P ROBLEM S TATEMENT
Commercial banks play a crucial role as financial intermediaries in the economy by mobilizing and lending money to households, firms, and other entities A significant challenge they face is credit risk, which arises when borrowers misuse the funds they have borrowed This risk often increases when banks lower their credit standards to attract more borrowers, potentially leading to a rise in non-performing loans (NPLs) in the future However, if the expansion of loans is driven by increased demand, it may not necessarily result in a higher incidence of bad loans.
Therefore, credit growth may be a reflection of credit quality under some circumstances
During a recession, the financial market, particularly the banking system, experiences negative impacts Commercial banks strive to increase credit growth for profit, while households and businesses face challenges in their operations This situation can lead to a decline in credit quality, raising concerns about whether the profit-driven approach of banks is effective or if it merely results in higher non-performing loans (NPLs), ultimately causing losses.
Understanding the impact of credit growth on credit quality is increasingly crucial for commercial banks, central banks, and policymakers For commercial banks, this knowledge aids in determining the optimal timing for adjusting credit standards and making informed decisions about expanding or restricting lending activities For central banks, it is essential for managing overall loan growth effectively.
To mitigate the risk of a banking crisis due to low credit quality, it is essential for commercial banks to monitor credit growth closely By comprehending the impact of credit expansion on overall credit quality, policymakers can better assess the economic landscape and implement more effective macroeconomic strategies.
According to a 2015 report by The State Bank of Vietnam, Dong Nai branch, the total debit balance of loans stands at approximately 100 billion VND, primarily concentrated in key sectors of Dong Nai province The overdue loan ratio is maintained at a safe 2.32 percent; however, businesses continue to face challenges as commercial banks implement new debt classification standards Additionally, some banks have been found to violate credit regulations, including careless appraisal of credit documents, misjudging customer financial capabilities and collateral, and inadequate supervision of capital usage Consequently, commercial banks in Dong Nai may encounter significant potential credit risks as their lending activities expand.
R ESEARCH O BJECTIVE AND Q UESTION
This research examines how the growth of bank credit affects credit quality while considering various bank characteristics The primary objective is to address the question of this influence.
Does a positive change in the commercial banks’ credit growth lead to a negative change in banks’ credit quality in the case of Dong Nai banking system?
R ESEARCH S COPE AND M ETHODOLOGY
The research is carried out in the scope of credit growth and credit quality of
The study examines 29 commercial banks in Dong Nai province, Vietnam, utilizing data collected from the third quarter of 2009 to the first quarter of 2014.
This study employs a quantitative analysis methodology, utilizing panel data to address the limitations of traditional estimation methods To effectively capture the dynamics of banks' credit growth and credit quality, the research applies the Difference GMM method for the dynamic panel data model.
Besides, the dynamic model could be used to generate long–run coefficients which reflect the equilibrium of the effects of credit growth.
T HESIS S TRUCTURE
This thesis comprises five chapters: Chapter 1 outlines the background and motivation behind the research on the impact of credit growth on credit quality Chapter 2 reviews relevant theories and previous empirical studies, establishing a conceptual framework for the research Chapter 3 details the data collection methods and quantitative techniques employed to derive the necessary results Chapter 4 presents the findings and discussions, while Chapter 5 summarizes the key research outcomes, discusses policy implications, acknowledges limitations, and suggests directions for future research.
This chapter explores two primary theories regarding the impact of credit growth on credit quality It begins by examining the macroeconomic context of banking to elucidate how these effects may manifest Additionally, it discusses the "three shifts" that detail the channels through which credit growth influences credit quality in the market The chapter also addresses various bank characteristics that serve as control factors affecting credit quality, alongside a review of relevant empirical studies and the conceptual framework guiding this research.
T HE M ACROECONOMIC C ONTEXT FOR B ANKING
Main Characteristics of Banks
Banks play a crucial role as financial intermediaries, significantly impacting the economy through their involvement in finance and monetary activities As a distinct industry, banks possess unique characteristics that set them apart from other sectors This research aims to analyze the banking system within the macroeconomic context by introducing three key characteristics of banks that are closely linked to credit growth and credit quality.
Banks operate with high leverage, primarily utilizing other people's money for their portfolios and mobilizing capital for lending activities According to Gavin and Hausmann (1996), this high leverage results in two key implications: first, banks are highly sensitive to macroeconomic volatility due to their thin capital, making them vulnerable to insolvency from minor economic downturns Second, the high leverage can create conflicts between the interests of bank shareholders and debt-holders.
5 managers often generate risky portfolio to bring the highest benefit for shareholders while debt–holders is limited in their capital recovery in case of insolvency
Banks can quickly face liquidity issues because the duration of their deposit liabilities often exceeds that of their loan assets Borrowers, such as businesses and households, typically require long-term financing, while depositors can withdraw their funds at any moment Even with time deposits, depositors may withdraw their money, albeit at the cost of receiving lower or no interest rates.
If banks try to manage loan terms, borrowers may still have extended repayment periods A temporary solution involves rolling over loans by taking new ones to pay off old debts within the same bank However, this practice can be restricted or even illegal, negatively impacting borrowers' profitability and leading to a decline in credit quality Consequently, banks must prepare for additional reserves to mitigate potential illiquidity caused by adverse macroeconomic conditions.
During the expansionary phase of the economy, banks struggle to accurately assess their borrowers' financial capabilities According to Gavin and Hausmann (1996), "good times are bad times for learning" about borrowers' true financial situations This economic prosperity often leads to lending booms, as borrowers can easily secure loans from banks to manage existing debts Consequently, many borrowers present themselves as financially stable, despite varying financial capacities As a result, banks face challenges in identifying which loans may eventually turn into non-performing loans (NPLs).
Boosting credit growth must be approached with caution, particularly in challenging macroeconomic conditions Banks face a significant risk of declining credit quality, which can lead to illiquidity and potential banking crises.
Shock and Vulnerability of Banking System
For banks to effectively lend mobilized money, it is essential that the growth rate of deposit liabilities exceeds the deposit interest rate If this condition is not met, banks can still address the issue by requesting borrowers to repay their outstanding debts.
Banks face limitations in their operations due to their reliance on borrowers' capabilities This reliance leads to a net transfer of resources from the banking system to depositors through withdrawals and interest payments A significant volume of this transfer can induce shocks within the banking system, and if the transfer is substantial enough, it may result in a collapse of the banking system, highlighting its inherent vulnerabilities (Gavin and Hausmann, 1996).
The Effect of Macroeconomic Determinants
A significant shift in net resource transfers can stem from economic changes When negative economic surprises arise from macroeconomic factors, two scenarios may occur: first, borrowers may struggle to meet their debt obligations due to decreased efficiency in their business operations.
Banks face limitations in their investment activities, particularly in lending, which can lead to illiquidity due to decreased deposit demand or increased withdrawals These situations can trigger a banking crisis characterized by insolvency A decline in credit quality may prevent banks from recovering sufficient principal and interest to meet their deposit liabilities, while increased withdrawal demands can strain their liquidity To mitigate these risks and enhance financial stability, banks should focus on improving their mobilization activities, thereby ensuring they have adequate liquidity to meet depositor withdrawals and more effectively manage non-performing loans (NPLs).
Credit Growth and Vulnerability of Banking System
Macroeconomic determinants indirectly impact the banking system primarily through the business environment and depositor behavior The fragility of the banking system in response to negative economic changes raises a critical question Borrowers' business activities are significantly affected by these economic shifts, and banks with strong relationships with their borrowers are equally influenced A rapid growth in credit serves as a clear indicator of this close relationship; as banks increase their lending, they become more dependent on their borrowers.
Boosting credit growth is closely related to the third characteristic of banks
Recognizing the positive attributes of borrowers encourages lenders to extend more credit, establishing a connection between credit growth and the banking system's vulnerability However, credit growth should be viewed as an indicator of economic development rather than a direct cause of risk The critical inquiry is under what conditions credit growth may exhibit negative consequences, which primarily relates to issues surrounding information asymmetry.
Gavin and Hausmann (1996) highlighted the challenges bankers face in assessing the creditworthiness of borrowers, noting that economic expansion can lead to borrowers performing well and generating positive cash flow This situation presents banks with opportunities to extend loans to both existing and new customers However, the limited information available about new borrowers increases the risk of misjudgment, potentially leading to a decline in credit quality in the future.
During economic expansion, the abundance of loan supply enables borrowers to access a wider range of lenders Banks provide these loans, which borrowers often use to settle debts with other financial institutions This practice can unintentionally and negatively affect the information of other banks.
8 information externality in the credit market This type of externality also leads to credit misevaluation and potentially low credit quality
Figure 2.1 illustrates the macroeconomic context for banking, focusing on the dynamics of lending and debt repayment The observed negative correlation between credit growth and credit quality may indicate two scenarios: first, adverse macroeconomic shocks can lead to inefficiencies in business operations, hindering debt repayment capabilities; second, positive signals during economic expansion may generate information externalities that assist banks in assessing their customers.
F IGURE 2.1 The macroeconomic context for banking
Macroeconomic effects Paying debts tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
C REDIT G ROWTH AND C REDIT Q UALITY THROUGH D IFFERENT S HIFTS
Credit growth and credit quality may not be directly related, as the increase in lending does not necessarily impact the quality of loans However, banks' lending decisions are significantly influenced by their own performance and that of their borrowers When banks underestimate borrower risk, they may increase lending, suggesting a potential relationship between credit growth and credit quality.
Clair (1992) highlights the theoretical connection between credit growth and credit quality, suggesting that banks may reduce their credit standards to attract more borrowers, which can result in diminished credit quality over time Additionally, if banks increase credit growth without implementing effective strategies to manage borrowers' loan usage, it can further lead to a decline in credit quality.
Clair (1992) suggested that during economic recovery or expansion, credit growth can have a positive relationship with credit quality This correlation may arise from structural changes in financial markets, such as lowering barriers between banks and borrowers, which can enhance credit growth and mitigate credit risk through diversification.
Keeton (1999) developed a theory to explore the impact of credit growth on credit quality, highlighting both the negative and positive relationships between these two factors.
This relationship is primarily driven by a shift in the loan market supply, where banks adjust their willingness to lend They can achieve this in two ways: by lowering the interest rates on new loans or by relaxing credit standards To facilitate easier lending, banks may overvalue collateral, approve loans for customers with lower financial capacity, or adopt more lenient lending practices.
Many projects fail to have their cash-flow statements thoroughly evaluated, which can compromise credit standards and increase the risk for banks This negligence often leads to lending to borrowers with poor creditworthiness, resulting in a portfolio of low-quality loans.
When banks lower lending rates and relax credit standards, it can lead to a supply shift that boosts credit growth, but this often results in a decline in credit quality.
F IGURE 2.2 Supply shift r e Expected rate of return from loans z Measure of credit standards
S Supply of loans from banks
D Demand of loans from the borrowers
Figure 2.2 illustrates the impact of supply shifts on total lending and credit standards On the left side, the expected return rate for banks is influenced by credit standards, represented by the variable \( z \) on the horizontal axis A higher \( z \) indicates that borrowers are in a strong position to service their debts, such as having substantial collateral or engaging in secure investment projects, which ultimately informs banks' lending decisions.
D r e ሺzሻ D z2 z1 r 1 e r 2 e r 1 e r 2 e tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
The expected rate of return from loans, represented on the vertical axis, is influenced by both the lending rate and the debt servicing capacity of borrowers Banks anticipate higher returns from good borrowers, potentially matching the lending rate Conversely, if there are indications of poor borrowing behavior, the expected rate of return for banks may fall below the lending rate.
Banks can determine a maximum expected rate of return based on credit standards, represented by the curve \( r_e(z) \) As the lending rate increases, the expected rate of return also rises; however, this increase has its limits Good borrowers can still manage to repay their debts despite higher rates, but if rates continue to climb, they may resort to riskier investments in hopes of higher returns This shift can lead to a decline in the quality of borrowers, preventing further increases in the banks' expected rate of return Consequently, the upward-sloping curve \( r_e(z) \) illustrates that banks anticipate greater returns from borrowers with better credit standards by offering them higher lending rates.
The curve r e ሺzሻ could also be analyzed from the side of expected rate return
In the loan market, banks establish a minimum credit standard level for borrowers based on the expected equilibrium interest rate, denoted as \$r_1^e\$ Borrowers with credit scores below a certain threshold, \$z_1\$, are deemed ineligible for loans, as the anticipated return would fall short of \$r_1^e\$, regardless of the lending rate Conversely, borrowers at or above \$z_1\$ can secure loans, which are priced to ensure banks achieve the desired return of \$r_1^e\$ Consequently, the minimum credit standard serves as a critical benchmark for banks' lending decisions, with higher expected returns prompting stricter credit requirements for potential borrowers.
The loan market, illustrated in Figure 2.2, plays a crucial role in determining banks' expected rates of return As banks anticipate higher returns from loans, their willingness to lend increases, resulting in an upward-sloping supply curve Conversely, the demand curve slopes downward for two main reasons: first, while higher lending rates can yield greater returns for banks, they also impose a higher cost of capital on borrowers, leading to reduced borrowing Second, as banks' expected rates of return rise, the credit standards become stricter, decreasing the number of borrowers who qualify.
The loan market reaches equilibrium when the supply of bank loans matches the demand for loans Initially, the supply curve is represented as S1S1 At this equilibrium point, banks anticipate a return rate of r1e, with the total loan amount being L1.
To expand their total loan offerings, banks must lower their credit standards to attract more borrowers, resulting in a rightward shift of the supply curve from S1S1 to S2S2 This shift increases total lending from L1 to L2 and decreases the expected rate of return from r1e to r2e, indicating that banks are willing to accept lower returns Consequently, banks not only offer lower lending rates to creditworthy borrowers but also reduce their credit standards to reach a broader audience This reduction in credit standards is illustrated by a downward movement along the curve r e (z) However, as credit standards decline, banks may face an influx of borrowers with lower debt servicing capacities, leading to an increase in non-performing loans (NPLs) and diminished credit quality.
Hypothesis A suggests that there may be a negative relationship between credit growth and credit quality.
C ONTROL F ACTORS FOR C REDIT Q UALITY
The hypothesis by Berger and DeYoung (1997) suggests that cost efficiency impacts credit quality Low cost efficiency may indicate poor management, leading to a weak credit appraisal process where banks might extend loans to high-risk borrowers or invest in projects with inflated cash flow Additionally, inaccurate valuation skills can result in underestimating collateral values, hindering the recovery of low-quality loans This issue may stem from a compromise between the banks' asset pricing department and borrowers, further complicating customer supervision.
Productivity shift Demand shift tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
Borrowers may misuse loans for inappropriate purposes, such as investing in high-risk, high-return projects rather than those evaluated by banks This misuse, along with other factors, can lead to a decline in credit quality.
Berger and DeYoung (1997) demonstrated a negative correlation between cost efficiency and credit quality in banks To prevent future non-performing loans (NPLs), banks may incur higher short-term operating costs for monitoring existing loans Consequently, minimizing these monitoring costs can result in diminished credit quality Thus, banks may experience lower cost efficiency if they invest in measures that help avert greater losses from NPLs in the long run.
However, this trade–off reflects a decrease in measured cost efficiency and an increase in credit quality afterward
The "Bad Management I" and "Skimping" hypotheses both pertain to measured cost efficiency, yet it is crucial to differentiate between them The key distinction lies in their cost implications: "Bad Management I" results in waste due to poor management practices, while "Skimping" serves as a strategic measure to prevent future losses.
Ineffective management in banks, characterized by deficiencies in credit appraisal, valuation skills, and customer oversight, often leads to poor performance, evident through low profitability on equity or assets Conversely, banks that excel in managing lending activities tend to demonstrate higher performance, as they are better equipped to minimize problem loans and enhance credit quality in the future (Louzis, Vouldis, and Metaxas, 2011).
“Pro–cyclical credit policy” hypothesis
Banks' credit policies aim to maximize profitability while maintaining a positive reputation Rajan (1994) notes that banks cannot effectively showcase their efficient loan portfolios or the performance of their customers; instead, the market only observes their earnings Consequently, banks may inflate earnings to uphold a favorable market perception, which is facilitated by a liberal credit policy This "pro-cyclical" approach aligns with demand conditions, leading banks to continue lending to high-risk borrowers despite previous loan issues While this strategy may temporarily enhance performance, it ultimately risks deteriorating credit quality and jeopardizing future stability.
According to Louzis et al (2011), banks can improve their credit quality through diversification opportunities, such as investing in promising business projects or high-potential stocks This approach can reduce the proportion of capital allocated to lending, thereby decreasing the likelihood of extending credit to high-risk borrowers However, while diversifying into non-lending investments may yield high returns, it often requires a long recovery period for both principal and interest Additionally, liquidity risk arises when customers withdraw funds, as banks primarily rely on liabilities for their capital structure Larger banks typically enjoy more diversification opportunities with minimal impact on liquidity, whereas smaller banks face greater challenges in diversifying due to heightened liquidity risks.
“Too–big–to–fail” hypothesis
Large banks, as noted by Stern and Feldman (2004), rely on government intervention during times of failure due to their significant impact on the financial market Consequently, these institutions are inclined to take on greater leverage and expand lending activities This credit expansion raises the likelihood of engaging with lower-quality borrowers, as highlighted by Louzis et al (2011).
Under this hypothesis, increased leverage negatively impacts credit quality, with larger banks experiencing a more significant decline than smaller ones This effect mirrors the lending aspect, where banks enhance credit growth.
Nguyen (2015) posited that the "too-big-to-fail" hypothesis is applicable primarily to the largest banks, suggesting that only these institutions are likely to receive government support during times of distress Consequently, the term "largest banks" serves as a viable alternative for bank size adjustment within the context of this hypothesis.
H YPOTHESIS C: The effect of credit growth on credit quality from large commercial banks might be larger than the smaller banks.
P REVIOUS E MPIRICAL S TUDIES
Clair (1992) is one of the earliest authors who investigated the relationship between credit growth and credit quality of banks in Texas using annual data from
From 1980 to 1990, the author assessed credit quality using the loan loss ratio and non-performing loan (NPL) ratio The independent variables were categorized into three groups: (i) credit growth, (ii) financial characteristics (including bank assets, bank equity, business loans, and real estate loans), and (iii) business conditions, specifically non-agricultural employment growth The model incorporated three types of credit growth: internal growth, growth through bank mergers, and growth through bank acquisitions.
The study examined the impact of credit growth using contemporaneous and lagged variables over one to three years, employing the ordinary least squares (OLS) method The findings revealed that both internal credit growth and growth through bank acquisitions significantly enhanced credit quality, as evidenced by improvements in the loan loss ratio and non-performing loan (NPL) ratio at contemporaneous and one-year lag periods Conversely, credit growth via bank mergers negatively affected the loan loss ratio, while showing no significant impact on the NPL ratio Additionally, various financial characteristics and business conditions were identified as effective control variables.
A study by Kunt and Detragiache (1998) identified a high ratio of non-performing assets, exceeding 10 percent, as a significant factor contributing to banking crises Analyzing annual data from market economies between 1980 and 1994 using multivariate logistic regression, the study categorized independent variables into three groups: macroeconomic (including GDP growth and inflation), financial (such as the ratio of money supply M2 to foreign exchange reserves), and institutional (like GDP per capita and law enforcement quality) Notably, the financial variable of credit growth, assessed with a two-year lag, showed a significant positive impact on the likelihood of banking crises, indicating that a 10 percent increase in credit growth could raise the crisis probability by 5.4 percent.
A study by Keeton (1999) examined the impact of credit growth on credit quality using quarterly time-series data from two distinct periods: 1967–1983 and 1990–1998, with a gap in data from 1984 to 1989 The data was sourced from the Senior Loan Officer (SLO) survey conducted by the Federal Reserve since 1967 The analysis employed vector auto-regression (VAR) methods, focusing on two VAR systems that included variables such as loan growth, credit standards, and the GDP gap.
The analysis reveals two key findings from the VAR systems regarding the relationship between loans, credit standards, and delinquency rates Firstly, while loan growth did not affect credit standards between 1990 and 1998, it significantly tightened credit standards from 1967 to 1983 Secondly, credit standards were found to reduce credit growth in both periods analyzed Additionally, the second VAR system indicates that increased loan growth in the past may result in a higher delinquency rate.
Salas and Saurina (2002) analyzed the macroeconomic and bank-specific factors influencing problem loans in Spanish commercial and saving banks from 1985 to 1997 Key bank-specific determinants include credit growth, measured by individual bank loan growth and branch network loan growth, along with inefficiency, collateralized loan rates, bank size, net interest margin, solvency ratio, and market share Macroeconomic variables examined are GDP growth rate, levels of indebtedness among families and banks, and a dummy variable for Spanish bank regulation in 1988 Utilizing a dynamic panel data estimator with the Difference GMM technique, the study considers lagged effects over 2, 3, and 4 years The findings indicate that credit growth positively impacts problem loans, with branch network loan growth increasing problem loan rates for commercial banks after 3 years, while saving banks experience rises in problem loans from bank loan growth and branch network growth after 3 and 4 years, respectively The study strongly confirms the significance of macroeconomic determinants and highlights the importance of certain bank-specific variables.
Jiménez, Lopez, and Saurina (2007) analyzed datasets from the Spanish banking system between 1988 and 2003 to investigate the link between competition and bank risk-taking, measured by the non-performing loan (NPL) ratio from borrowers Their model incorporates three categories of variables: a competition index based on the total value of bank loans, macroeconomic controls, and additional relevant factors.
This study examines the relationship between real GDP growth rate and bank-specific variables, such as return on equity, bank size, and loan ratio, using the difference GMM regression method The findings reveal limited evidence supporting a link between competition and bank risk-taking, with the exception of the Lerner index, which demonstrates a significant inverted U-shaped effect on the non-performing loan (NPL) ratio Specifically, as Spanish banks adopt more competitive loan policies, indicated by a lower Lerner index, they initially experience an increase in NPLs, followed by a subsequent reduction in risk Additionally, the analysis shows that a higher loan ratio is strongly and negatively correlated with the NPL ratio, suggesting that banks with a high level of lending specialization encounter lower risks from bad borrowers.
Foos, Norden, and Weber (2010) analyzed Bankscope annual data from over 16,000 banks across 16 major countries, including the United States, Canada, Japan, and 13 European nations, from 1997 to 2007 to assess the impact of loan growth on bank risk The study utilized two key dependent variables to measure credit quality: the loan losses provision ratio and the loan losses to net interest income ratio The primary independent variables included abnormal loan growth (with four lags), the equity to total assets ratio, and total customer loans The first model, which focused on the loan losses provision ratio, employed a dynamic panel data approach, utilizing both OLS and system GMM for estimation The findings indicated that, after controlling for the specified variables, abnormal loan growth significantly and positively influenced loan losses provision after 2 and 3 years.
The study employs a fixed effects model to analyze macroeconomic conditions and calculates loan growth as the average growth rate over previous years Findings indicate that past average loan growth is positively and significantly related to the ratio of loan losses to net interest income Additionally, the research explores the interaction effects of bank mergers and acquisitions on the relationship between loan growth and loan losses, confirming that the influence of loan growth remains positive and significant.
Bank mergers and acquisitions lead to a reduction in the impact of loan growth, resulting in all coefficients of lagged loan growth becoming negative This indicates that the loan growth of banks involved in mergers and acquisitions is inversely related to loan losses.
Valverde, Ibanez, and Fernandez (2011) conducted a study on bank lending and credit quality during the 2008 financial crisis, utilizing a sample of Spanish banks from the first quarter of 2000 to the first quarter of 2010 Their research employed three simultaneous GMM models, with loan growth, non-performing loan (NPL) ratio, and bank rating as the dependent variables.
Bank conditions, including metrics like equity over total assets, cost over income ratio, and deposit over total liabilities, along with market fundamentals such as GDP growth, real housing price growth, and the EURIBOR rate, serve as important control variables The findings indicate that the growth of bank lending significantly and positively impacts the non-performing loan (NPL) ratio with a lag of two years Additionally, certain control variables demonstrate significance, particularly within the bank rating model.
Caporale, Colli, and Lopez (2014) found that the growth of bank credit in Italy during expansionary periods leads to an increase in non-performing loans (NPLs) during contractionary phases Their study utilized a structural vector auto-regression model (SVAR) analyzing 17 monthly time-series data from June 1998 to June 2012, which included various loan types, NPLs, and macroeconomic indicators such as the unemployment rate, consumer price index, housing price index, and EURIBOR rate.
T ABLE 2.2 Summarization of the literature about the effect of credit growth on credit quality
Year Author Sample Period Methodology Key findings
1992 R T Clair Annual data of banks in Texas 1980 – 1990 Ordinary least squared
Loan growth improves loan quality at the lag of 0 and 1 year
Annual data of banks in all market economies
Credit growth causes the decline in credit quality after 2 years
1999 W R Keeton Quarterly time–series surveyed data from Senior Loan Officer (SLO)
Loan growth tightens credit standards in and leads to high delinquency rate
2002 V Salas and J Saurina Annual data of commercial and saving banks in Spain
1985 – 1997 Difference GMM Loan growth increases the rate of problem loans after 3 and 4 years
Annual data of banks in Spain 1988 – 2003 Difference GMM High loan ratio leads to a contemporaneous increase in NPLs
Bankscope annual data of more than 16,000 banks in 16 major countries
1997 – 2007 OLS and System GMM Loan growth of banks engaged in mergers and acquisitions negatively relate to loan losses
Quarterly data of banks in Spain
2000 – 2010 Simultaneous GMM Loans growth positively related to
NPL ratio after two years
Monthly data of banks in Italy 1998 – 2012 Structural VAR Credit growth causes an increase in
NPLs tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
T HE C ONCEPTUAL F RAMEWORK
Based on the literature review of relevant theories and empirical analyses, this research proposes a conceptual framework illustrating the impact of credit growth on credit quality, as depicted in Figure 2.6 The primary focus of the study is on credit growth, while explanatory variables will be utilized with lags to account for past effects.
F IGURE 2.6 The effect of credit growth on credit quality
Supply shift (–) Demand shift and Productivity shift (+) Too–big–to–fail (–)
Bad management I (+) or Skimping (–) Bad management II (+) or Pro–cyclical credit policy (–) Diversification (+)
Cost efficiency is crucial for enhancing profitability in banking The size of the bank plays a significant role in its ability to leverage resources effectively By optimizing these factors, banks can improve their financial performance and ensure sustainable growth.
This chapter introduces credit quality measurement, data and methodology
This research measures credit quality using the loan classification system of The State Bank of Vietnam The data is sourced from detailed balance sheets and income statements available in the database of The State Bank of Vietnam, Dong Nai branch The methodology employs a dynamic panel data estimator, utilizing the difference GMM technique to calculate long-run coefficients and establish the econometric model specification.
M EASURING C REDIT Q UALITY
Various factors can be utilized to assess credit quality, with many prior studies employing the non-performing loan (NPL) ratio as an indicator of poor credit quality Additionally, Keeton (1999) utilized loan losses as a measure of credit quality.
The definition of Non-Performing Loans (NPL) varies by country, but it generally focuses on overdue loans, typically those that are 90 days or more past due In Vietnam, this research adopts the definition outlined in Circular No 02/2013/TT–NHNN, issued by the State Bank of Vietnam on January 21, 2013 According to this definition, bank loans are categorized into five distinct groups.
Current debts that can be fully and promptly recovered, including both principal and interest, are those that are overdue by less than 10 days, provided the borrower is capable of repaying the total amount on time.
G ROUP 2: Debts which need attention
Debts can become overdue within a timeframe ranging from 10 to 90 days.
Debts are overdue for a period of between 91 days and 180 days
Debts are overdue for a period of between 181 days and 360 days
Debts are overdue for a period of more than 360 days
Credit quality could be measured from the ratios calculated from five groups above There would be three typical ratios as follows:
The non-performing loan ratio is determined by dividing the total amount of loans classified in groups 3, 4, and 5 by the total amount of loans A higher non-performing loan ratio indicates a decline in credit quality.
The loan loss ratio, determined by dividing the total amount of loans classified in group 5 by the total amount of loans, serves as a key indicator of credit quality; a higher ratio signifies lower credit quality.
The standard loan ratio is determined by dividing the total amount of loans classified in group 1 by the total amount of loans A higher standard loan ratio indicates better credit quality.
Different measurements play distinct roles in lending activities: standard loans are characterized by high quality, while non-performing loans (NPLs) face challenges in recovering both principal and interest, prompting banks to seek timely solutions Loan losses, which are likely to reflect low credit quality, can significantly impact banks' expenses Notably, many banks report zero loan losses over extended periods, indicating that the loan loss ratio may not be a suitable variable in the model For banks with similar NPL ratios, this discrepancy becomes even more pronounced.
A higher standard loan ratio indicates improved credit quality Consequently, this research focuses on the ratio of non-performing loans (NPLs) to standard loans, rather than total loans, to more accurately assess credit quality This approach serves as an inverse measure of credit quality.
D ATA C OLLECTION M ETHOD
Data for this study were sourced from the State Bank of Vietnam, encompassing detailed balance sheets and income statements of 29 commercial banks from the third quarter of 2009 to the first quarter of 2014 The data collection method adhered to the System of Bookkeeping Accounts for Credit Institutions, as outlined in Decision No 479/2004/QĐ–NHNN, issued on April 29, 2004 This system categorizes various account types, formatted in the XXXX digit structure, with Table 3.1 illustrating the necessary items corresponding to these account types.
T ABLE 3.1 Necessary items and their account type
Customer loans in five groups From 2XX1 to 2XX5
Total assets From 1XXX to 3XXX and 5XXX
Liabilities From 40XX to 49XX
Table 3.2 presents the variables associated with their hypotheses and the anticipated signs for assessing credit quality, reflecting the inverse relationship between credit quality and cost efficiency These expected signs differ from those shown in Figure 2.6 For a comprehensive understanding, the specific formulas for these variables can be found in Table 3.3.
Currently, Dong Nai province is home to 53 commercial banks However, this research focuses on 29 bank branches due to the availability of collected data and the recent opening of new branches.
T ABLE 3.2 The expected signs of variables used in the research
CREDIT with LARGEST Too–big–to–fail +
LEVERAGE Too–big–to–fail +
T ABLE 3.3 The calculation of variables
Total loans in group 3, 4 and 5 Total loans in group 1
Credit growth CREDIT Total loans t − Total loans t−1
Profitability ROA Total incomes − Total expenses
Bank size SIZE Total assets
Total assets of all banks
The three largest banks by total assets have been identified, highlighting their significant financial standing For the latest updates and comprehensive information, please refer to the provided resources.
E CONOMETRIC M ETHODOLOGY
Dynamic Panel Data Estimator
Research by Salas and Saurina (2002) and Louzis et al (2011) indicates that dynamic panel data, which incorporates the lagged effects of explanatory variables, can be effectively structured.
In the equation presented, 𝐘 𝐢𝐭 and 𝐘 𝐢𝐭−𝟏 represent the credit quality of bank i at time t and its previous period, respectively The term 𝛃(𝐋) signifies a 1 × k lag polynomial vector, capturing the lagged effects of explanatory variables, while 𝐗 𝐢𝐭 consists of k × 1 vectors representing other explanatory variables excluding 𝐘 𝐢𝐭−𝟏 The unobserved individual characteristics of banks are denoted by 𝛈 𝐢, with 𝛂 𝟎 as the intercept and 𝛆 𝐢𝐭 as the error term Given the nature of accounting data and the non-contemporaneous nature of banks' decisions and their impacts, the current level of explanatory variables is excluded from the model.
Econometric Problems
The common methods for estimating dynamic panel data models often face significant endogeneity issues due to the correlation between independent variables and the error term, denoted as 𝛆 𝐢𝐭 Additionally, unobserved cross-sectional characteristics, or fixed effects, are included in the error term To address these challenges, Arellano and Bond (1991) proposed transforming the data into first differences, which helps eliminate unobserved factors and fixed effects This approach, further generalized by Arellano and Bover (1995) and Blundell and Bond (2000), utilizes the Difference GMM method, resulting in more efficient estimations of the model.
In the context of the equation (2), the first difference operator, denoted as 𝚫, indicates that 𝚫𝐘 𝐢𝐭−𝟏 may be correlated with the error term 𝚫𝛆 𝐢𝐭, potentially resulting in biased estimations Louzis et al (2011) suggest that using the two-period lag of the dependent variable, 𝐘 𝐢𝐭−𝟐, which is expected to correlate with 𝚫𝐘 𝐢𝐭−𝟏 but not with the error term 𝚫𝛆 𝐢𝐭, serves as an appropriate instrument in the model.
To obtain an efficient estimator, it is essential that the explanatory variables are exogenous However, endogeneity can arise due to the correlation between the explanatory variables, denoted as \$\Delta X_{it}\$, and the error term, represented as \$\Delta \epsilon_{it}\$ In such instances, the inclusion of lagged variables becomes necessary.
𝐗 𝐢𝐭 (in level, not taking first difference) could become valid instruments (Louzis et al., 2011)
The Difference GMM method includes two techniques: one-step and two-step GMM The one-step GMM provides consistent parameters under the assumption of independent and homoscedastic residuals, while the two-step GMM does not adhere to this assumption However, the two-step GMM can result in downward biased standard errors Research by Arellano and Bond (1991) and Blundell and Bond (2000) indicates that the efficiency gained from two-step GMM is minimal, even in the presence of heteroscedastic errors Consequently, one-step GMM estimation may be the preferable option.
The estimated model must undergo testing for serial autocorrelation and the validity of instrument variables For serial autocorrelation, the equations \$\Delta \epsilon_{it} = \epsilon_{it} - \epsilon_{it-1}\$ and \$\Delta \epsilon_{it-1} = \epsilon_{it-1} - \epsilon_{it-2}\$ indicate that both share the same \$\epsilon_{it-1}\$ Consequently, the presence of first-order autocorrelation, AR(1), is anticipated, necessitating further testing for second-order autocorrelation.
AR(2) is more important as it considers the correlation between 𝛆 𝐢𝐭−𝟏 (in 𝚫𝛆 𝐢𝐭 ) and
The null hypothesis for both the AR(1) and AR(2) tests posits the absence of autocorrelation If the AR(2) test rejects this null hypothesis, it indicates that the validity of the instrumental variables is compromised Therefore, a higher p-value in the AR(2) test suggests a stronger case for the absence of autocorrelation.
33 better For the validity of instrument variables, Sargan Test and Hansen Test are used with the null hypothesis of valid instrument variables (Roodman, 2006).
Estimating The Long–run Coefficients
In dynamic panel data models, it is crucial to account for the long-run coefficients of explanatory variables, as they indicate the cumulative impact on the dependent variable Louzis et al (2011) highlighted the potential for multicollinearity among the lags of these variables, which can compromise the precision of individual coefficients Consequently, utilizing long-run standard errors for these coefficients may provide a more reliable estimation.
Previous studies indicate that credit growth typically affects credit quality after a one-year period Therefore, this research utilizes four lags for the explanatory variables, given the quarterly data The model will be presented as follows:
On the supposition that all 𝚫𝐗 𝐢𝐭−𝐣 (j ranges from 1 to 4) increases in one unit, the total short–run effect on 𝚫𝐘 from 𝚫𝐗 at present (period t) would be:
In the period t + 1, because of the term 𝛂𝚫𝐘 𝐢𝐭−𝟏 , the total effect would be:
In the period t + 2, the overall impact can be summarized as follows: the total effect is significant, and it is essential to stay updated with the latest information and resources available.
The long–run effect would capture all short–run effects Therefore, it is equal to the sum of (3), (4) and (5) and more The formula can be written as:
The absolute coefficient of 𝛂 is typically less than 1, leading to the convergence of 𝛃 𝐋𝐑 This value represents the sum of an infinite geometric progression with a common ratio of 𝛂 The formula for 𝛃 𝐋𝐑 can be simplified from equation (6).
To sum up in the formula (7), 𝛃 𝐣 is the coefficient of one explanatory variable of lag 𝐣 𝛂 is the coefficient of 𝚫𝐘 𝐢𝐭−𝟏
According to Stuart and Ord (1998, p 351), the variance of 𝛃 𝐋𝐑 could be calculated as follow:
𝐣=𝟏 tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
Econometric Specification
According to the literature, the dynamic panel data model about the effect of credit growth on credit quality would be generally formed as:
In equation (9), 𝐍𝐏𝐋_𝐒𝐋 indicates credit quality, with the variable 𝐣 representing lags from 1 to 4 The coefficients are denoted by 𝛂, 𝛄, 𝛅, and 𝛃 The control factors 𝐗 include cost efficiency (𝐂𝐄𝐅𝐅), profitability (𝐑𝐎𝐀), bank size (𝐒𝐈𝐙𝐄), and leverage (𝐋𝐄𝐕), with leverage interacting with a dummy variable for the three largest banks (𝐋𝐀𝐑𝐆𝐄𝐒𝐓) Due to the stability of bank size, using its lags as explanatory variables may result in high multicollinearity; therefore, 𝐒𝐈𝐙𝐄 will be utilized at its current level.
By using the “restricted” GMM procedure suggested by Judson and Owen
(1999), control variables are added into the model (9) in turn The baseline model (include only credit growth) and the unrestricted model (include all 4 control variables) are also estimated.
Hypothesis testing
The impact of credit growth on credit quality is determined by four individual lagged coefficients (𝛄 𝐣), with their significance assessed through the p-value in the estimation results.
The "too-big-to-fail" hypothesis plays a significant role in the marginal effect of credit growth on credit quality The coefficient of credit growth, denoted as 𝐂𝐑𝐄𝐃𝐈𝐓, along with its interactive term 𝐂𝐑𝐄𝐃𝐈𝐓 × 𝐋𝐀𝐑𝐆𝐄𝐒𝐓, provides crucial insights into this relationship.
If 𝐋𝐀𝐑𝐆𝐄𝐒𝐓 = 1, the coefficient of 𝐂𝐑𝐄𝐃𝐈𝐓 would be (𝛄 𝐣 + 𝛅 𝐣 ) and the variance is calculated as 𝐕𝐚𝐫(𝛄 𝐣 + 𝛅 𝐣 ) = 𝐕𝐚𝐫(𝛄 𝐣 ) + 𝐕𝐚𝐫(𝛅 𝐣 ) + 𝟐𝐂𝐨𝐯ሺ𝛄 𝐣 , 𝛅 𝐣 ሻ
The long–run coefficient of 𝐂𝐑𝐄𝐃𝐈𝐓 corresponding to 𝛄 𝐣 and 𝛅 𝐣 is calculated as formula (7), the variance is in formula (8) In the case of long–run coefficient of
(𝛄 𝐣 + 𝛅 𝐣 ), the formula would be calculated as:
The corresponding variance of this long–run marginal effect (11) would be:
𝐂ሺ𝟏 − 𝛂ሻ + 𝐕𝐚𝐫ሺ𝛂ሻ ሺ𝟏 − 𝛂ሻ 𝟐 ] , ሺ12ሻ tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
Table 3.4 presents four null hypotheses which need testing as follow:
H 0 1 : γ j = 0 Credit growth does not have any effect on credit quality at lag j
H 0 2 : γ j + δ j = 0 Credit growth does not have any effect on credit quality at lag j for the three largest banks
H 0 3 : γ LR = 0 Credit growth does not have any long–run effect on credit quality
The hypothesis \( H_0: \gamma_{LR} + \delta_{LR} = 0 \) suggests that credit growth does not significantly impact the long-term credit quality of the three largest banks.
This chapter presents the estimation results and discusses the main findings
Credit growth negatively affects credit quality at lag levels of 3 and 4, with the long-run coefficient of credit growth also being statistically significant This indicates the presence of supply shifts and information externalities in the credit market, along with adverse impacts stemming from macroeconomic conditions.
Figure 4.1 compares the growth rate of deposit liabilities in the Dong Nai banking system, focusing on commercial banks from 2010 Q1 to 2014 Q4, with the quarterly deposit interest rate in Vietnam The data indicates that when the deposit growth rate exceeds the deposit interest rate, it suggests a stable banking environment with no significant shocks or vulnerabilities Consequently, credit growth was not a major concern for commercial banks during this period However, the figure also highlights that if credit growth falters, it could pressure banks to enhance their mobilization efforts, potentially disrupting the balance between deposit growth rates and interest rates, and reversing the flow of net resources.
F IGURE 4.1 Deposit growth rate and deposit interest rate
Source: The State Bank of Vietnam, Dong Nai branch
International Monetary Fund eLibrary Data
Table 4.1 below presents a summary of descriptive statistics for the sample used in this empirical study
Variable Observation Mean Standard Deviation Minimum Maximum
NPL_SL 551 0.0348 0.0394 0.0000 0.1977 CREDIT 551 0.0414 0.1570 –0.6192 0.5467 COST_EFF 551 0.9050 0.1627 0.4036 1.4639
The article discusses the deposit growth rate and the impact of the deposit interest rate on financial stability It highlights the importance of understanding these rates for effective financial planning Additionally, it mentions resources for downloading the latest research and academic papers related to these topics.
Table 4.2 displays the estimation results from the one-step Difference GMM method, utilizing robust standard errors The AR(2) test for second-order autocorrelation indicates that \$\epsilon_{it-1}\$ does not correlate with \$\epsilon_{it-2}\$, as evidenced by a high p-value Conversely, the AR(1) test for first-order correlation yields a small p-value (below 0.05), confirming the validity of the instrument variables Additionally, both the Sargan Test and Hansen Test show very high p-values, supporting the null hypothesis that the instrument variables are valid.
These tests indicate that the problem of endogeneity would be solved and the coefficients in the model could be used for further analysis
The estimation results indicate that the lagged independent variable (∆𝐍𝐏𝐋_𝐒𝐋 𝐢𝐭−𝟏) has a positive and significant coefficient, suggesting that current credit quality is influenced by its past performance This implies that higher or lower credit quality can lead to similar outcomes in the following quarter Additionally, the lagged variables of credit growth are also positive and significant at lag levels 3 and 4, indicating that the decision to expand lending activities by commercial banks in Dong Nai negatively impacts credit quality after three quarters to one year.
According to Circular No 02/2013/TT–NHNN, customer loans are classified as non-performing loans (NPLs) if they are overdue for 91 days or more, which is two quarters shorter than the significant lag levels of credit quality Consequently, new loans from commercial banks in Dong Nai face a high likelihood of becoming NPLs within two quarters after reaching the overdue threshold Additionally, the impact of credit growth on credit quality is not statistically significant within the shorter time frames of one and two quarters.
Dependent variable Baseline model Model (1) Model (2) Model (3) Model (4) Full model
–0.0254 (0.2058) tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
2.4969 (3.1992) tot nghiep down load thyj uyi pl aluan van full moi nhat z z vbhtj mk gmail.com Luan van retey thac si cdeg jg hg
Testing for autocorrelation and validity of instruments
Standard errors are presented in parentheses
indicate significance levels of 10%, 5%, and 1%, respectively For the latest full download of the thesis, please contact via email at vbhtj mk gmail.com.
Table 4.2 indicates that the coefficients of credit quality variables interacting with the dummy variable ∆(𝐂𝐑𝐄𝐃𝐈𝐓 𝐢𝐭−𝐣 × 𝐋𝐀𝐑𝐆𝐄𝐒𝐓 𝐢𝐭) are not statistically significant, suggesting that the impact of credit growth on the credit quality of the three largest commercial banks is similar to that of other banks Additionally, most bank characteristics used as control variables in this study do not significantly explain changes in credit quality.
The significance of ∆𝐋𝐄𝐕𝐄𝐑𝐀𝐆𝐄 𝐢𝐭−𝟏 at 10% is noted, although it does not meet expectations Additionally, while most coefficients are insignificant, with the exception of ∆𝐂𝐎𝐒𝐓_𝐄𝐅𝐅 𝐢𝐭−𝟐, the positive signs of cost efficiency and profitability across all lag levels suggest a potential presence of "Skimping" and "Pro-cyclical credit policy" hypotheses within the Dong Nai banking system.
In the long run, the positive and significant coefficient of ∆𝐂𝐑𝐄𝐃𝐈𝐓 suggests that the credit growth of commercial banks in Dong Nai may lead to a gradual decline in credit quality The overall impact of credit growth is significantly greater than the effects observed from individual lag levels Additionally, the long-run coefficient of ∆(𝐂𝐑𝐄𝐃𝐈𝐓 × 𝐋𝐀𝐑𝐆𝐄𝐒𝐓) is not significant, indicating that the effect of credit growth on credit quality remains consistent across both the largest banks and their smaller counterparts.
The estimation results indicate a negative relationship between credit growth and credit quality among commercial banks in Dong Nai, evident in both short-run and long-run analyses This suggests two key implications: first, commercial banks may be lowering their credit standards to enhance lending activities, increasing the likelihood of extending loans to borrowers with poor financial capacity Second, adverse macroeconomic conditions may negatively impact the business operations of economic entities, particularly households and small to medium enterprises.
45 enterprises They would hardly gain positive cash flows to finance their liabilities
Overdue loans are on the rise due to information externalities in the loan market, leading commercial banks in Dong Nai to potentially misjudge new customers Borrowers, including firms and households, often have access to multiple lending sources, allowing them to use funds from one loan to settle debts with another.