Bank risk and performance in an emerging market setting: the case of Bangladesh

Over the past 20 years, several banking crises have taken place and increased concerns to the regulators about the financial stability system. Academicians, researchers and regulators identified several causes for maintaining the financial stability and urged for the stringent prudential regulation. The capital adequacy regulations play a prominent role for the prudential purposes compared to other tools proposed by the regulators (Altunbas et al., 2007).

To maintain the stable position in the international financial system, Basel Capital Accord named Basel I was introduced first in 1988. Because of the weaknesses of the Basel I, Basel II introduced in 2004. The prime objective of Basel Accord is to strengthen the bank’s capital position and minimization of risks. Investigations into whether the Basel Capital Accord work effectively or not proliferated after the global financial crisis 2007-2008. As a result, in response to the global financial crisis, Basel III was introduced in 2010 by the Basel Committee on Banking Supervision. Regulators in most of the countries around the world are going to implement the Basel III step by step with varying timelines and methodologies. Already, many countries were adopted Basel II, for example, the European Union in 2008, Singapore in 2008, India in 2009 and Bangladesh in 2010. Needless to say, the recent financial crisis raised questions to the mind of the policy maker about the effectiveness of capital regulation and its effect on the bank risk and performance. It is thus no surprise to increase the concern about the investigation on the linkage between bank capital, risk and performance. Most of the empirical literature focuses on the relationship between capital, risk and profitability in US and European contexts (Demirguc-Kunt et al., 2013; Altunbas et al., 2007). However, the recent empirical research provides novel insights into the banking sector of alternative countries or regions, for example, the MENA (Middle Eastern and North African) regions (Bitar et al., 2016; Naceur and Omran, 2011), the Sub-Saharan African (Flamini et al., 2009), the Tunisia banking sector (Bennaceur and Goaied, 2008), the Egypt banking sector (Ben Naceur and Kandil, 2009) and the Asian banking sector (Zheng et al., 2017; Tan and Floros, 2013; Lee and Hsieh, 2013; Lee et al., 2015).

Despite the ongoing debate on the necessity of financial reforms, there is scant research on this topic, particularly in Asian countries. The research in the banking sector of the Asian countries is very important because banks are the main source of finance in the private sector of this region Lee and Hsieh (2013) and Deesomsak et al. (2004). Therefore, the banking sector of the Asian countries is the key laboratory for the investigation. To extend the research on Asian countries, this study endeavors to investigate the impact of bank capital on bank performance and risk in the banking sector of emerging economy. This investigation has considered Bangladesh as an ideal laboratory to examine the impact of bank capital on bank performance and risk. The study has considered Bangladesh as a benchmark economy for the investigations as it is one of the emerging economies. The study acknowledges with the study of Khan et al. (2013) that the definition of an “emerging economy” is problematic. However, Bangladesh has considered being the largest economies in the twenty-first century as one of the “next 11” emerging economies along with the BRIC countries by the global asset management company (Goldman Sachs) in 2005 (www.goldmansachs.com). In Bangladesh, several capital regulations reform has undergone over the last two decades, and the country achieved a consistent economic growth over six per cent after the millennium period (Zheng et al., 2017). Thus, the study recommends the significance of the country selection as Bangladesh. In addition, the investigation claims that the differences of national culture give different banking practices with different research findings (Zheng and Ashraf, 2014; Ashraf, 2017; Ashraf et al., 2016b). Therefore, the findings from Bangladesh will contribute to the existing literature significantly and will be effective for the other countries with a similar culture. To the best of my knowledge, this is the unique investigation on Bangladesh.

The previous empirical literature shows mixed results on the association between bank capital and risk. Some of the prior studies indicate there is a positive association between capital and risk, means the higher the capital ratio leads to the higher risk which is supported by the regulatory hypothesis (Iannotta et al., 2007; Demirgüç-Kunt and Huizinga, 2000). Existing theories suggest that the vital reason for introducing capital regulations is referred in the point of Moral Hazard Hypothesis (Demirgüç-Kunt and Kane, 2002; Hussain and Hassan, 2005). According to this hypothesis, there is a negative relationship between capital regulations and risk (Agusman et al., 2008; Jacques and Nigro, 1997). The State Preference Model uses by Liu et al. (1996) and Lin (1994) also finds a negative association between capital and risk. They claim that the level of risk exposures is reduced because of stricter capital management. In contrast, Lin et al. (2005), Blum (1999) and Keeley and Furlong (1990) used portfolio theory and questioned regarding the effectiveness of the capital management. They argue that banks are forced to have a greater level of risk exposures when the capital regulations tend to be strict. However, the puzzle between capital and risk, as suggested by Altunbas et al. (2007) and Hughes and Mester (1998), is that the association between Capital regulation and bank risk-taking is affected by the amount of profitability. Goddard et al. (2004) claim that the high ratio of capitalized bank earns a high profit and faces less risk.

From the above theoretical literature, it has observed that there is a puzzle between the bank capital, risk, and performance. The literature also shows inconclusive results on the relationship. Hence, there is still debate whether bank capital positively or negatively impacts the bank performance and risk. By considering the above fact, thus, this study attempts to find out the answer to the two questions. What is the impact of bank capital on the bank performance of Bangladesh? What is the impact of bank capital on the bank risk of Bangladesh?

This study is expected to play significance role for the academician, researcher, and policy maker because it contributes to the existing literature in several ways to fill the gap. First, this is the unique study addressing the impacts of capital requirement on both risk and performance simultaneously in the Bangladeshi banking sector. Also, this study is different from other studies because it uses two measures of capital, i.e. risk-based capital and non-risk-based capital. The existing literature indicates a debate about the capital requirements effectiveness. For example, Anginer and Demirgüç-Kunt (2014) and Demirguc-Kunt et al. (2013) argue what types of capital should use by the bank and what would be the structure of it. This is supported by the other researchers, Cathcart et al. (2015) and Dermine (2015), who shed doubts on the effectiveness of the risk-weighted assets in determining risk exposure of the bank. So, this paper adds value to the extant literature on highlighting the capital requirements effectiveness by examining the effect of risk-based and non-risk-based capital on bank risk and performance in Bangladesh. The study uses various measures of risk as well such as default risk, credit risk and overall risk.

Second, the prior literature on Bangladesh indicates a traditional association between capital and risk, risk and performance, but this study extends by adding capital and profit relationship with risk. The study is the complements of the recent studies of Bitar et al. (2016), Zheng et al. (2017), Lee et al. (2015) and Lee and Hsieh (2013).

Third, this research considers long period (2000-2015) of unbalanced panel data set of 30 commercial banks of Bangladesh.

Fourth, the study uses simultaneous equation modeling with generalized methods of moments (GMM) technique for the regression analysis to examine the simultaneous effect of bank capital requirement on risk and performance and testing the robustness of the results by using two-stage least squares (TSLS) regression.

Finally, the study trying to draw some suggestions regarding the influence of bank capital on the performance and risk in Bangladesh which will be beneficial not only for Bangladesh but also for other countries of similar economic nature as well as for the academicians, researcher, and policy maker.

The remaining section of this study proceeds as follows. Section 2 provides about related literature. Section 3 shows the research methodology. Section 4 includes the analysis and findings. Section 5 summarizes the concluding remarks and indicates directions for the future researchers.

This study divides the existing literature into three parts. The first part displays the existing literature related to the linkage between bank capital and performance. The second part of the study focuses on the related literature based on the association between bank capital and risk. Finally, the third part provides the literature related to variables used in the study.

The prior literature indicates that bank performance is considered as an influential factor in the relationship between risk and capital (Lee and Hsieh, 2013; Moon and Hughes, 1997; Altunbas et al., 2007). Many researchers finds a positive association between capital and bank performance (Jacques and Nigro, 1997; Lin et al., 2005; Goddard et al., 2004; Rime, 2001; Lee and Hsieh, 2013; Naceur and Omran, 2011; Ben Naceur and Kandil, 2009; Mbizi, 2012). The positive relationship of prior literature emphasizes to hold more capital to increase bank performance. For example, Iannotta (2006) claims that banks are required to increase their capital ratios commensurably with the amount of risk taken by following the Basel guidelines. The author has mentioned that higher capital ratios diminish banks’ leverage taking behavior and thus leads to their expected profits. Fiordelisi et al. (2011) argue that lower return induce bank management to choose a higher point on the efficiency frontier to improve their profits which leads to investments in riskier portfolios. As a result, a positive association should be expected between bank capital and performance. Zhang et al. (2008a) suggest that commercial banks, under the constraint of capital, should allocate resources and augment their business to increase liquidity and profitability. In contrast, differing from the positive relationship, Altunbas et al. (2007) find that inefficient European banks appear to hold more capital. Goddard et al. (2013) also concluded a negative association exists between profitability and capital in banks of nations from European Union member for the period 1992 to 2007. In addition, Guidara et al. (2013) conclude that there is no strong evidence that changes in capital buffer affect the profitability measured by ROE. Therefore, mixed results have found in the previous literature. Table I summarizes the previous literature in a more scientific way as follows.

In the context of the implications of regulatory policies from the regulator, examination of the associations between risk and capital is considered as one of the important issues today (Lee and Hsieh, 2013). It is assumed that higher capital will have a positive impact on risk of the banking sector (Lee and Chih, 2013), but empirical results are mixed. Findings of some research reveal that there is a positive association between risk and capital (Shrieves and Dahl, 1992; Blum, 1999; Rime, 2001; Altunbas et al., 2007; Laeven and Levine, 2009; Kim and Santomero, 1988; Koehn and Santomero, 1980; Kahane, 1977; Lin et al., 2005). On the other hand, some studies find a negative relationship between risk and capital (Lee and Hsieh, 2013; Lee and Chih, 2013; Ho and Hsu, 2010; Zhang et al., 2008; Agoraki et al., 2011). Demirgüç-Kunt and Kane (2002) argue that the negative association between capital and risk may be referred as moral hazard hypothesis (MHH) which implies that banks with undercapitalized value take more risk to take advantage of existing horizontal deposit insurance schemes. Altunbas et al. (2007) and Agusman et al. (2008) found that inefficiency is positively related with risk-taking which advocates the MHH, i.e. banks with a low-performance level are more vulnerable than banks with a high-performance level. However, no relationship between bank capital and risk has been found by Guidara et al. (2013). By using a mean-variance approach, Kim and Santomero (1988) and Koehn and Santomero (1980) showed that single capital ratio regulation is not enough to control risk. For ignoring the option value of deposit insurance, the mean-variance is not appropriate (Keeley and Furlong, 1990; Furlong and Keeley, 1989). They used a contingent-claims model and showed that enlarged capital would not raise portfolio risk of banks. The reason behind this is that the value of the deposit insurance reduces as the capital increases. Table II summarizes the prior literature.

Prior literature uses various measures of performance such as return on assets (ROA), return on equity (ROE), net interest margin, earning per share, Tobin’s Q, etc. Previous literature also indicates that performance measures can be calculated in various ways. For example, ROA can be calculated by considering before tax profit, or after-tax profit, or total assets, or average total assets, or total earning assets or average earning assets. This study uses ROA before tax as a measure of performance by following the study of Tan (2016), Demirgüç-Kunt and Huizinga (1999) and others.

This study uses three types of risk, namely default risk, credit risk and overall risk. The study measures default risk by following the study of Iannotta et al. (2007), where they measure default risk using natural logarithm of Z-score. The Z-score is used by a number of empirical studies as the risk or stability indicator in the banking sector (Liu and Wilson, 2013; Liu et al., 2013; Tan, 2016) and it is calculated by using the sum of a bank’s ROA and equity to total assets ratio (ETAR) over the standard deviation of the bank’s ROA. A higher Z-score indicates that there is a higher stability and lower risk (Tan, 2016). By following (Agoraki et al., 2011), the study measures credit risk by the non-performing loans to total loans. The higher non-performing loans to total loans ratio indicate that there is more risk for losses from loans defaults (Zhang et al., 2013). This investigation also measures overall risk by the loan loss provision to net interest revenue ratio, used by Baselga-Pascual et al. (2015).

This study uses two types of capital measures:

1. the traditional non-risk-based capital (total shareholders’ ETAR) which is known as actual capital; and

2. the risk-based capital (capital adequacy ratio [CAR]) which is known as regulatory capital.

Many prior empirical studies use non-risk-based capital (Shrieves and Dahl, 1992; Altunbas et al., 2007; Lee and Hsieh, 2013; Tan and Floros, 2013; Iannotta et al., 2007; Sufian and Habibullah, 2009; Sufian, 2012; Amidu and Hinson, 2006; Athanasoglou et al., 2008; Dietrich and Wanzenried, 2011). Many studies use regulatory capital measures (Aggarwal and Jacques, 2001; Jacques and Nigro, 1997; Ediz et al., 1998).

This study uses the ratio of total loans to total assets as a measurement of liquidity in the performance equation by following the study of Goddard et al. (2013). The largest ratio indicates a lower level of liquidity. Thus, a large amount of loans to the customer indicates more interest revenue from them. Hence, the negative impact of liquidity on bank performance is expected. In contrast, Bourke (1989) argues that the higher liquidity level of a bank generates higher profit. The higher volume of loans sometimes leads to decline the bank performance if the bank does not have a good risk management system. Therefore, it is expected that there may be a positive or negative impact of liquidity on bank performance.

The study uses operating expenses to total assets as cost inefficiency measurements by following the study of Dietrich and Wanzenried (2011). These measures have been used extensively in the existing literature (Liu and Wilson, 2010; Kosmidou, 2008; García-Herrero et al., 2009). The existing literature shows mixed results. For example, Molyneux and Thornton (1992) find the positive impact of cost inefficiency on bank performance. Naceur (2003) also find a positive relationship between cost inefficiency and bank performance in the Tunisian banking industry. The positive relationship between cost inefficiency and performance also supports the efficiency wage theory. The existing literature also indicates that there is a negative relationship between cost inefficiency and bank performance. For example, Athanasoglou et al. (2008) claim that well-managed banks have the ability to reduce operating costs, which in turn increases bank performance. This opinion is supported by the study of Bourke (1989) and Jiang et al. (2003a). Therefore, it is expected that there may be a positive or negative impact of cost inefficiency on bank performance.

The prior literature indicates cost inefficiency is a source of risk (Poghosyan and Čihak, 2011; Männasoo and Mayes, 2009). Louzis et al. (2012), in their bad management hypothesis, suggest that cost inefficiency is positively associated with the increases of future non-performing loan which ultimately leads to bad management and thus poor skills in credit scoring. Therefore, a positive relationship is expected between cost inefficiency and bank risk.

This investigation uses net profit after tax over a total number of employees as a measurement of labor productivity. This variable has been widely used by the researcher in the prior literature to investigate its impact on bank performance (Tan and Floros, 2013; Tan and Floros, 2012b; Tan and Floros, 2012a, Tan and Floros, 2012c; Athanasoglou et al., 2008). The higher the labor productivity ratio indicates banks efficient management which reflects banks performance. Therefore, this study predicts a positive association with performance.

This study uses off-balance sheet items to total assets as non-traditional activity measures. This variable has been included in the performance equation. Bank can increase their sources of income by the off-balance sheet activities without changing the capital structure which ultimately affects the performance (Apergis, 2014; Deelchand and Padgett, 2009). Rahman et al. (2015) find a significant negative association between non-traditional activity and bank profitability. Hence, no prior expectation of this variable.

The study uses the ratio of non-interest income to total income as a measure of income diversification. A bank engaged in different types of businesses is able to generate more income (Tan and Floros, 2012b). In addition, the more diversified banks can reduce their costs easily from economies of scope. Jiang et al. (2003b) find a positive association between income diversification and performance. However, others find (Demirgüç-Kunt and Huizinga, 1999; Gischer and Jüttner, 2001) negative association of income diversification with performance because there is stronger competition to generating free income compared to traditional interest income activity which ultimately reduces bank performance. Thus, the study finds mixed results and no prior expectation regarding the relationship between income diversification and performance.

This study includes deposit ratio in the risk equation and total deposit to total assets has considered as a measure of deposit ratio. As deposits are insured, a higher deposit ratio raises moral hazard of banks to fund in risky investments (Soedarmono et al., 2010). In addition, the higher deposit ratio increases leverage risk. Therefore, a positive relation is predicted between deposit ratio and bank risk.

This study uses total liabilities to total assets as a measure of leverage and includes this variable in the risk equation. Higher leverage may be both beneficial and cost to the banks. For example, the high leverage ratio indicates more debt in the capital structure which indicates higher interest payments and thus create pressure on the bank income (Zheng et al., 2017), as well as increase risk. On the other hand, the higher leverage may increase profit by reducing risk because interest payments provide a tax shield. Chaibi and Ftiti (2015) claim that risk is affected by capital structure. Highly leveraged capital leads to a tendency to a higher risk-taking because of the need to produce higher returns with lower capital. Thus, no prior expectation of this variable on the impact of bank risk.

This study used concentration ratio as a measure of market power and included in the performance equation. According to structure conduct performance (SCP) hypothesis, the banks’ behavior is affected by market power, while the market structure plays a decisive role in bank performance. The idea of this hypothesis is based on the fact that in a more concentrated market where significant shares are occupied by a few banks, the competitive condition is lower, while higher concentration leads to greater market power, and the resultant increase in the collusive behavior leads to higher profits. These indicators were recently used by (Fu et al., 2014; Al-Muharrami et al., 2006) to measure competition in the banking industry.

The study measures economic growth by the annual GDP growth rate. Because of increasing the demand for lending during the cyclical upswings, some researchers claims GDP growth rate has a positive impact on the bank performance (Athanasoglou et al., 2008; Bikker and Haaf, 2002; Demirgüç-Kunt and Huizinga, 1999). However, Tan and Floros (2012a) investigate the association between GDP growth and bank profitability in China and finds a negative association between both variables. They argue that the economic growth improves the environment of the business, and which lowers the entry barriers of banks. As a result, the competition increased dampens the performance of banks. Thus, this study has no prior expectation regarding this variable.

This study uses a bank lending interest rate variable in the risk equation. Geng et al. (2016) examine the effects of the interest rates on bank risk in China by using the ratio of interest income to total loans as a measure of lending interest rate. Their findings indicate the lower level of bank lending interest shows lower interest income from loans, which causes reduction of bank earnings. They also claim that the low lending interest rate shows that some loans fail to be recovered and become non-performing loans, which causes an increasing risk. Therefore, to achieve the target profit, banks have to invest in high risky projects or financial instruments. Hence, the study expects a negative relation between bank lending interest rate and bank risk.

This study uses annual inflation rate to see the effect of it on the bank risk. Arpa et al. (2001) argue that if inflation increases then the banking sector share of risk provisions in the total loans also increases. Hussain and Hassan (2005) also find a positive association between inflation and bank risk.

This study mainly uses secondary sources of data. As sources of data, the study uses Bankscope database (Bitar et al., 2016; Demirguc-Kunt et al., 2013). Also, to fill up the missing data, the investigation relied on the annual audited financial statement of each bank collected from the bank websites and Dhaka Stock Exchange, Bangladesh. The industry and macroeconomic data have gathered from the Bangladesh Bank website (www.bb.org.bd) and World Bank database (data.worldbank.org). The study also uses journals, books, different websites, library facility for the desk and extensive study as secondary sources of data.

To meet the desired objectives, the study has considered an unbalanced panel data of 30 sample banks of Bangladesh during the period from 2000-2015. The details about the sample have shown in Table III.

The study investigates the effect of risk-based capital and traditional non-risk based capital on the bank risk and performance of Bangladesh. For this purpose, the study uses two simultaneous equations by following the previous literature (Zheng et al., 2017; Tan and Floros, 2013; Vollmer and Wiese, 2013; Francis and Osborne, 2012; Fiordelisi et al., 2011; Ben Naceur and Kandil, 2009; Altunbas et al., 2007; Nier and Baumann, 2006; Lee and Hsieh, 2013; Hussain and Hassan, 2005; Ayuso et al., 2004; Jacques and Nigro, 1997) as follows:

In the above two equations, the subscript i indicates the cross-sectional dimension across banks, t refers to years and ε_it denotes the random error term which includes the unobserved bank-specific effect and idiosyncratic error.

The first equation uses to tests whether the performance of banks is affected by the bank capital. In the equation, P_it indicates bank performance using ROA before tax (ROAB); β₀ is the constant term; P_it-1 is the one period lagged performance variable; C_it is the bank capital has measured by the CAR and Total Shareholder’s ETAR; R_it is the Bank Risk has measured by the default risk (LNZSCORE), credit risk (Non-performing loans to Total loans, NPLTL), and overall risk (Loan loss provision to Net interest revenue, LLPNIR); LIQD_it is the liquidity has measured by the total loans to total assets ratio; OETA_it is the cost inefficiency measures has calculated by the operating expenses to total assets ratio; LP_it is the labor productivity has calculated by the net profit after tax to total number of employees; OFBSTA_it is the non-traditional activity measures by the off-balance sheet items to total assets ratio; INCD_it is the income diversification has calculated by the ratio of non-interest income to total income; CONC3_it is the market power has measured by the concentration ratio of the three largest bank assets to total assets; and GGR_t is the annual GDP rate (per cent).

The second equation uses to tests whether the risk taking the behavior of banks are affected by the bank capital. In the equation, R_it, C_it, and P_it indicates same meaning as defined in the first equation; γ₀ is the constant term; R_it-1 is the one period lagged risk variable (LNZSCORE_it−1, NPLTL_it−1, LLPNIR_it−1); OETA_it is the cost inefficiency measures has calculated by the operating expenses to total assets ratio; DPR_it is the deposit ratio has measured by the total deposit to total assets ratio; LEV_it is the leverage has calculated by the total liabilities to total assets ratio; BLINT_it is the bank lending interest rates (per cent); INFLR_t is the annual inflation rate (per cent).

The descriptions of the variables and their expected effects on the bank performance and risk have shown in Tables IV and V.

The previous empirical literature uses a variety of methods to solve the simultaneous equation, but not all methods are appropriate for all cases. The study examines the impact of bank capital on performance and risk by using Two-Step system GMM estimator developed by Arellano and Bover (1995) and Blundell and Bond (2000). GMM estimator has applied here because some problems have faced by the simultaneous equations, namely, endogeneity issue, serial correlation and heteroscedasticity. Lee and Hsieh (2013) argue GMM technique is more effective compared to OLS technique. OLS technique provides biased results in the estimation of a dynamic model (Nickell, 1981). To be more specific, two types of variants observed in the GMM: differenced GMM estimator and system GMM estimator. The study prefers the later one compared to the previous one because the system GMM estimator provides more precise results than difference GMM by solving the critical issues. GMM has considered superior to dynamic panel model because it provides estimates of fixed effects and OLS (Bond, 2002).

To run the regression using GMM with Eviews-8, the study first tests the identification of simultaneous equation to see whether the equation is over-identified or not because it is one of the pre-condition of using GMM. The investigation also seeks whether the models of each equation suffers from endogeneity, serial correlation and heteroscedasticity issues. For these issues, the study employs Durbin–Wu–Hausman endogeneity test, Breusch–Godfrey serial correlation LM test and white heteroscedasticity test. The study also considers Hausman test for fixed/random effect to see which is appropriate for the model. Also, to test the validity of the instruments and over-identifying restrictions the Sargan j-test applies here.

The study also provides the descriptive statistics like mean, minimum, maximum, and standard deviation value to know the depth understanding about the study variables. The Pearson’s correlation provides here to confirm that the study variables have not suffered any multicollinearity issues.

Finally, this study has applied robustness checks by introducing TSLS instead of GMM to justify the findings whether the results of the two methods are consistent or not.

This chapter has divided into four different sections. The first section highlights the descriptive statistics of all variables of the total sample year observations. The second section provides Pearson correlation matrix among the independent variables used in the study. The third section indicates regression analysis. Finally, the fourth section provides evidence of the accuracy of study findings by the robustness check.

Table VI shows the descriptive statistics of the study variables which includes the minimum value, maximum value, mean value, standard deviation value and the number of observations of the sample variables. Among the study variables, it has been observed that the bank lending interests rate (BLINT) has a greater amount of variation among the samples (standard deviation = 0.977) followed by economic growth (GGR) with a standard deviation of 0.834, bank risk (LNZSCORE) with a standard deviation of 0.504, and others. The only performance variable ROA before tax (ROAB) shows a mean value of 0.023 with a maximum value of 0.065, the minimum value of 0.001, and standard deviation of 0.010. There is not so much variation of ROAB among the samples. Among the three risk measures of LNZSCORE, NPLTL and LLPNIR, LNZSCORE shows the greater variations among the samples indicate variations of risk or financial stability. The minimum value of NPLTL (0) indicates one or some of the sample has no non-performing loan which is a good sign and the evidence finds it for the one bank. The ratio of LLPNIR indicates a minimum value of negative figure (−1.331) which means one or some of the samples have a shortage of provision or negative net interest revenue. The average CAR (11.28 per cent) indicates higher capital adequacy has maintained by the sample banks on an average as per requirements of the BASEL II accord. But, the minimum value (5.65) indicates one or some of the banks has not maintained the minimum capital requirements of 10 per cent as per BASEL II accord. The equity to capital ratio (ETAR) indicates a mean value of 0.074 which is far from the maximum value of 0.154 and a minimum value of 0.002. So, there is a greater deviation from the average value of ETAR.

Table VII shows the degree of relationships among the independent variables used in the study. This study uses Pearson correlation matrix to examine whether any multicollinearity issues exists between independent variables.

Among the study variables, the highest correlation is 0.61 between the two capital ratios CAR and total shareholders’ ETAR. Kennedy (2003) claims that the correlation value above 0.70 indicates multicollinearity problem exists between the variables. Another academician, Gujarati (2009), states that the correlation value above 0.80 between the two variables creates serious multicollinearity problem. Thus, this study indicates the correlations among the variables are not so strong, which suggest non-existence of multicollinearity issues.

Table VIII reports the empirical results of the equation one where six model uses for the regression analysis. The table provides the impact of bank capital on the performance of the Bangladeshi banking sector. The first three models (1-3) depict the simultaneous effect of bank capital (CAR) and risk (default risk, LNZSCORE; credit risk, NPLTL; and overall risk, LLPNIR) on the bank performance (ROA before tax, ROAB) of Bangladesh. The second three models (4-6) show the simultaneous effect of bank capital (ETAR) and risk (default risk, LNZSCORE; credit risk, NPLTL; and overall risk, LLPNIR) on the bank performance (ROA before tax, ROAB) of Bangladesh. The regression findings indicate that the lag dependent variable (ROAB_t-1) is positively and statistically significant with the performance across all the six models. Thus, it is evidence that there is a persistence of performance from one year to next year. The study highlights that the main independent variable bank capital (CAR; and ETAR) has a significant and positive impact on bank performance (ROAB) using all the three types of risk measures across all the six models. Thus, the higher the capital ratio indicates higher the bank performance. The findings of the study are in line with Psillaki and Mamatzakis (2017), Zheng et al. (2017), Casu et al. (2017), Bougatef and Mgadmi (2016), Lee and Hsieh (2013), Mbizi (2012), Naceur and Omran (2011), Ben Naceur and Kandil (2009), Pasiouras and Kosmidou (2007), Lin et al. (2005), Goddard et al. (2004), Rime (2001) and Jacques and Nigro (1997). The findings are inconsistent with the views that higher capital requires higher opportunity cost of holding and thus jeopardize the performance of the banks (Goddard et al., 2013; Karels et al., 1989; Brewer III and Lee, 1986). The study finds that all the three measures of risk across the six models indicate a negative impact on bank performance. Among all control variables, all the variables significantly affect the bank performance except economic growth (annual GDP growth rate, GGR). The study includes this variable as excluding it weaken the model. The study finds liquidity ratio (LIQD) is positively and significantly impact on bank performance. Higher liquidity ratio indicates a lower level of liquidity. Thus, the study finds a negative relationship of liquidity with a performance which indicates higher the loans leads to higher interest revenue and improve bank performance. The result is consistent with the findings of Molyneux and Thornton (1992) and Goddard et al. (2013); it is inconsistent with Bourke (1989). Cost inefficiency (OETA), i.e. the ratio of operating expenses to total assets, is positively and significantly affects the bank performance which means more expenditure on salary, wages to the staff increases their productivity and thus performance increases. The findings have supported the efficiency wage theory (Tan, 2016) and also consistent with the findings of Naceur (2003) and Molyneux and Thornton (1992). However, the negative impact of cost inefficiency on performance has found by Dietrich and Wanzenried (2011), Athanasoglou et al. (2008) and Bourke (1989). Labor productivity is positively and significantly impact on bank performance. The higher level of labor productivity indicates a higher level of bank efficiency and which leads to higher profit. The findings are supported by the study of Zheng et al. (2017). The non-traditional activity (off balance sheet items to total assets, OFBSTA) implies negative and significant impact on profitability which means as off balance sheet items increases then performance decreases. The results is consistent with Rahman et al. (2015), but the findings of Apergis (2014) has found positive impact of non-traditional activity on bank performance. This study finds that income diversification (INCD) has a significant and positive impact on performance which indicates the more diversified of a bank faces less risk and increases performance. The findings are consistent with Jiang et al. (2003b), but opposite of the findings of Tan (2016), Gischer and Jüttner (2001) and Demirgüç-Kunt and Huizinga (1999). The findings of this study provide that Market power (CONC3) i.e. the concentration ratio is positively and significantly impact on the bank performance which supports the market SCP hypothesis (Tan, 2016).

Table IX reports the empirical results of the equation two where six model uses for the regression analysis. Table IX provides the impact of bank capital on the risk of the Bangladeshi banking sector. The first three models (1-3) depict the simultaneous effect of bank capital (CAR) and performance (ROA before tax, ROAB) on three measures of risk (default risk, LNZSCORE; credit risk, NPLTL; and overall risk, LLPNIR). The second three models (4-6) show the simultaneous effect of bank capital (ETAR) and performance (ROA before tax, ROAB) on three measures of risk (default risk, LNZSCORE; credit risk, NPLTL; and overall risk, LLPNIR).

The regression findings indicate that the lag dependent variable (LNZSCORE_t−1, NPLTL_t−1, LLPNIR_t−1) is positively and statistically significant with the risk across all the six models. Thus, it is evidence that there is a persistence of risk from one year to next year.

The study highlights that the main independent variable bank capital (CAR) has a significant and negative impact on bank risk using all the three types of risk measures across the first three models (1-3). Thus, the higher the CAR indicates lower the bank risk. The findings of the study are in line with Soedarmono and Tarazi (2016), Lee and Chih (2013), Agoraki et al. (2011), Zhang et al. (2008), Hussain and Hassan (2005), Jacques and Nigro (1997). The findings are contrary to the studies of Bitar et al. (2016), Lin et al. (2013), Altunbas et al. (2007), Lin et al. (2005), Bichsel and Blum (2004), Rime (2001), Blum (1999) and Shrieves and Dahl (1992).

The shareholder’s ETAR is negatively and significantly impact on bank risk using all the three types of risk measures across the last three models (4-6). This result indicates the higher the capital ratio tends to lower the risk. This finding is supported by the studies of Bougatef and Mgadmi (2016), Hao and Zheng (2016) and Tan and Floros (2013). However, the positive relationship between ETAR and bank risk finds by the study of Bitar et al. (2016).

The study also provides evidence that bank performance negatively and significantly affects the bank’s risk taking behavior across all models, which indicate that the higher the performance of a bank, the lower the risk and vice-versa.

Among the control variables, cost inefficiency (OETA) is positively and significantly impact on risk across all the models which indicate higher the operating expenses leads to high risk. The findings are consistent with the study of Baselga-Pascual et al. (2015). Deposit (DPR) is positively and significantly affects the risk across all the models which indicate higher the deposit leads to higher risk. This result is supported by the study of Ashraf et al. (2016a) and Soedarmono and Tarazi (2016). Leverage (LEV) is positively and significantly impact on risk across the models three, four, and six which indicates the higher the liabilities leads to higher risk. Bank lending interest (BLINT) is significantly and negatively associated with risk except for model-3 which means higher the percentage of lending rate lower the risk. This result supported by the study of Geng et al. (2016). The findings indicate that annual inflation rate (INFLR) has a positive and significant impact on bank risk across all the models which mean that as the rate of inflation increases in an economy, the risk also increases. The result is consistent with the findings of Arpa et al. (2001), but Tan and Floros (2013) have found an insignificant negative impact of inflation on bank risk.

Robustness checking indicates check the robustness of the main study findings by using different techniques such as changing the dependent variable with an alternative measure, changing the methods of regression analysis, changing the alternatives control variables. In this study, I have used 2SLS methods instead of GMM to check the results effectiveness. Table X shows the robust results of the impact of bank capital on performance using 2SLS with random effects model. The robust finding of the study indicates that all of the variables are significant except GGR. Table XI shows the impact of bank capital on risk using the 2SLS methods with random effects. The study finds that all the variables are significant and the results largely remain same with the main findings.

After the implementation of BASEL I, BASEL II and BASEL III in the international arena, the empirical research on Bangladesh failure to give the answers to the following questions: What is the impact of bank capital on the bank performance? andWhat is the impact of bank capital on the bank risk taking behavior? To the best of my knowledge, little attention has given by the researcher to give the answer of these questions on Bangladesh. Thus, to search the answer of the above questions, this study will be a complement study of other studies and add value to the existing literature.

This paper examines whether bank capital affects to the bank performance and risk of Bangladeshi Banking Sector. To achieve the predetermined objectives, the study uses an unbalanced panel data set of 30 banks from 2000 to 2015 yielding 413 bank-year observations. The study has considered two capital measures of bank, namely, CAR and total shareholder’s ETAR; ROA before tax has considered as performance measurement tools; three measures of risk such as default risk (natural logarithm of score), credit risk (non-performing loans to total loans) and overall risk (loan loss provision to net interest revenue) have been considered. The study also has considered some bank control measures, and industry and macro-economic factors to see the effect of those factors on capital. Liquidity ratio, cost inefficiency, labor productivity, off-balance sheet items to total assets ratio, income diversification, market power (concentration ratio) and economic growth (annual GDP growth rate) have been considered as significant determinants of bank performance. Bank cost inefficiency, deposit ratio, leverage, bank lending interest rates and the annual inflation rate has considered detecting the effect of those factors on bank risk. Using two-step system GMM, the regression analysis results solve the endogeneity issues, serial correlation problems, heteroscedasticity issues and other unobserved bank-specific issues. The study findings suggest that bank capital is positively and significantly affect the bank performance which means the higher the capital of a bank leads to higher performance. The regression results also indicate that bank capital has a significant and negative impact on bank risk which means the higher capitalized banks have a lower risk. The study also has observed that the simultaneous result indicates a significant and negative impact of risk on performance and vice-versa. Among the bank control, and industry and macroeconomic factors, cost inefficiency, labor productivity, income diversification and market power have significant and positive impact on bank performance. Bank off balance sheet items and liquidity also significantly affect the performance, but the study has found a negative impact on performance. The study has not found any significant impact of economic growth, i.e. annual GDP growth rate on bank performance. The regression result also displays that cost inefficiency, deposit ratio, leverage and the annual inflation rate is positively and significantly impact on bank risk. Bank lending interest rate also has a significant impact on bank risk but has a negative impact on risk. This investigation has also found that some of the state-owned banks were not maintained minimum capital requirements as per BASEL Capital Accord, and not maintained enough provisions against bank risk exposures.

The main objectives of introducing the BASEL Capital Accord I, II and III are to strengthen the bank capital position to maintain financial stability through reducing risk and improving performance. This investigation also aims whether this study supports with the objective of BASEL regulations. The study finds that capital is a significant and positive factor for the bank performance and reduces bank risk of Bangladesh which supports the initiatives of BASEL regulations as well as the Central Bank of Bangladesh policy decisions to implement the BASEL guidelines. Thus, the study suggests implementing the BASEL III as soon as possible. As the higher capital ratio increases performance and reduces risk, so the banks of Bangladesh should maintain the minimum capital requirements as per BASEL guidelines and Bangladesh Bank should have to set strict decision to follow the BASEL requirements. Cost inefficiency (Operating expenses), labor productivity and income diversification significantly and positively affect the bank performance. Thus, the bank policy maker should consider those factors properly. Liquidity also significantly and negatively impact on performance which indicates more loans to customers more profit generation but at the same time bank should aware about the ability and project selection of the customer. The study finds a significant negative relation of off-balance sheet items on bank performance, so, the bank should be aware of off-balance sheet items. Specially, the Government of Bangladesh should give more attention to the state-owned banks properly, as this investigation finds those banks were not maintained capital adequacy properly and also faces negative net revenue and negative provisions against loans. The government also should take necessary steps to reduce inflation, as the study finds a significant and positive correlation between inflation and bank risk. Each bank should report the capital adequacy as per BASEL requirements and should also measure and report the different measures of risk and performance results in the annual report. Finally, the study suggests that higher capital ratio leads to a bank to set up careful investment strategies through reducing risk and improving performance.

Although this study is the unique study for examining the impact of bank capital on bank performance and risk, the study faces limitations as well. The future researcher may consider the following avenues for the further study by considering the limitations of this study: first, this study has considered 30 sample banks on Bangladesh; the future researcher may consider more sample banks. Second, this study has taken data during 2000-2015, in future the latest data period may consider. Third, this investigation has been done only on one country named Bangladesh; the future researcher may include other countries which are similar nature of economic and political conditions like South Asian or African countries. Fourth, the study has considered one measure of performance, two measures of capital, and three measures of risk; in future, other measures of performance such as ROA after tax, return on average assets, return on average earning assets, ROE, NIM, earnings per share, Tobin’s Q, etc. may be considered. The future researcher may also consider other capital measures such as tier-1 capital to risk-weighted assets, tier-2 capital to risk-weighted assets besides total regulatory capital to risk-weighted assets. The other measures of risk such as market risk, liquidity risk may consider in future. Fifth, the future studies may include more bank-specific, industry and macro-economic factors besides this study such as taxation, cost of intermediation, management efficiency, reserve, bank size, unemployment rate, Herfindahl index of industry concentration and revenue diversification, Lerner index of industry competition, banking sector development measure, stock market development measure, etc. Sixth, this investigation includes three main variables capital, risk, and performance; in future efficiency, competition, politics and globalization factors may add as main variable. Finally, this study has used GMM and TSLS with Eviews software for the regression analysis; in future, other best-fit methods with the best software may consider. Also, regardless of limitations, this investigation predicts that the empirical findings of this study will be beneficial for the future researcher and policymaker to develop the theory on the link between bank capital, risk and performance.

This paper has been adopted from the author’s unpublished Master thesis. The authors acknowledge the very constructive insights from the thesis committee members, reviewers of the thesis and to the Supervisor Dr Xiaojing Li of the Donlinks School of Economics and Management at the University of Science and Technology Beijing on earlier versions of this paper.