AN INVESTIGATION INTO BANK ASSET QUALITY

Get Complete Project Material File(s) Now! »

CHAPTER 3: THEORETICAL FOUNDATION AND HYPOTHESIS DEVELOPMENT: DETERMINANTS OF BANK ASSET QUALITY

INTRODUCTION

The preceding chapter explained the functioning of banks and how banking activities impact bank asset quality. However, the theories underpinning the relationships between NPLs and the various determinants were not addressed in the preceding chapter.
This chapter aims to provide the theoretical foundation for this study. The theoretical foundation enables the connection between the bank functions and the theories on how NPLs impact the range of banking activities. These linkages allow the development of hypotheses which are analysed in a later chapter.
This chapter consists of two sections, namely, the theoretical foundation and the hypothesis development. These sections, Sections 3.2 and 3.3, are divided into subsections. Section 3.2.1 presents the theoretical association, Section 3.2.2 the theoretical shortcomings, and Section 3.2.3 the research contribution, while Section 3.2.4 presents the research objectives. Section 3.3 provides the hypothesis development, and the subsequent subsections (Section 3.3.1 – 3.3.5) present the five research hypotheses, before the chapter is concluded with a chapter summary.

THEORETICAL FOUNDATION

The 19th and 20th centuries were plagued by banking and stock market crashes (Bordo & Haubrich, 2010). These crashes were preceded by changes in monetary policies or various shocks as a result of war, real estate crashes or boom cycles (Bordo & Haubrich, 2010). According to Messai and Jouini (2013) before a bank failure occurs, the banks display excessive levels of NPLs, causing economic deterioration. The construction of a theoretical foundation will allow the researcher to investigate the existing theory to identify the existing gaps in the theory. This will magnify the need for the current study and will also contribute to the research questions (Gentner, 1993).
Scholte, Van Teeffelen and Verburg (2015) also made use of a theoretical framework to present which determinants or characteristics impact the object of their study.
The theoretical foundation of this study is used to:
Emphasise the contribution made by this study;
Discuss the findings and shortcomings in past research;
Present a comparative table of the determinants impacting NPLs; and Present the research objectives that guide the hypothesis development.
As an introduction to some theories that have an important impact on this study, Table 3.1 provides an overview of past findings. Table 3.1 also serves as a summary of the past findings discussed in sections 3.2.1 to 3.2.4.

Theoretical associations

In recent years, NPLs became a popular area for research with various authors, such as Bofondi and Ropele (2011), Dash and Kabra (2010), Erdinç and Abazi (2014), Ghosh (2015), and Saba et al. (2012), contributing to the body of knowledge on bank asset quality. The results show conclusively that there are different determinants of NPLs for banks in the developed economies and in the EMEs.
The importance of the determinants has also been debated, leading to disagreements regarding the importance of the macroeconomic determinants in comparison to the microeconomic determinants of bank asset quality. Klein (2013) was of the opinion that the microeconomic variables are significant, although they did not add to the overall explanatory power as determinants of NPLs. Klein (2013) expanded by saying that NPLs are impacted by macroeconomic and microeconomic conditions, irrespective of the bank type (retail, commercial or savings). However, Makri et al. (2014) reported that the microeconomic and macroeconomic determinants both contribute significantly to the NPLs of a bank.
The macroeconomic determinants that predominantly impact NPLs are: GDP growth, inflation rate, and the prevailing interest rate. Table 3.1 presents an abundance of evidence that GDP growth has a negative impact on NPLs, because when the GDP increases more funds are available to borrowers to service their debt (Beck et al., 2015; Buncic & Melecky, 2013; De Bock & Demyanets, 2012; Espinoza & Prasad, 2010; Ghosh, 2015; Klein, 2013; Louzis et al., 2012; Makri et al., 2014; Messai & Jouini, 2013; Nkusu, 2011; Saba et al., 2012; Salas & Saurina, 2002; Škarica, 2013). Erdinç and Abazi (2014) explained that the inverse relationship between GDP and NPLs is proof of the business cycle dependence of NPLs, as well as the sensitivity of NPLs to macroeconomic shocks.
Inflation and interest rates in countries are amongst the determinants that were evaluated by different researchers as key contributors to NPLs (Beck et al., 2015; Buncic & Melecky, 2013; De Bock & Demyanets, 2012; Espinoza & Prasad, 2010; Ghosh, 2015; Klein, 2013; Louzis et al., 2012; Makri et al., 2014; Messai & Jouini, 2013; Nkusu, 2011; Saba et al., 2012; Salas & Saurina, 2002; Senawi & Isa, 2014; Škarica, 2013). However, researchers found contradicting evidence when determining the impact and level of significance that the inflation and interest rate determinants have on NPLs. The theory suggests that as interest rates increase, growth slows and the ability to repay borrowed funds subdues (Espinoza & Prasad, 2010; Ghosh, 2015). Central banks that use an inflation-targeting approach increase interest rates when inflation rises, resulting in higher levels of NPLs during times of rising inflation (Espinoza & Prasad, 2010). However, inflation causes the real amount that was borrowed to decrease, and in theory, borrowers should find it easier to service their debts during these periods (Ghosh, 2015). The influence that interest rates and inflation have on NPLs and each other seem to rest on contradicting theories.
Banks and borrowers alike are vulnerable to external shocks, such as the GFC, which often impact the macroeconomic conditions that the borrowers, lenders, and banks operate in (Erdinç & Abazi, 2014). Although the NPLs performed consistently during the early years of the millennium, the GFC in 2007 caused a sharp increase in the NPLs (Beck et al., 2015; Klein, 2013). Škarica (2013) reported that despite the interventions by central banks and regulators, high levels of NPLs remained for years after the GFC, and the effect may linger on even longer in EMEs.
In the past, regulators intervened when it came to credit extension. Espinoza and Prasad (2010) argued that these interventions reduce the negative impact of macroeconomic and microeconomic risks. The interventions are either in the form of regulatory reforms or changes in the capital requirements of banks (Espinoza & Prasad, 2010). The introduction of stronger prudential measures provides more stability in the banking sector, potentially reducing the NPLs when banks have to abide by stricter capital adequacy requirements (Erdinç & Abazi, 2014; Makri et al., 2014). However, according to Saba et al. (2012), NPLs increase when there are inadequate credit controls. Often after the introduction of stricter credit controls, NPLs also increase as it becomes more difficult for borrowers to qualify for funds, while the borrowers who obtained credit before the introduction of the new credit controls are still debtors to the bank (Salas & Saurina, 2002). Thus prudential and market conduct regulation impacts the NPLs in different ways.
Rapid credit extensions are the most critical cause of poor bank asset quality (Salas & Saurina, 2002). The prevailing economic conditions influence credit extensions, which in turn, impact the NPLs (Bofondi & Ropele, 2011). During the GFC, credit growth reduced to zero, and in some cases, contracted to such an extent that it became negative (Buncic & Melecky, 2013). Salas and Saurina (2002) reported that the credit extension rate indicates the willingness of banks to extend credit to borrowers that are ‘risky’ borrowers. The risk appetite of a bank changes depending on the economic cycle, for example, they might engage in riskier lending practices during economic expansion periods, while they become risk adverse during economic contraction periods.
Saba et al. (2012) reported that high NPLs are a result of risky bank behaviour. Bank management is ultimately responsible for the risk appetite of a bank, and the management of a bank impacts the NPLs (Erdinç & Abazi, 2014). The ROA measures the management behaviour of a bank, as the ROA provides information on the willingness of a bank to engage in risky lending activities (Messai & Jouini, 2013). Higher ROA ratios provide management with fewer incentives to engage in risky transactions (Messai & Jouini, 2013). This implies that good management is paramount to reducing NPLs (Erdinç & Abazi, 2014).
Whether banks engage in personal banking, commercial transactions or international banking, the exposure to credit risk remains. Although credit risk is the main contributor to high NPLs (Salas & Saurina, 2002), banks may also provide funds in foreign currencies which expose the banks to currency risk that also impacts the NPLs (Buncic & Melecky, 2013). According to Senawi and Isa (2014), gold could be a substitute for foreign currency, implying that gold could impact the NPLs in a fashion similar to that of the foreign exchange rates. De Bock and Demyanets (2012) found that the asset quality of a bank has a close relationship with commodities, of which gold is one.

Theoretical shortcomings

According to the BIS, the standard loan classifications are defined as: passed, special mention, substandard, doubtful and loss loans. Laurin and Majnoni (2003) maintained that whatever the classification of the loan, the classification of NPLs differs from country to country. According to Filip (2015), ‘standard’ is the highest quality level of a loan and anything other than ‘standard’ is a potential loss. The ‘standard’ classification is the only consistent evaluation of NPLs across different countries.
Although it is widely acknowledged that there are different NPL definitions in different countries, the majority of scholars (Beck et al., 2015; Buncic & Melecky, 2013; De Bock & Demyanets, 2012; Espinoza & Prasad, 2010; Klein, 2013; Makri et al., 2014; Messai & Jouini, 2013; Nkusu, 2011; Škarica, 2013) based their research on NPLs on more than one country, while only a few (Ghosh, 2015; Louzis et al., 2012; Saba et al., 2012; Salas & Saurina, 2002; Senawi & Isa, 2014) based their research on a single country. Of the studies based on a single country, two studies focused on the USA, and these studies were disaggregated by the states in the USA. Nkusu (2011) stated that the differences between the regulatory, supervision and accounting practices also constrain the inferential abilities of cross-country research on NPLs.
The macroeconomic conditions within countries are the drivers for policy and management decisions within banks (Nkusu, 2011). Louzis et al. (2012) agreed that there is an abundance of literature on the macroeconomic variables, again stating that GDP growth, unemployment rates and interest rates have a significant impact on NPLs, this is, however, applicable to a multi-country perspective. Salas and Saurina (2002) were pioneers in combining macroeconomic and microeconomic variables into a single study, while assessing savings and commercial banks. Although the research included macroeconomic and microeconomic variables, the research by Salas and Saurina (2002) was conducted on various different bank types and not on banks that offer comprehensive banking services, similar to South African banks. As a result, the impact of macroeconomic variables measures for two different types of banks, while the current study evaluates the macroeconomic variables for comprehensive banks. Messai and Jouini (2013) stated that future studies should evaluate the macroeconomic and microeconomic determinants of bank asset quality from a single-country perspective, as is done in the current study.
The importance of the microeconomic determinants of NPLs should not be discounted, as some research (Makri et al. 2014; Saba et al., 2012) suggested that the microeconomic determinants have a similar explanatory power to that of the macroeconomic determinants. However, Klein (2013) disagreed stating that although the microeconomic determinants are consistent, their explanatory power was negligible. Past research (Klein, 2013; Makri et al., 2014; Saba et al., 2012) provides inadequate consistent information on the microeconomic NPL determinants, therefore allowing the current study to fill a void in the literature related to these determinants, and thereby making an original contribution to the South African body of knowledge on bank asset quality.
To further enhance the knowledge on the determinants of NPLs, this study provides more information on bank asset quality from an EME perspective by focusing on the:
GFC resilience;
Regulatory changes; Profitability measures;
Impact of gold sales as an asset of last resort;
Country-specific bank-per-bank perspective; and
Macroeconomic and microeconomic determining variables.
Louzis et al. (2012) suggested that the microeconomic determinants of NPLs may have been influenced by the GFC and recommended that future research should evaluate the impact of the GFC on microeconomic determinants. NPLs remain one of the most common causes of a deteriorating economic environment and bank failures (Messai & Jouini, 2013). Due to the developed economies experiencing the largest shock during the GFC, Nkusu (2011) researched the impact that the GFC had on NPLs in these advanced economies.
The credit risk appetite of a bank is a primary driver of NPLs for banks who engage in riskier lending practices (Saba et al., 2012). Ghosh (2015) recommended that regulators intervene in the banking environment to ensure that irresponsible lending does not take place. Chipeta and Mbululu (2012) found that in South Africa, the NCA reduced the credit being extended to individual lenders for credit cards, bank overdrafts and conventional loans, while the total credit extended grew. Erdinç and Abazi (2014) recommended that a study of the determinants of NPLs should allow regulators to enact appropriate regulatory measures to manage NPLs and reduce risk in the banking sector. Chipeta and Mbululu (2012) recommended that future research should focus on how the GFC and NCA impacted credit extended by South African banks. This study is an excellent platform to study asset quality stemming from their recommendations to determine the impact of the GFC and the NCA on NPLs.
Jang and Kataoaka (2013) concluded that commodity exports have an impact on NPLs in New Zealand. This is because those banks have significant credit exposure in the commodity markets (Jang & Kataoaka, 2013). According to Baxter (2009), South Africa has the largest exposure in the mineral export market, especially in the gold commodity market. Senawi and Isa (2014) found that in the Malaysian banking sector, the value of gold impacts the NPLs in the country. A shortcoming of the study by Senawi and Isa (2014) is regarding the spot price of gold as a determinant of NPLs and not the value of gold exports or gold sales. In this study, gold sales are the determinant of NPLs, because the spot price of gold only determines the price at which gold trades, while the gold sales allow this study to capture the income generated by gold sales.
The shortcomings in in the regulatory environment and the potential impact of gold sales on NPLs lay the foundation for studying the impact of the risk behaviour by banks. From a regulatory perspective, the NCA only reduced the individual client credit, while the total credit extended by banks increased, irrespective of the introduction of the NCA regulations. Chipeta and Mbululu (2012) found that the increase in credit was primarily because more credit was extended to the private sector (this would include mining firms). Messai and Jouini (2013) measured the incentive of risky lending activities by introducing the ROA as a measure of the banks’ incentive to generate income from risky lending practices. Support for engaging in risky activities also comes from shareholders, and not only from bank managers, motivating that research should also determine which determinant would have a more significant impact, shareholders or management. Irrespective of the driver behind reckless credit extension, Saba et al. (2012) concluded that risky lending behaviour causes larger NPLs.
This section identified the potential caveats of past research, highlighting the potential for new research. Although the contribution is evident, a summary of the contributing factors is presented in the following section.

Summary of research contribution

This research aims to contribute to the body of knowledge by investigating five gaps that exist in the research regarding the determinants of NPLs in the South African banking sector. The five gaps in the body of knowledge pertain to research on the determinants of NPLs from a single country perspective, where a consistent definition of NPLs is used throughout the study. This is in contrast to existing research where a variety of countries were considered in the study, which invariably results in differences in the definitions of NPLs and differences in the regulatory environment (Beck et al. 2015; Nkusu, 2011).
This research also contributes to the existing body of knowledge on the impact that structural breaks, macroeconomic conditions and microeconomic conditions have on the determinants of NPLs for comprehensive banks, thus providing consideration for the international competitiveness of the South African banks (Klein, 2013; Messai & Jouini, 2013). A gap remains in the knowledge about the impact of the GFC on the NPLs of comprehensive banks in EMEs, since most studies on these determinants were conducted in developed economies (Klein, 2013; Nkusu, 2011).
There have been numerous regulatory changes in the South African financial markets since the 1980s. The first changes took place in 1987 when the De Kock Commission stated that the banking sector was over-regulated; this was soon followed by the introduction of a new banking act in 1990 and another government sponsored commission, the Melamet Commission in 1993 (Botha & Makina, 2011). The most recent changes were the introduction of the Twin Peaks financial supervision model and the introduction of the NCA (Botha & Makina, 2011; Chipeta & Mbululu, 2012), this study is undoubtedly unique and contributes to the knowledge on the determinants of NPLs during times of regulatory changes.
The literature indicates that this study is, to the best of the author’s knowledge, the first to include gold sales as a determinant of NPLs. Past studies (Jang & Kataoaka, 2013; Nikolaidou & Vogiazas, 2017; Senawi & Isa, 2014) did not use gold sales but rather used the spot price of gold or the production output of gold. The study by Jang and Kataoaka (2013) use agricultural commodity spot prices as these are the predominant exports in New Zealand.
Saba et al. (2012) warned that risky lending activities results in poor asset quality and found that bank managers drive the risky activities. This study contributes to the existing knowledge by reviewing the impact of the risky lending activities from a shareholder perspective, as well as a managerial perspective, by introducing the measures ROE and ROA, respectively.
The five primary contributions of this study are based on the theoretical shortcomings as presented in Section 3.2.2. These five contributions also lay the foundation for the research objectives of this study.

Research objectives

Similarly to the five contributions presented in Section 3.2.3, this study also has five research objectives. These research objectives introduce the five hypothesis. The research objectives aim to fulfil the shortcomings of past studies and satisfy the recommendations for future research by authors of past research.
The five research objectives set by this study are to:
Assess the impact of the GFC on bank asset quality in an EME (cf. Klein, 2013). Determine whether the NCA is effective in reducing NPLs (cf. Chipeta & Mbululu, 2012; Salas & Saurina, 2002).
Conduct a single-country study on the macroeconomic and microeconomic determinants of NPLs in an EME (cf. Saba et al., 2012).
Establish whether a primary commodity export of an EME has an impact on NPLs (cf. Jang & Kataoaka, 2013; Senawi & Isa, 2014).
Determine whether interest income on loans has superior explanatory power regarding NPLs, in comparison to the ROA and ROE (cf. Makri et al., 2014; Messai & Jouini, 2013; Saba et al., 2012).
Considering the five objectives of this study, the following hypotheses were developed to achieve the outcomes of the research objectives.

HYPOTHESIS DEVELOPMENT

An important part of conducting empirical research is to approach a research question through hypothesis testing (Banarjee, Chitnis, Jadhav, Bhawalkar & Chaudhury, 2009). As such, Banarjee et al. (2009) stated that hypotheses have three important characteristics, namely, they are simple, specific and developed when planning a research project.
Five hypotheses (as discussed below) were developed for this study, according to the above-mentioned characteristics of hypotheses.

TABLE OF CONTENTS
DECLARATION
ACKNOWLEDGEMENT
ABSTRACT
ABSTRACT: AFRIKAANS
ABSTRACT: NORTHERN SOTHO
TABLE OF CONTENTS
LIST OF FIGURES
LIST OF TABLES
LIST OF ABBREVIATIONS
CHAPTER 1: INTRODUCTION: THE QUALITY ASSET DILEMMA
1.1 INTRODUCTION
1.2 BACKGROUND
1.3 EXPLORATORY LITERATURE REVIEW
1.4 RESEARCH PROBLEM
1.5 RESEARCH CONTRIBUTION
1.6 STUDY OUTLINE
1.7 SUMMARY
CHAPTER 2: LITERATURE REVIEW: BANKS AND QUALITY OF ASSETS
2.1 INTRODUCTION
2.2 AN INVESTIGATION INTO BANK ASSET QUALITY
2.3 BANK TYPES
2.4 BANK FINANCIAL REPORTING
2.5 BANK REGULATION AND SUPERVISION
2.6 BANK CREDIT RATINGS
2.7 SUMMARY
CHAPTER 3: THEORETICAL FOUNDATION AND HYPOTHESIS DEVELOPMENT: DETERMINANTS OF BANK ASSET QUALITY
3.1 INTRODUCTION
3.2 THEORETICAL FOUNDATION
3.3 HYPOTHESIS DEVELOPMENT
3.4 SUMMARY
CHAPTER 4: RESEARCH DESIGN AND EMPIRICAL METHODOLOGY
4.1 INTRODUCTION
4.2 RESEARCH DESIGN
4.3 DATA
4.4 STATISTICAL ANALYSIS
4.5 OVERVIEW OF PANEL DATA MODELS
4.6 PANEL DATA MODEL SPECIFICATIONS
4.7 PANEL DATA SPECIFICATION AND DIAGNOSTIC TESTS
4.8 PROPOSED PANEL DATA REGRESSIONS
4.9 ETHICAL CONSIDERATIONS
4.10 SUMMARY
CHAPTER 5: BANK ASSET QUALITY VARIABLE DESCRIPTIONS
5.1 INTRODUCTION
5.2 DATA REPRESENTING DETERMINANTS OF BANK ASSET QUALITY
5.3 DEPENDENT VARIABLE
5.4 INDEPENDENT VARIABLES
5.5 SUMMARY OF EXPECTED VARIABLE SIGNS
5.6 SUMMARY
CHAPTER 6: EMPIRICAL RESULTS
6.1 INTRODUCTION
6.2 DESCRIPTIVE PANEL DATA EXPLORATION
6.3 REGRESSION RESULTS
6.4 SUMMARY
CHAPTER 7: SUMMARY, CONCLUSIONS AND DIRECTION FOR FUTURE RESEARCH
7.1 INTRODUCTION
7.2 RESEARCH PURPOSE
7.3 SUMMARY OF RESULTS
7.4 CONTRIBUTION
7.5 RECOMMENDATIONS
7.6 LIMITATIONS
7.7 FUTURE RESEARCH
7.8 SUMMARY
REFERENCES
GET THE COMPLETE PROJECT