which is highly susceptible to exchange rate fluctuations Currency depreciation increases the cost of imported inputs and raises the burden of foreign-currency-denominated debt for firms, while appreciation undermines the profitability and competitiveness of export-oriented businesses Consequently, firms face challenges in servicing their debt obligations, and banks' credit risk exposure is closely tied to the net import-export rate.
Refer to above discussions, the exchange rate have related to credit risk for commercial banks and the following hypotheses are proposed:
H4: Exchange rate appreciation has a significant relationship with bank credit risk in the five studied ASEAN countries.
The unemployment rate serves as a crucial indicator of a country's economic health and significantly influences credit risk It reflects borrowers' capacity to meet loan obligations and highlights their debt levels, which in turn presents challenges for banks in their lending practices aimed at stimulating economic growth.
Research indicates that rising unemployment rates adversely affect household income, leading to reduced consumption and savings as families struggle to manage debt This decline in spending can lower market demand, subsequently decreasing overall production levels and firm revenues Additionally, increased production costs, including wages, may hinder firms' ability to repay loans Consequently, higher unemployment rates are associated with an elevated risk of credit defaults, suggesting a positive correlation between unemployment and bank credit risk.
H5: Unemployment rate has a significantly positive impact on bank credit risk in the five studied ASEAN countries.
DATA AND METHODOLOGY
Data collection
This study analyzes the Non-Performing Loans (NPLs) rate of individual commercial banks as the dependent variable, due to the limited availability of country-level NPL data Key macroeconomic factors, including Real GDP growth, inflation, real interest rates, exchange rates, and unemployment rates, are utilized as independent variables, alongside various bank-specific determinants as control variables Data for macroeconomic indicators is sourced from the World Bank, while NPL rates and bank-specific factors are obtained from the Bank Scope-Fitch’s International Bank Database The analysis focuses on a sample period from 2002 to 2015, covering five ASEAN countries to ensure data validity and sufficient time length.
This research focuses exclusively on commercial banks, including state-owned, joint-stock, city, rural, and foreign banks, ensuring that data is available for at least three consecutive years Commercial banks play a crucial role in the banking industry by offering a wide range of financial services and investment products aimed at profitability, unlike state banks, investment banks, and Islamic banks, which have different objectives The study encompasses a balanced panel sample of 162 banks and 1,263 annual observations from five ASEAN member countries, with detailed bank counts provided in Appendix 1.
Econometric methodology – The NPLs measurement
In this research, selecting an appropriate methodological approach is crucial, aligning with the theoretical frameworks and empirical results while accommodating data constraints This paper aims to identify effective cross-country modeling techniques that leverage both time series and cross-sectional dimensions of the dynamic panel sample to yield optimal results.
Several studies have employed cross-country modeling techniques; however, the diverse issues, institutional contexts, and governance systems across nations can introduce uncertainty into cross-country board research To address this challenge, collecting data from individual banks and analyzing the overall economic environment of each country can enhance the understanding of both temporal and cross-sectional dimensions of dynamic board performance.
This research utilizes a balanced panel data set due to limitations in data collection Following recent literature on panel data, a dynamic approach is employed to address the time persistence in the structure of non-performing loans (NPLs) by incorporating a lagged variable of NPLs The analysis adopts a dynamic panel data specification to effectively capture these relationships.
Where: δ is a constant term; the subscripts i = 1,…,N and t = 1,…,T respectively indicate the cross sectional and time dimension of the panel; ��� � � is the dependent variable
NPLs; � � is a k×1 vector of explanatory variables (macroeconomic and microeconomic factors) other than � � −1 ; β is a k×1 vector of coefficients; � � the unobserved individual
(country and bank specific) effects and � � is the error term.
Traditional panel data estimators can introduce econometric bias, as the OLS estimator becomes biased and inconsistent even without serial correlation In dynamic panel data models, the RE estimator also exhibits bias, while the FE estimator remains consistent with a larger time dimension (T) Given that this study examines a relatively small time frame of 14 years, the correlation between the lagged dependent variable and unobserved effects can lead to significant bias, making the use of these estimators inappropriate for this analysis.
To address biases in traditional estimators, econometric methods like instrumental variables and first-differencing can be employed Prior research (Louzis et al., 2012; Castro, 2013; Chaibi & Ftiti, 2015) highlights the generalized method of moments (GMM) estimator, introduced by Arellano and Bond (1991) and enhanced by Arellano and Bover (1995) and Blundell and Bond (1998), as more reliable for dynamic panel models incorporating lags of the dependent variable This study utilizes the GMM estimator to effectively mitigate the identified biases By applying the first difference transformation of the initial equation, the dependent variable with a lag of order j+1 can be integrated to ensure the relevant moment conditions, resulting in a revised equation.
After applying differencing in equation (3), the individual level effects are removed; however, a new bias emerges due to potentially endogenous explanatory variables Consequently, the dependent variable is now correlated with the new error term.
Arellano and Bond (1991) propose that under the assumption of serially uncorrelated error terms and predetermined explanatory variables, the moment conditions can be satisfied by using lags of the dependent variable of order two or higher, along with current and lagged values of the explanatory variables.
The orthogonality restrictions outlined are fundamental to one-step GMM estimation, ensuring consistent parameter estimates under the assumption of independent and homoscedastic residuals across both cross-sectional and temporal dimensions This methodology, known as Arellano–Bond, effectively addresses the challenges of estimating parameters in dynamic panel data models.
First-order autocorrelation (AR1) in error terms does not lead to inconsistent estimates Both the dependent and predetermined or endogenous variables are instrumented using their lagged levels, while the differences of strictly exogenous independent variables are instrumented with their own values It is essential to ensure the absence of second-order autocorrelation in the differenced equation, as highlighted by Arellano and Bond (1991) The presence of instruments is crucial for obtaining consistent GMM estimates in the moment conditions.
Arellano and Bond (1991) introduced a two-step GMM estimator that initially assumes independence and homoscedasticity of the error term across countries and over time This method utilizes recovered residuals to create a consistent variance-covariance matrix for the moment conditions The two-step GMM estimator is more efficient and alleviates the homoscedasticity assumption compared to the one-step estimator However, Blundell and Bond (1998) highlight that lagged levels of persistent explanatory variables serve as weak instruments in the regression equation in differences, potentially inflating coefficient variance Conceptually, the differencing process of the two-step estimator eliminates individual-level influences and time-invariant explanatory variables, such as leverage and size, which are often nearly constant over time Consequently, the first-difference of these variables lacks informative value, leading to unidentified parameters in the first-differenced system.
The two-step GMM estimator shows limited improvements in efficiency, even when considering heteroscedastic errors (Judson and Owen, 1999) Additionally, its dependence on residual measurements from the one-step estimator can introduce bias in standard errors, leading to potentially unreliable asymptotic statistical inferences (Bond, 2002; Bond and others).
Page | 30Windmeijer, 2002) This problem should be considered especially in the case that the extent of cross-section data is
Page | 31 relatively small (Arellano and Bond, (1991); Blundell and Bond, 1998), which is correctly similar to the case of this research.
Roodman (2009) highlights that the difference and system GMM can lead to instrument proliferation, where bias arises from internal instruments generated from past observations of the instrumented variables This presents a trade-off between the lag distances for creating internal instruments and the sample depth for two-stage least squares (2SLS) analysis Roodman also notes that larger sample sizes can increase the number of instruments, potentially resulting in misleading asymptotic outcomes for estimators and specification tests, particularly in small sample studies Additionally, Chaibi & Ftiti (2015) identify issues from two perspectives: classical concerns regarding numerous instruments overfitting endogenous variables, and modern challenges related to feasible efficient GMM (FEGMM), which uses sample moments to estimate an optimal weighting matrix These issues can lead to both invalid and valid results simultaneously due to weakened specification tests.
Chaibi & Ftiti (2015) propose two methods to reduce the number of instruments generated in difference and system GMM The first method involves using specific instruments instead of all available lags, resulting in a linear instrument count in relation to time (T) while still covering the number of instruments per period This approach constrains the coefficients on certain lags to zero during the projection of regressors onto the complete instrument set The second method combines instruments by aggregating smaller sets, which preserves more information without dropping any lags This technique applies constraints to project regressors onto subsets of instruments with identical coefficients To address issues of instrument proliferation and overfitting, Roodman (2009) recommends using the moment conditions of the standard difference GMM.
This new moment condition is similar to the orthogonality of ��� � and �� �
However, the estimator is only required to play down the magnitude of the empirical moments
In the context of econometric analysis, instruments are utilized for each individual variable rather than pairing them with additional variables This approach leads to the creation of collapsed instruments, which keeps the instrument count linear in relation to T Missing values are substituted with zeros, as noted by Beck and Levine (2004) and Carkovic and Levine (2005) Furthermore, by collapsing instruments and restricting lag depth, it is possible to maintain a consistent instrument count regardless of T To ensure the consistency and efficiency of the GMM estimator, a set of moment conditions must be applied, contingent upon the assumptions of no serial correlation in the error terms and the validity of the instruments used.
The variables definition and measurement
3.3.1 The dependent variable – the Non-performing loans:
This paper selects the non-performing loans (NPLs) ratio as a key indicator of bank credit risk, following the methodologies established in previous studies (Louzis et al., 2011; Thiagarajan, 2013; Messai and Jouini, 2013; Klein, 2013; Chaibi & Ftiti, 2015) The NPLs ratio, defined as the proportion of impaired loans to total bank assets, significantly influences a bank's performance A rising NPLs ratio may signal increased credit risk, potentially leading to asset deterioration and negatively impacting both individual banks and the broader economic landscape.
The GDP growth rate represents the market value of all final goods and services produced within a country's borders over a specific period, serving as a key indicator of economic health and the economic cycle A decline in this growth rate often signals rising prices, reduced output, and increased unemployment, which can subsequently lead to a rise in non-performing loans (NPLs) for banks The real GDP growth rate is frequently utilized in academic research to analyze these economic trends (Badar and Jabid, 2013; Messai, 2013; Louzis et al., 2013; Castro, 2013).
This study will analyze the impact of the previous year's economic performance on current non-performing loans (NPLs) by incorporating the lagged real GDP growth rate into the model.
The relationship between inflation rates and non-performing loans (NPLs) is complex and multifaceted Higher inflation may negatively impact banks' asset quality and borrowers' incomes, potentially leading to an increase in NPL ratios Additionally, monetary policies aimed at controlling inflation can result in elevated interest rates, further exacerbating NPL issues Conversely, inflation can diminish the real value of loans, potentially enhancing borrowers' ability to repay Previous studies, including those by Bochahova (2008), Castro (2013), and Chaibi & Ftiti, provide insights into these dynamics.
2015), the inflation rate as GDP deflator will be applied in this paper.
Interest rates are a key factor in bank loans, significantly influencing market risk and forming an integral part of national monetary policy aimed at achieving monetary objectives set by central banks to counter inflation This study utilizes the real interest rate as a proxy for the interbank interest rate applied by commercial banks, highlighting its positive correlation with non-performing loans (NPLs) due to the increased debt burden on borrowers The relationship between real interest rates and NPLs has also been explored in the research conducted by Jimenez and Saurina (2006) and Bochahova.
The exchange rate plays a crucial role in analyzing the movement of Non-Performing Loans (NPLs), particularly when loans are in foreign currency This variable can have both positive and negative impacts This study focuses on the Real Effective Exchange Rate (REER), which reflects the competitiveness of a country's debt servicing capabilities in relation to its import and export sectors An appreciation of the REER signifies a strengthening of the domestic currency This indicator has been widely referenced in research by Bochahova (2008), Castro (2013), and Chaibi & Ftiti (2015).
The unemployment rate is closely linked to the economic cycle, influencing cash flow and consumer spending on goods and services This relationship highlights the positive correlation between non-performing loans (NPLs) and economic conditions.
The undeniable unemployment rate has garnered significant attention from researchers such as Bochahova (2008), Nkusu (2011), Castro (2013), and Chaibi & Ftiti (2015) This paper aims to explore the impact of unemployment on non-performing loans (NPLs) in five ASEAN countries.
3.3.3 Microeconomic variables – bank-specific determinants:
While macroeconomic variables are generally viewed as exogenous to the banking system, understanding the determinants of Non-Performing Loans (NPLs) within this context is crucial Individual banks' characteristics and management policies play a significant role in enhancing efficiency and risk management, which can influence NPL trends The selection of these variables is guided by theoretical frameworks and data availability, incorporating both bank-specific and macroeconomic factors to enhance the model's explanatory power.
Several studies, including those by Berger and DeYoung (1997), Salas & Saurina (2002), Louzis et al (2012), Castro (2013), and Chaibi & Ftiti (2015), explore the link between bank-specific factors and non-performing loans (NPLs) These studies analyze the causal relationship between loan quality and bank efficiency in terms of cost and management This article will incorporate and discuss relevant microeconomic variables in relation to the hypotheses presented in earlier research.
The solvency ratio indicates a bank's capital strength, calculated by the proportion of equity capital to total assets A higher solvency ratio suggests that a bank is better equipped to withstand potential shocks in the credit market, as noted by Berger and DeYoung.
Research from 1997 supports the 'moral hazard' hypothesis, indicating a negative relationship between bank capital and non-performing loans (NPLs) This hypothesis suggests that bank managers may take on higher risks in their loan portfolios despite having limited capital Conversely, Curak et al (2013) argue that even well-capitalized banks might engage in riskier lending practices, leading to potential loan losses due to the moral hazard behavior of managers Consequently, the impact of the solvency ratio on loan losses remains unclear.
Cost inefficiency, measured by the cost to income ratio, plays a significant role in understanding the efficiency-risk hypothesis proposed by Berger and DeYoung (1997), which examines the link between cost inefficiency and banks' credit risk The relationship between cost inefficiency and non-performing loans (NPLs) can vary, demonstrating either a positive or negative correlation This relationship has been analyzed through three hypotheses: "Bad management," "Bad luck," and "Skimping," which explore the causality between these variables.
The "bad management" hypothesis suggests that banks may struggle with loan appraisal and borrower monitoring due to insufficient skills and experience in managing internal costs This ineffective management results in low cost efficiency, which is likely to contribute to an increase in future non-performing loans (NPLs).
The "bad luck" hypothesis suggests that bad loans occur due to unforeseen events that surpass banks' expectations and control Consequently, banks must reallocate resources and incur additional expenses to manage these problematic loans, leading to decreased cost efficiency.
Econometric strategy – The system GMM estimator
The system GMM (SGMM) estimator, introduced by Arellano and Bover (1995) and Blundell and Bond (1998), builds upon the difference GMM (DGMM) method established by Arellano and Bond (1991) When dealing with small sample sizes, SGMM demonstrates superior estimation performance compared to DGMM.
The SGMM approach enhances the estimation of a dynamic exchange rate model by addressing endogeneity and accommodating fixed effects, as established by Arellano & Bond (1991), Arellano and Bover (1995), and Roodman (2006) This method processes variables to eliminate fixed effects, ensuring that the instruments remain exogenous Additionally, Blundell and Bond (1998) refine the original DGMM by simultaneously evaluating differences and levels, where the differenced equation utilizes lagged values as instruments, and the levels equation is instrumented by the first-difference of lagged values (Roodman, 2009) Consequently, the SGMM retains fixed effects and accurately accounts for bank heterogeneity in its estimations.
This study utilizes a limited sample period, focusing solely on the first lag of the dependent variable while treating other lags of regressors as instruments All bank-specific factors and the first lag of non-performing loans (NPLs) are classified as endogenous variables, whereas macroeconomic variables are considered exogenous Roodman developed the SGMM syntax and implemented it using the xtabond2 command in STATA The optimal model is determined based on criteria established by Arellano and Bond (1991) and Roodman (2006, 2009), as detailed in Appendix 2 The final model of this paper is constructed using selected variables and references from previous studies.
� 10 ���� � −1 + � 11 ������ � + � � (7) Where: δ is a constant term; the subscripts i = 1,…,N and t = 1,…,T indicate the cross sectional and time dimension of the panel respectively; � � is the error term;
��� � −1 is the dependent variable with a lagged;
SVCR, LEVER, INEFF, NOINTINC, LOGSIZE, and ROA are critical bank-specific factors that include the solvency ratio, leverage ratio, cost inefficiency, non-interest income, the logarithm of total assets, and returns on total assets.
UNEMP, RINT, REER, GDPG and INFGDP: is macroeconomic determinants, which respectively are unemployment, real interest rate, real effective exchange rate, realGDP growth rate with 1 lag and inflation rate.
RESULTS AND DISCUSSIONs
Summary statistics
Table 1 summarizes the statistics of key variables across five ASEAN countries, revealing a disparity in observation numbers, with macroeconomic variables reaching a peak of 2,268 and non-performing loans (NPLs) at a minimum of 1,607 The NPLs rate fluctuates between -0.449% and 68.838%, averaging around 4.879%, indicating banks' management of impaired loans Furthermore, the real GDP growth rates reflect significant economic advancement, with an average of approximately 5.392% Unemployment and inflation rates stand at 5.629% and 6.919%, respectively, showcasing the current economic landscape of these nations Notably, real interest rates exhibit considerable variation across the region.
The study of five countries reveals significant variations in bank-specific determinants, including leverage ratio, cost efficiency, average equity returns, and non-interest income These differences can be attributed to the unique development stages, institutional frameworks, and cultural contexts of each nation.
Variable Observations Mean Std Dev Min Max
Unit root tests
This paper utilizes a balanced and small sample dataset, making the Fisher-type unit root tests, specifically the augmented Dickey–Fuller and Phillips-Perron tests, the most suitable choice These tests are performed with a time trend, panel mean, and one lag for all variables, with the null hypothesis stating that "all panels contain unit roots." Given the small and balanced nature of the sample, individual panel tests are conducted before combining the p-values to derive the overall test results Only the inverse chi-squared statistics are reported in Table 3, which demonstrates the rejection of the unit root null hypothesis for all variables, confirming their stationarity Additionally, correlation tests presented in Appendix 3 reveal no serial correlation among the variables analyzed in this study.
Unit root tests for NPLs estimations variables
Variable Augmented Dickey-Fuller Phillips-Perron
Note: * indicates that the unit root null hypothesis is rejected at the 1% significance level
Empirical results
To begin with, this research will following the three major principles as noticed byKlein (2013) Firstly, differences countries have difference regulations and
Page | 42 accounting systems, thus each country has its own separated classification of theNPLs or bad loans.
The inconsistency of Non-Performing Loans (NPLs) across countries leads to the assumption that their classification remains unchanged within individual nations, which can limit issues when unobserved fixed effects are accounted for in analysis Additionally, the true extent of NPLs may be obscured in some countries due to a significant number of restructured loans or those sold to asset management companies, potentially biasing estimation results and underestimating the actual health of the banking system Furthermore, this analysis is unable to account for various unobserved compositional impacts across countries, such as differences in housing versus consumption or corporate versus retail sectors, which may influence the determinants of NPLs.
The dynamic model (7) will be analyzed using both Fixed Effects (FE) and System Generalized Method of Moments (SGMM) estimation techniques The FE estimator offers superior control over heterogeneity and unobserved individual effects, such as bank-specific factors, compared to traditional Ordinary Least Squares (OLS) However, it may introduce endogeneity issues related to the lagged variable and fixed effects in the error term The SGMM, developed by Arellano and Bover (1995) and Blundell and Bond (1998), enhances efficiency in samples with limited time dimensions and high persistence by differencing the data and using lagged variables as instruments Following Roodman (2009), the analysis considers only the first lag of the dependent variable due to small sample sizes, ensuring the persistence of credit risk and addressing potentially unmeasured explanatory variables In this framework, lagged Non-Performing Loans (NPLs) and bank-specific factors are treated as endogenous, utilizing GMM-style instruments, while macroeconomic variables are considered strictly exogenous and instrumented as "IV style." The results from the FE and SGMM estimations are presented in Table 4.
Results with SGMM and fixed-effect estimations
All models were estimated with a constant, and robust t-statistics are provided in parentheses The significance levels for rejecting the null hypothesis are indicated as ***, 1%; **, 5%; and *, 10% The one-step SGMM estimator with difference lag was utilized for the model estimation Each regression includes the number of observations (No Obs.) The Arellano–Bond tests for first (AR1) and second-order (AR2) autocorrelation in first-differenced errors are presented, along with the statistics and p-values (in square brackets) for the Sargan-test of over-identifying restrictions and the Hansen-test for uncorrelation between instruments and residuals An appreciated real effective exchange rate indicates a strengthened domestic currency.
The analysis presented in Table 4 reveals that non-performing loans (NPLs) are affected by both macroeconomic and bank-specific factors, with varying impacts across different estimators Most coefficients align with expected signs, indicating their predictive power, while non-significant variables show opposite signs The residuals demonstrate no unit root or serial correlation, and the Hansen test confirms that the applied instruments are not correlated with the residuals in the SGMM estimator Furthermore, the Arellano-Bond tests indicate that while the first-order autocorrelation (AR1) hypothesis is rejected, the second-order autocorrelation (AR2) hypothesis holds true, which is anticipated due to the nature of differencing.
In comparing the FE and SGMM estimators, the signs of bank-specific factors differ, while macroeconomic factors remain consistent The fixed effect model excludes lagged NPLs due to endogeneity issues, although its coefficient is statistically significant at 0.55, suggesting a potential influence on the banking system, possibly due to prior year write-offs In the SGMM estimator, coefficients for variables such as SVCR, LEVER, LOGSIZE, and ROA show marginal increases, whereas the explanatory power of SVCR, LOGSIZE, RINT, lagged GDPG, and INFGDP significantly decreases The collected variables serve as short-term indicators, indicating that significant variables have immediate impacts on NPL rates in the surveyed year, while others demonstrate slower effects or no influence on NPL movements Detailed estimations will be addressed in the following section.
Table 4 indicates that the equity-to-assets ratio (SVCR) has a significantly positive impact on non-performing loans (NPLs) across various estimators, supporting the "moral hazard" hypothesis Regardless of whether banks in the studied countries are thinly or highly capitalized, NPLs are likely to rise due to poor management practices characterized by reckless risk-taking and a focus on short-term reputation (Louzis et al., 2013) Consequently, this behavior leads banks to undertake greater risks, resulting in an increase in NPLs.
The leverage ratio serves as a negative determinant of credit risk across five countries, demonstrating significance at the 5% level in the SGMM estimator This suggests that higher liabilities relative to equity correlate with a reduced likelihood of impaired loans, contradicting the "too big to fail" hypothesis and previous research findings, except for the case of French banks Conversely, it aligns with the insights of Salas and Saurina, as well as Zribi & Boujelbene, indicating that effective credit risk management practices enhance banks' efficiency and diversification, thereby mitigating the risk of bad loans Additionally, the leverage ratios of the studied banks cluster around a specific threshold, which bolsters their capital structures, improves cost efficiency, and ensures secure financial claims, addressing the need for liquidity amidst credit risk concerns.
The negative and insignificant sign of inefficiency (INEFF) in both estimators indicates that cost inefficiency does not significantly impact non-performing loans (NPLs) in banks across five countries This finding contradicts the "bad management" hypothesis, which posits that high cost inefficiency leads to increased NPLs, aligning instead with the results of Chaibi and Ftiti (2015) regarding the German banking system Here, banks appear to implement effective credit strategies and policies that manage credit risk, reduce operating costs, and enhance operating income This supports the "skimping" hypothesis, suggesting that banks allocate substantial resources to maintain loan quality Additionally, the results also endorse the "bad luck" hypothesis, indicating that banks may need to reallocate resources to address problem loans, particularly during unexpected financial crises like that of 2007.
The analysis of non-interest income (NOINTINC) as a proxy for the "Diversification" hypothesis reveals a negative correlation with non-performing loans (NPLs) in SGMM estimators, although this relationship is statistically insignificant This suggests that the additional business activities of banks in the five countries do not significantly contribute to NPLs These findings align with previous research by Salas and Saurina (2002), Louzis et al (2012) for GIPSI countries, and Chaibi & Ftiti (2015) in France, indicating that banks can mitigate the risk of NPLs by generating income from diverse financial activities beyond traditional loan granting.
The SIZE indicator, determined by the natural logarithm of total assets, shows a positive correlation with non-performing loans (NPLs) in the studied countries according to the SGMM, but a negative and significant relationship at the 10% level in the FE estimator This positive association aligns with findings by Louzis et al (2012) and Chaibi & Ftiti (2015), supporting the "too big to fail" hypothesis, where larger banks, by increasing leverage, engage in excessive risk-taking, resulting in higher NPLs Additionally, Laeven et al (2014) indicate that larger banks face greater credit risk due to the private interests of shareholders or managers, reinforcing the "moral hazard" and "bad management" theories In their extensive study of 52 countries, Laeven et al (2014) note that large banks dominate the interbank system, benefiting from economies of scale, while smaller banks rely on liquidity from these larger institutions Furthermore, due to lax regulations and expectations of government bailouts, large banks are more inclined to engage in riskier activities and rely on short-term financing, which can lead to liquidity shocks and potential bank failures.
5 service charges on deposits, trust fees, advisory fees, servicing fees, net trading profits from trading books, and commissions and fees from off balance sheet items.
The profitability of banks, measured by Return on Assets (ROA) ratios, significantly influences their risk-taking behavior, exhibiting a negative relationship at a 5% significance level This finding aligns with previous research by Godlewski (2005), Louzis et al (2012), Messai and Jouini (2013), and Asfaw and Veni (2015) Higher profitability is associated with fewer high-risk activities, supporting the notion that well-managed banks achieve greater cost efficiency and asset quality By implementing effective credit risk management systems, these banks reduce pressure from credit activities, thereby minimizing exposure to problem loans.
In summary, the effects of bank-specific factors on non-performing loans (NPLs) differ significantly, creating a trade-off for banks seeking to mitigate risks associated with poor loans Consequently, the influence of bank managers emerges as a crucial element in enhancing the efficiency of banks within a country's financial system.
This study explores the relationship between the macroeconomic environment and non-performing loans (NPLs), focusing on credit risk indicators within commercial banks across five ASEAN countries The findings provide valuable insights into how macroeconomic factors influence the levels of NPLs, highlighting the importance of understanding these dynamics for effective risk management in the banking sector.
The findings presented in Table 4 demonstrate that a rise in the unemployment rate, indicative of an economic downturn, correlates with increased non-performing loan (NPL) ratios, with both estimators showing significant results at the 5% level Specifically, the System Generalized Method of Moments (SGMM) analysis reveals that a 10 percentage point increase in the unemployment rate is associated with a notable rise in NPL ratios.
2.28 percent point increase in NPLs Therefore, the null hypothesis is rejected and theH5 is accepted This finding supports to the strong connection of the business cycles and the banking sector’s resilience and also validates to the results of prior papers(Nkusu, 2011; Messai and Jouini (2013); Vogiazas and Nikolaidou, 2013 and Bucur and Dragomirescu, 2014) For more specific, as the unemployment rate goes up, the income of both households and firms are squeezed so that they loss their capacity to serve the loan payment As a results, banks face the increase in the NPLs.
Similar to the unemployment rate, the results of the real effective exchange rate appreciation in both estimators indicates a statistically significant influence to the loan
OTHER ANALYSIS AND ROBUSTNESS CHECK
This section presents additional analyses and robustness checks using the SGMM estimator to ensure model consistency Initially, macroeconomic variables influencing credit risk were substituted with related variables to assess similar effects Despite these substitutions, the impact of the economic environment on non-performing loans (NPLs) remained consistent with regression (2), as detailed in Appendix 4 To further evaluate the robustness of the findings concerning data and estimation methods, alternative estimators and analyses involving time and level restrictions are provided, with results available in Appendix 5.
5.1 The Money supply (M2) vs the real GDP growth:
The Money supply (M2) serves as a substitute for real GDP growth due to their strong short-term correlation According to standard macroeconomic theory, an increase in the money supply leads to lower interest rates, which stimulates consumption and borrowing activities This dynamic typically results in a rise in total output and spending, subsequently boosting GDP Additionally, enhanced economic productivity increases the value of money in circulation, allowing for the exchange of more valuable goods and services per currency unit The outcomes of this analysis are detailed in Appendix 4.
The analysis reveals that all original variables maintain their directional consistency with the previous regression results Notably, the M2 variable, expressed as a percentage of GDP, exhibits a positive correlation with real GDP growth and significantly impacts non-performing loans (NPLs) at a 5% significance level These findings align with earlier research conducted by Fofack (2005), Badar & Javid (2013), and Nursechafia & Abduh.
An increase in the overall money supply can lead to a decline in the quality of a country's bank portfolio, primarily due to rising inflation, which in turn results in a higher rate of non-performing loans (NPLs).
5.2 The Interest rate spread vs the real interest rate:
The interest rate spread, which represents the difference between deposit interest rates and loan interest rates, is a primary source of income for banks, according to research by Castro.
In this study, the long-term interest rate is substituted with the interest rate spread (INTSP) due to the limited data sample, which makes the long-term rate unsuitable for analysis The sensitivity test reveals that while the coefficient on INTSP is positive and insignificant—similar to the real interest rate—the overall positive impact of interest rates on non-performing loans (NPLs) in the five ASEAN countries remains consistent Additionally, factors such as SVCR, LEVER, UNEMP, and REER continue to significantly influence NPLs, although their impact has diminished Despite these findings aligning with previous research, it is noted that, as suggested by Castro (2013), the long-term interest rate is more appropriate for assessing its effect on NPLs, given that loans are typically extended over longer periods, highlighting a limitation of this paper.
5.3 The GDP per capita growth vs the real GDP growth:
Few studies utilize GDP per capita as a proxy for the economic cycle to assess the macroeconomic impact on non-performing loans (NPLs), with some incorporating both real GDP growth and GDP per capita in their analyses (Fofack, 2005; Bochahova, 2008) GDP per capita serves as an indicator of overall income levels in a country, reflecting productivity and living standards However, it is essential to conduct sensitivity tests when substituting real GDP growth with GDP per capita.
According to Appendix 4, while the UNEMP variable decreases its significance from 1% to 5%, it remains positive, and the sign and explanatory power of other variables in regression (2) remain unchanged In comparison, the GDP per capita growth variable is also positive but insignificant when evaluated alongside the real GDP growth variable in regression (2).
Accordingly, movement of the NPLs in the banking system of the five ASEAN countries are not influenced by the changes of the GDP per capita.
The analysis focuses on the period from 2007 to 2015, revealing a reduction in observations to 912, as detailed in Appendix 5 Despite some changes in the explanatory power of variables, key relationships remain intact Notably, SVCR, LEVER, and REER have become insignificant, suggesting that banks have implemented effective credit risk management strategies amid shifts in capital structures Additionally, ROA and UNEMP continue to influence NPLs, while INEFF has gained significance at the 10% level These findings support the "skimping" hypothesis, indicating that during the financial crisis, banks utilized various resources and strategies to maintain loan portfolio quality, enhance cost efficiency, and mitigate the rise in NPLs.
In the upcoming analysis, Indonesia will be omitted from the sample As highlighted in the "Asian Development Outlook 2015" by the Asian Development Bank (ADB), Indonesia stands out as the largest economy in the region and has the highest net petroleum import rate among its peers Furthermore, Indonesia accounts for the largest share of banks in the sample, with a total of 71 banks, resulting in a significant reduction of observations to 838 and a total of 91 groups.
The analysis reveals that while the overall sign of variables remains unchanged, the UNEMP variable demonstrates increased explanatory power, and notably, RINT, LOGSIZE, and INEFF are significant in four countries In the macroeconomic context, rising real interest rates elevate debt burdens for individuals and firms Additionally, bank-specific factors highlight the significant influence of LOGSIZE and INEFF on non-performing loans (NPLs), suggesting that the banking systems in these countries may implement effective credit strategies.
Page | 54 and policy or increases their investment to manage the risk on the movement and quality of
The concept of "too big to fail" highlights the moral hazard associated with large banks and the regulatory challenges faced by countries in managing these financial giants This situation underscores the significant impact that these banks have on the overall health of the financial system, particularly in relation to the fluctuations in Non-Performing Loans (NPLs).
The robustness checks will evaluate data behavior using various estimators, starting with the two-step SGMM estimator introduced by Arellano and Bond (1991) This estimator maintains a similar number of observations and groups as the SGMM estimator Notably, the explanatory power of variables such as SVCR, ROA, and LEVER has increased, while LOGSIZE and INFGDP also show enhanced significance, in contrast to the decreased influence of UNEMP and REER The two-step estimator effectively eliminates individual-level effects, mitigating cross-bank relationships between bank-specific variables and non-performing loans (NPLs), as well as accounting for time-invariant explanatory variables typically found in cross-sectional data However, some variables, like leverage and size, exhibit near time-invariance Furthermore, the two-step GMM estimator does not significantly enhance efficiency, even when considering heteroscedastic errors (Judson and Owen, 1999) Additionally, its reliance on residuals from the one-step estimator introduces bias in standard errors, raising concerns about the validity of asymptotic statistical inferences (Bond, 2002; Bond and Windmeijer).
2002), which is worth to be considered in this research with small sample data.
In the analysis of small sample data with high persistence, the SGMM estimator demonstrates greater efficiency than the DGMM estimator However, the DGMM estimator's results are also utilized to assess the significance of any differences As shown in Appendix 5, the number of observations increased to 1,053 across 152 groups, with all variable signs remaining consistent with those found in the SGMM estimator.
Recent analysis indicates notable changes in the explanatory power of certain variables Specifically, the SVCR and LEVER variables have seen a decline, while the ROA has shown an increase The UNEMP variable has become insignificant, whereas the REER continues to maintain its relevance Additionally, the SGMM estimator proves to be more suitable than DGMM for variables characterized as "random walk" variables, according to Roodman.
CONCLUSION, POLICY IMPLICATIONS & LIMITATIONS OF THE
Main findings
Recent global financial sector challenges, particularly within the banking system, have negatively impacted asset quality and capital, increasing the risk of losses This has sparked significant interest among researchers and policymakers in understanding the causes of banking crises, primarily linked to credit risk, which can lead to illiquidity and insolvency The rise in credit risk is often reflected in the increase of non-performing loans (NPLs) on banks' balance sheets, particularly when borrowers struggle to repay Consequently, numerous studies have utilized various panel estimation techniques to analyze macroeconomic factors as key determinants of NPL growth This paper aims to investigate the effects of macroeconomic determinants on the NPLs of commercial banks in five ASEAN countries from 2002 to 2015 Additionally, it employs dynamic panel data methodology, specifically the SGMM estimator, to explore several bank-specific factors and their influence on risk management concerning NPL rates.
This paper conducts additional analyses and robustness tests to validate the consistency of results derived from the SGMM estimator The main findings remain robust despite alternative examinations and time-period restrictions, offering new perspectives Notably, macroeconomic factors, particularly the unemployment rate and exchange rate, significantly influence the movements of non-performing loans (NPLs) in commercial banks across five ASEAN countries An increase in NPL rates correlates with both the appreciation of the domestic currency and rising unemployment, signaling an economic downturn However, due to the limited sample size and short-term data, other variables like real GDP growth, real interest rates, and inflation show no significant impact on NPL rates Interestingly, the unexpected positive correlation with real GDP growth suggests complexities within the banking systems studied.
Page | 57 countries have an awareness to control their lending activities based on the movement of the business cycle.
The study reveals that bank-specific determinants of credit risk are significantly influenced by the banking system's solvency ratio, leverage, and returns on total assets in both pre-crisis and post-crisis periods These findings align with the "moral hazard" and "bad management" hypotheses, indicating a direct relationship between these indicators and the quality of bank management, which in turn affects asset quality and non-performing loans (NPLs) on bank balance sheets Improved management quality, coupled with reduced moral hazard incentives, is associated with lower levels of NPLs.
Policy implications
The findings suggest that certain variables may act as early warning indicators for future bank credit risk, highlighting the need for effective policy measures by both bank managers and national regulators At the country level, implementing appropriate policies is essential to foster a stable economic environment, which is crucial for maintaining the stability of the banking system For instance, ensuring price stability through monetary policy can help achieve optimal interest rates and control inflation.
To enhance productivity and bolster steady growth and employment, structural reforms are essential for improving a country's external competitiveness Regulatory authorities across the five ASEAN nations must intensify oversight in the financial sector, focusing on effective risk management systems It is crucial to monitor banking performance closely and identify institutions at risk of impaired loans and financial instability, enabling timely interventions to avert potential banking crises Additionally, policymakers should adopt a proactive stance on credit risk by removing tax and legal barriers, allowing banks to optimize portfolio management and strengthen their capacity to absorb losses.
To enhance management quality and strengthen credit risk management, banks should prioritize improving the efficiency of credit risk analysis and loan monitoring By doing so, they can effectively identify potential increases in bad debts, ensuring better financial stability and risk mitigation.
Page | 58 quality assets In addition, it is recommended that banks should limit excessive lending as well as loans denominated in foreign currency and maintain high credit and capital standards.
Limitations
This study acknowledges several limitations, primarily the small sample size and limited time span, which restrict the accuracy of the findings The reliance on annual data may not provide sufficient detail compared to quarterly or monthly data, potentially hindering a comprehensive understanding of banks' immediate responses to short-term monetary policy amid rapid interest rate changes Additionally, the research does not account for the varying impacts of country-specific institutions on banks' credit risk dynamics Furthermore, the focus remains on commercial banks in general, leaving out critical distinctions in credit risk determinants between private and state-owned banks.
Future research recommendation
To enhance the reliability of future research findings, it is advisable to increase the sample size Additionally, exploring credit risk determinants using advanced statistical methods like structural equation modeling or investigating the effects of potential nonlinear credit market frictions could provide valuable insights.
Future research could explore the regulatory, institutional, and legal factors influencing non-performing loans (NPLs) across different loan types—such as business, mortgage, and consumer loans—in five ASEAN countries, providing valuable insights for academic studies.
Investigating specific types of banks, including private and state-owned banks, will yield more accurate insights into bank credit risk factors This targeted analysis will enable policymakers to focus on and manage credit risk more effectively and appropriately for each banking category.
Anastasiou, D., Louri, H., Tsionas, M (2016) Determinants of non-performing loans:
Evidence from Euro-area countries Finance Research Letters. http://dx.doi.org/10.1016/j.frl.2016.04.008
Asian Development Bank (2015) Asian Development Outlook 2015 – Financing
Ahmad, N., Ariff, M., (2007) Multi-country study of bank credit risk determinants.
International Journal in Banking and Finance, 5 (1), 135–152.
Akinlo, O & Emmanuel, M (2014) Determinants of Non-Performing Loans in
Nigeria The Journal of Accounting & Taxation Vol 6, No 2, pp 21-28.
Ali, A., Daly, K., (2010) Macroeconomic determinants of credit risk: Recent evidence from a cross country study International Review of Financial Analysis, 19, 165– 171.
Anjom, W & Karim, A M (2016).Relationship between Non-Performing Loans and
Macroeconomic Factors with Bank Specific Factors: A Case Study on Loan Portfolios – SAARC Countries Perspective ELK Asia Pacific Journal of Finance and Risk Management, Volume 7, Issue 2.
Arellano, M., (2003) Modelling optimal instrumental variables for dynamic panel data models Centro de Estudios Monetarios y Financieros, Working Paper
Arellano, M., Bond, S., (1991) Some tests of specification for panel data: Monte
Carlo evidence and an application to employment equations Review of Economic Studies 58, 277.297.
Arellano, M., Bover, O., (1995) Another look at the instrumental variable estimation of error-component models Journal of Econometrics 68, 29.51.
Asari, M., Ahmad, Latif, A and Jusoff, K., (2011) An Analysis of Non- Performing
Loan, Interest Rate and Inflation Rate Using Stata Software World AppliedSciences Journal, 12, 41-48.
Asfaw, A H & P Veni, P (2015) Determinants of Credit Risk in Ethiopian Private
Commercial Banks International Journal of Accounting and FinancialManagement Research Vol 5, Issue 3, 1-14.
Aver, B., 2008 An empirical analysis of credit risk factors of the Slovenian banking system Managing Global Transitions, 6 (3), 317–334.
Badar, M., & Javid, A Y., (2013) Impact of macroeconomic forces on nonperforming loans: an empirical study of commercial banks in Pakistan WSEAS Transactions on Business and Economics, 10(1), 40-48.
Beck, R., Jakubik, P., Piloiu, A., (2013) Non-performing loans-What matters in addition to the economic cycle? Working Paper Series, European Central Bank, No 1515.
Berge, T.O., Boye, K.G., (2007) An analysis of bankfs problem loans Norges Bank
Berger, A., DeYoung, R., (1997) Problem loans and cost efficiency in commercial banks Journal of Banking and Finance 21, 849.870.
Blankespoor, E., Linsmeier, T J., Petroni, K R and Shakespeare, C., (2012) Fair value accounting for financial instruments: Does it improve the association between bank leverage and credit risk? Forthcoming at The Accounting Review.
Blundell, R., Bond, S., (1998) Initial conditions and moment conditions in dynamic panel data models Journal of Econometrics 87, 115–143.
Blundell, R., Bond, S., and Windmeijer, F (2000) focused on enhancing the effectiveness of the standard Generalized Method of Moments (GMM) estimator in dynamic panel data models Their work is featured in the book "Nonstationary Panels, Panel Cointegration and Dynamic Panels," edited by Baltagi, B., and published in the Advances in Econometrics series, volume 15, by JAI Press, Elsevier Science.
The banking sector is significantly influenced by macroeconomic factors, as highlighted in Bohachova's (2008) empirical assessment across various countries Additionally, Bond (2002) provides insights into dynamic panel data models, offering a comprehensive guide to microdata methods and their practical applications in economic research Together, these studies underscore the importance of understanding both macroeconomic influences and data methodologies in evaluating banking sector risks.
Bond, S., Windmeijer, F., (2002) Finite Sample Inference for GMM Estimators in
Linear Panel Data Models: A Comparison of Alternative Tests Mimeo,Institute for fiscal studies, London.
Bucur, I A & Dragomirescu, S E., (2014) The Influence of Macroeconomic
Conditions on Credit Risk: Case of Romanian Banking System Studies and Scientific Researches Economics Edition, No 19.
Carkovic and Levine (2005) explore the relationship between foreign investment and economic growth in their contribution to the book "Does Direct Foreign Investment Promote Development?" edited by Mortan, Graham, and Blomstrom Their research, published by the Institute for International Economics and the Center of Global Development, provides insights on whether foreign direct investment serves as a catalyst for economic development, highlighting key findings on its impact on growth dynamics.
Castro, V., (2013) Macroeconomic determinants of the credit risk in the banking system: The case of the GIPSI Economic Modelling, Volume 31, March 2013,
Cifter, A., Yilmazer, S., Cifter, E., (2009) Analysis of sectorial credit default cycle dependency with wavelet networks: evidence from Turkey Economic Modelling 26, 1382–1388.
Chaibi, H., & Ftiti,Z., (2015) Credit risk determinants: Evidence from a cross country study Research in International Business and Finance, 33, 1–16.
Curak, M., Sandra Pepur, S., Poposki, K., (2013) Determinants of non-performing loans – evidence from Southeastern European banking systems Banks and Bank Systems, Volume 8, Issue 1.
Dash, M., Kabra, G., (2010), The determinants of non-performing assets in Indian commercial bank: An econometric study Middle Eastern Finance and Economics, 7, 94-106.
DeAngelo, H and Stulz, R M., (2013) Why High Leverage is Optimal for Banks NBER
De Bock, R & Demyanets, A., (2012) Bank Asset Quality in Emerging Markets;
Determinants and Spillovers IMF Working Papers 12/71.
Demirguc-Kunt, A., Detragiache, E., (1998) The Determinants of Banking Crises and
Developed Countries IMF Staff Papers, 45 (1), 81-109.
Demirguc-Kunt, A., Huizinga, H (1999) Determinant s of commercial bank interest margins and profitability: some international evidence The World BankEconomic Review, 13(2), 379-408.
Farhan, M., Sattar, A., Chaudhry, A B., & Fareeha K., (2012) Economic determinants of non-performing loans: perception of Pakistani bankers European Journal of
Festic, M., Kavkler, A., Repina, S., (2011) The macroeconomic sources of systemic risk in the banking sectors of five new EU member states Journal of Banking and Finance 35, 310–322.
Fofack, H., (2005) Non-performing loans in sub-Saharan Africa: Causal Analysis and
Macroeconomic Implications World Bank Policy Research Working Paper no. 3769.
Gambera, M., (2000) Simple forecasts of bank loan quality in the business cycle.
Issue Series, No 3, Federal Reserve Bank of Chicago, Chicago.
Gunsel, N., (2012) Micro and macro determinants of bank fragility in North Cyprus economy African Journal of Business Management, 6(4), 1323-1329.
Hoggarth, G., Sorensen, S and Zicchino, L (2005) Stress tests of UK banks using a
VAR approach Working Paper no 282 Bank of England.
Hu, J., Yang, Li., Yung-Ho, C., (2004) Ownership and non-performing loans: evidence from Taiwan’s banks Developing Economies 42, 405.420.
Jakubik, P., (2007) Macroeconomic environment and credit risk Czech Journal of
Jiménez, G., Saurina, J., (2006) Credit cycles, credit risk and prudential regulation.
International Journal of Central Banking 2 (2), 65–98.
Judson, R., Owen, L., (1999) Estimating dynamic panel data models: a guide for macroeconomists Economics Letters 65, 9–15.
Juks, R., (2010) Why banks prefer leverage? Sveriges Riksbank, Economic Review
Garr, D, K., (2013) Determinants of Credit Risk in the Banking Industry of Ghana.
Garr, D, K., (2013) Macroeconomic and Industry Determinants of Interest Rate
Spread- Empirical Evidence Developing Country Studies, 3(12), 90-99.
Godlewski, C., (2005).Bank capital and credit risk taking in emerging market economies Journal of Banking Regulation 6(2), 128-145.
Kalirai, H and Scheicher, M (2002) Macroeconomic stress testing: preliminary evidence for Austria Financial Stability Report 3, 58-74.
Kelly, R (2012) House Prices, Unemployment & Irish Mortgage Losses Central
Klein, N., (2013) Non-Performing Loans in CESSE: Determinants and Impact on
Macroeconomic Performance IMF Working Paper, European Department,
Koch, T W and McDonald (2003) Bank Management 3rd edition,Sydney, Thomson. Lawrence, E., (1995) Default and the life cycle model Journal of Money Credit and
Laeven, L., Ratnovski, L and Tong, H (2014) Bank Size and Systemic Risk IMF
Lindgren, C.-J., Garcia, G., and M I Saal (1996) Bank Soundness and
Louzis, D., Vouldis, A., and Metaxas, V., (2012) Macroeconomic and Bank-specific
Determinants of Non-performing Loans in Greece: A Comparative Study of Mortgage, Business and Consumer Loan Portfolios Journal of Banking & Finance, Elsevier, 36, 1012.
Makri, V., Tsagkanos, A., Bellas, A., (2014) Determinants of Non-Performing Loans:
The Case of Eurozone PANOECONOMICUS, 2014, 2, pp 193-206
Messai, A and Jouini, F., (2013) Micro and Macro Determinants of Non-performing
Loans International Journal of Economics and Financial Issues 3, 852.
Mishkin, F (1996) Understanding Financial Crises: A Developing Country
Modigliani, F., Miller, M., 1958 The cost of capital, corporate finance, and the theory of investment American Economic Review 48, 261-297.
Nkusu, M., (2011) Nonperforming Loans and Macro financial Vulnerabilities in
Advanced Economies International Monetary Fund, WP No 11/161.
Nursechafia, N & Abduh, M (2014) The Susceptibility of Islamic Banks’ Credit
Risk towards Macroeconomic Variables Journal of Islamic Finance Vol 3, No.1.
Pesola, J., (2005) Banking fragility and distress: an econometric study of macroeconomic determinants Bank of Finland Research Discussion Papers,
Peyavali, J S (2015) The Impact of Macroeconomic Determinants on Non- performing Loans in Namibia International Review of Research in Emerging Markets and the Global Economy, 1 (4), 612-632.
Podpiera, J., Weill, L., (2008) Bad luck or bad management? banking market experience Journal of Financial Stability 4, 135–148.
Poudel, R P S., (2013) Macroeconomic determinants of credit risk in Nepalese banking industry In Proceedings of 21st International Business Research Conference.
Quagliariello, M., (2007) Banks riskiness over the business cycle: a panel analysis on
Italian intermediaries Applied Financial Economics, 17(2), 119-138.
Rajan, R., (1994) Why bank policies fluctuate: a theory and some evidence.
Ratnovski, L., (2013) Liquidity and Transparency in Bank Risk Management.
Rinaldi, L., Sanchis-Arellano, A., (2006) Household Debt Sustainability: What
Explains Household Non-performing Loans? An Empirical Analysis ECB Working Paper.
Roodman, D., (2006) How to do xtabond2: An introduction to “difference” and
“system” GMM in Stata Center for Global Development Working Paper No. 103.
Roodman, D., (2009) Practitioners' corner A note on the theme of too many instruments Oxford Bulletin of Economics and Statistics, 71(1), 135.158.
Salas, V., Saurina, J., (2002) Credit risk in two institutional regimes: Spanish commercial and savings banks Journal of Financial Services Research 22,
Shu, C., (2002) The impact of macroeconomic environment on the asset quality of
Hong Kong's banking sector R D Economic Research Division, Hong KongMonetary Authority.
In their 2011 study published in the Kuwait Chapter of the Arabian Journal of Business and Management Review, Solarin et al explore the factors influencing nonperforming loans within Malaysia's Islamic banking system using an Autoregressive Distributed Lag (ARDL) approach The research highlights the critical determinants that affect loan performance, providing valuable insights for improving financial stability in Islamic banking.
Stiglitz, J E and A Weiss (1981) Credit Rationing in Markets with Imperfect
Stiroh, K & Rumble, A., (2006) The dark side of diversification: The case of US financial holding companies Journal of Banking and Finance, 30, 2131-2161 Thiagarajan, S., (2013) Determinants of Credit Risk in the Commercial Banking
Sector of Belize Research Journal Of Social Science And Management, 03,
Thiagarajan, S., S Auuapan, et al (2011) Credit risk determinants of public and
Private sector banks in India European Journal of Economics, Finance and Administrative Science(34): 147-154.
Vogiazas, S D., (2015) Determinants of Credit Risk in the Bulgarian and the
The Romanian banking system has been significantly influenced by the Greek financial crisis, as explored in research conducted by the South East European Research Centre at Sheffield University Vogiazas and Nikolaidou (2011) analyze the determinants of credit risk within the Bulgarian banking sector, highlighting the interconnectedness of regional economies during the Greek twin crises This study underscores the importance of understanding external economic factors that impact banking stability in Southeast Europe.
Washington, G K., (2014) Effects of macroeconomic variables on credit risk in the
Kenyan banking system International Journal of Business and Commerce,
Windmeijer, F., (2005) A finite sample correction for the variance of linear efficient two-step GMM estimators Journal of Econometrics 126, 25–51.
Williamson, S (1987), Financial Intermediation, Business Failures, and Real Business
Cycles Journal of Political Economy, 95(6), 1196–1216.
Zribi, N., & Boujelbene, Y., (2011) The factors influencing bank credit risk: the case of Tunisia Journal of Accounting and Taxation, 3(4), 70-78.
Number of banks in each country: 162 banks in total
APPENDIX 2 xtabond2 model selection criteria
F-test - Reject the null hypothesis that independent variables are jointly equal to zero
Arellano-Bond test - First-order serial correlation (AR1 ≤ 0.05) but no second-order serial correlation (AR2 ≥ 0.1) in the residuals (Arellano & Bond, 1991)
Sargan Test - Sargan statistic is biased in one-step estimator with ‘Robust’ option (Roodman, 2006) Therefore, Sargan Test is not considered.
Hansen J-statistic - P value ≥ 0.25 (Roodman, 2009) and p ≤ 0.80 over-identifying restrictions does not reject the null at any conventional level of significance.
Difference-in-Hansen - P value of 1 is the sign of inappropriate model (Roodman,
2009) Steady state - The estimated coefficient on the lagged dependent variable should have a value less than (absolute) unity (Roodman, 2009) Number of instruments
- The number of instruments should not exceed the number of groups (i.e number of banks) (Roodman, 2009)
Roodman (2006, 2009) recommends reporting the optimal number of instruments in regression analysis The standard approach involves setting lag-limits, starting from lag2 for endogenous variables and lag1 for exogenous and predetermined variables, extending to the most available lag To adhere to Stata's size limit for instruments, the 'collapse' option is utilized Additionally, various regressions are estimated by modifying the upper and lower lag-limits, with the regression that meets all specified criteria being chosen as the optimal regression.
Source: Roodman, 2006; Roodman, 2009; Arellano & Bond, 1991.
NPL SVCR DEBT INEFF LOGSIZE NOINTINC ROA UNEMP RINT REER GDPG INFGDP
Additional analyses and Robustness checks
All models were estimated using a constant, with robust t-statistics provided in parentheses The significance levels for rejecting the null hypothesis are indicated as ***, 1%; **, 5%; and *, 10% The one-step SGMM estimator with difference lag was employed for each regression, and the number of observations (No Obs.) is reported Additionally, the Arellano–Bond tests for first (AR1) and second-order (AR2) autocorrelation in first-differenced errors are included, along with the statistics and p-values for the Sargan test of over-identifying restrictions and the Hansen test for instrument-residual uncorrelation in the AB estimations An appreciated real effective exchange rate signifies an appreciation of the domestic currency.
Additional analyses and Robustness checks