Capital Adequacy, Asset Quality and Bank Performance in Bangladesh

Abstract

Banks are depository financial institutions connecting the savers and users of fund. These mediators are interpolated between the final borrowers and lenders allowing them well organized allocation of funds in the economy. Entities having excess funds can advance them for rational return to entrepreneurs and other economic units who need funds to take the advantage of economically and financially feasible investment ventures. The presence of financial markets and financial organizations allows such transfer of financial resources. Thus, both the borrowers and lenders are well off compared to without financial organizations and intermediaries. It is argued that financial establishments have a progressive role in funding and investment in a multidimensional practice linking the difficulty of numerous interconnected and inter-reliant factors of differentiated nature. It is difficult to assess the contribution of each factor independently. The pivotal purpose of financial organization with other non-depository institutions is to support in the distribution of country‘s scarce capital among several alternative investment areas. Thus, the financial market plays a twin role, providing numerous types of investment fund and disciplining businesses, which are incompetent and fail to follow profitable income objectives. Thus, it is observed that financial institutions especially commercial banks if rightly organized and directed can help expansion of our economy. In the context of Bangladesh has an option, efforts to be designed at nurturing banking activities for accelerating the economic wheel of the country. Notwithstanding its significant merits, it is also not desirable to overlook the problems of the nation‘s crisis oriented banking structure, which requires appropriate guideline of the banking procedure in order to safeguard effective use and watching of business funds. i The appearances of a non-performing supervisory structure are evident in the banking area of Bangladesh. Because of the inefficient and corrupt-ridden banking structure, there was fear that a huge part of the bank credit would turn into classified and defaulted loans. The prevailing extensive spread of default culture has to increase the costs of financial intermediation by banks and financial institutions revealed in recent years. The motives for this default culture on a huge scale are also intensified by politically motivated and influenced credit provided by public sector loans and given to sponsor-director by private sector banks and due to the flaw of legal and organizational arrangements for defaulted and outstanding loan recovery. These aspects badly influence the financial segment and its setting for successful operation for achieving the desired goals. Therefore, the flow in credit distributions need to be disciplined to avoid more worsening in the financial and banking sector for upholding quality of commercial bank advancing by developing the organizational setting and that is considered as one of the prime apprehension of banking sector. Over the previous years, varieties of theories and different analyses have appeared in bank management arena. These developments viewed that bank management issues to be determined by extensive range of aspects i.e., profitability, capital adequacy, asset quality and other related factors. In this thesis studies related to the above mentioned areas have been thoroughly analyzed and discussed in a sequential manner. The sequence of analysis suggests that the results are diverse in nature reflecting the models and methodologies used in different countries and dependent on the financial and regulatory structure of the countries under study. ii In this thesis an attempt has also been made to give an insight into the different types of banks operating in Bangladesh, their performance with respect to profitability, return on asset, return on equity, classification of loans and capital adequacy under the different structural settings and different rules and regulations at domestic and international settings. An insight into the above mentioned areas of banking system of Bangladesh revealed that bank profitability, return on equity, return on asset and capital adequacy ratios have wide variations during the study period. Moreover, it is documented that non-performing loan has been increasing trend over the years and capital adequacy ratio of different banks are not uniform. There exists gap between the regulatory requirements and the amount of capital maintained by banks. Additional dimension has also been observed with the introduction of Basel in the banking sector of Bangladesh. Diverse results have been observed for the capital base during the pre-Basel and post Basel era. We are optimistic that the country‘s banking sector will able to overcome the existing problems in the banking sector with the introduction of different government rules and regulations and efficient governance measures in the banking sector. In the ‗Research Design‘ chapter of the study, at first, strands of literatures on research philosophies, research designs, research approaches, and research methods were discussed. Then the researcher specified the chosen research methods, and also provided the rationale behind such choices. The researcher then focused on the data collection mechanism, sampling framework and major choices made during the data management process. At the later segment of this chapter, the ethical dilemmas related to data collection phase were highlighted. In the conceptual framework segment, the econometric challenges related to multiple regression analysis were presented. The researcher has also discussed how those challenges were managed. iii The researcher followed positivism as the chosen research philosophy. As per the positivism philosophy it is believed that there exists only one state of reality at a given point in time; respondents‘ cognitive biases do not affect the decision-making process; and researcher can objectively detach him/her from the research process. It was a secondary data driven study and the researcher did not collect primary data through FGD, survey or interview. So, the collected data was free from standard survey biases and the research results were not subject to social scientist‘s interpretation. Deduction was the chosen research approach. In a deductive approach, tentative null hypotheses are formed and these are tested using the collected data. Theory formation is not the researcher‘s objective, rather researchers try to test the empirical validity of a theory in deductive research. The researcher tested research hypotheses [constructed based on the established theory] in Bangladeshi context. In case of qualitative research, research inputs and outputs are non numeric. On the other hand, in case of quantitative research, research inputs and outputs are numeric. In case of mono-qualitative research, researchers use only one qualitative tool like interview, FGD etc. In case of multi-qualitative research, researchers use only more than one qualitative tool. In case of mono-quantitative research, researchers use only one quantitative tool like descriptive statistics, regression etc. In case of multi-quantitative research, researchers use only more than one quantitative tool. A number of quantitative methods were used in this thesis. It was basically a multi-method quantitative business research. Research designs are of different types – archival research, case study, focus group discussion, survey etc. Survey, interview, and focus group discussion etc. are popular ways to collect primary data. Case study, and archival research etc. are popular ways to collect secondary data. This research is based on archival research. The data depository used in the iv research is based on an in-house constructed excel template. As already mentioned, the researcher used secondary data for this research. The data was collected from annual reports of Bangladesh-based commercial banks. Since annual reports are available in the public domain there is no need to seek for prior permission. Macro-economic data was downloaded from Bangladesh bank website. In a standard survey and interview-based research, generally researchers face a number of ethical concerns. Participation in the survey and interview needs to be voluntary; there should not be any discrimination based on gender, income level, and religious belief; and participation of the survey respondents should remain anonymous. Since it was secondary data-based research and the data was collected from a publicly available source with no pre-extraction and post-extraction clauses, the researcher fraught limited number of ethical challenges while conducting the research. The researcher employed purposive sampling [also known as judgemental or subjective sampling] for to select traditional commercial banks. In a purposive sampling, all the economic unit does not have the equal chance to be selected; so basically, it is a non probability-based sampling. The research time period spanned from 2011 to 2023. The constructed panel database had both cross-sectional and time-series variations – these variations were later exploited while building models. All the baseline regressions were run under the OLS (ordinary least square) framework. Before running the regressions, the researcher ensured that the pre-conditions [linearity in parameter, random sampling, consistency, no full rank issues] were met. The researcher went through the empirical literatures and then selected the set of independent variables. That is why, the researcher believes that the omitted variable concern is partially mitigated. Instrumental variables were used to check out whether ‗reverse causality channel‘ is a valid v source of endogeneity into the model or not. Robustness of the estimated effects were evaluated against measurement issues and heterogeneity concerns. The researcher has used multiple definitions of independent variables to mitigate proxy variable measurement issues. Robustness of the estimated effects were also tested by splitting the dataset into two portions [70% and 30% split-up]. Since the researcher dealt with panel data, there were two choices before the researcher – either to run a fixed-effects model or a random-effects model. As per the Hausman test result, the researcher used fixed-effects model [firm-fixed effects] in case of every specification. Moreover, fixed-effects model is more flexible with its treatment related to cross-sectional heterogeneity. In the ‗Empirical Analysis‘ chapter of the thesis, the researcher has at first presented the baseline regression results to better understand the profitability determinants. The sign and the magnitude of the regression coefficients were the key area of interest. Commercial bank‘s profitability was defined through three perspectives – accounting profit (measured through ROA), economic profit (measured through residual income), and market‘s perception of profit (measured through CAPE). It was evident that business size, activity mix, cost management, interest rate, GDP growth rate, asset quality, net interest margin positively influenced the accounting profitability of commercial banks. It was also evident that capital adequacy and inflation rate negatively influenced the accounting profitability of commercial banks. For the first baseline regression equation, most of the regression coefficients were both economically and statistically significant. It was evident that business size, activity mix, cost management, interest rate, GDP growth rate, asset quality, net interest margin positively influenced the economic profitability of commercial banks. It was also evident that capital adequacy and inflation rate negatively influenced the economic profitability of commercial banks. For the second baseline regression equation, vi most of the regression coefficients were both economically and statistically significant. It was evident that business size, activity mix, cost management, interest rate, GDP growth rate, asset quality, net interest margin positively influenced the market-based profitability measures of commercial banks. It was also evident that capital adequacy and inflation rate negatively influenced the market-based profitability of commercial banks. For the third baseline regression equation, most of the regression coefficients were both economically and statistically significant. In the baseline models, fixed effects models [firm-fixed effects] were run. The choice of firm-fixed effects was inspired by Hausman-test results and the conceptual flexibility embedded in the model. The estimated effects are robust to model preference as the regression sign does not flip and level of significance does not change when the researcher use random effects model. Baseline regression results were extracted based on OLS [ordinary least square] framework; OLS is a special case of GLS and its applicability is certainly quite limited. Most of the inbuilt assumptions of OLS are not realistic like linear in parameter, homoscedasticity etc. Regression errors are normally distributed only under some very specific circumstances. That is why it was important to test whether the regression results hold if different estimation techniques are employed. Similar types of results can be extracted if MLE or GMM estimation techniques are introduced instead of the OLS framework. The estimated effects are robust to out-of-the-sample contexts as the regression sign does not flip when the researcher built the model using 70% data and later tried to predict the remaining 30% data using the estimated model. The researcher did not observe any signs of cross-sectional heterogeneity in the estimated effects. Based on commercial bank‘s size [measured by asset value], commercial banks were divided into vii two groups: large-size banks and small-size banks. Profitability determinants in case of large-sized banks were not different from that of small-sized banks. The researcher then managed endogeneity concerns revolving the baseline results. Endogeneity in the baseline regression can stem from – omitted variable bias, reverse causality channel and measurement error in the independent variable. Omitted variables become part of the regression error and this error can be correlated with the set of independent variables – resulting into endogeneity. Similarly, presence of reverse causality and measurement error in the independent variables would make regression errors strongly connected with the error – resulting into endogeneity. After reviewing the literature, the researcher has identified a number of bank-specific and macro-factors that may have influenced commercial bank‘s profitability. There are at least three aspects of profitability namely liquidity management, management efficiency, and labor efficiency which were omitted from the baseline models. It was evident that asset quality and banks‘ performance is positively related and the regression coefficient is significant in case of all the baseline regression models, once the omitted variables were introduced into the models. It was also evident that capital adequacy and banks‘ performance is negatively related and the regression coefficient is significant in case of all the baseline regression models, once the omitted variables were introduced in the models. As already mentioned, measurement errors in the dependent variables cannot lead to endogeneity problem, but it can increase the variance of the estimators. There are three dependent variables used in the baseline model namely ROA, economic profit and CAPE. In order to mitigate the inflated variance concerns, the researcher used alternative measurement for all these three variables. Instead of using ROA, the researcher used ROE; instead of economic profit, the researcher opted for scaled residual earnings [scaled by bank-level interest income] and instead of 3-year viii moving average based CAPE, the researcher used 5-year moving average based CAPE. It was evident that asset quality and banks‘ performance is positively related and the regression coefficient is significant in case of all the baseline regression models, once the alternative definition of dependent variables were introduced into the models. It was also evident that capital adequacy and banks‘ performance is negatively related and the regression coefficient is significant in case of all the baseline regression models, once the alternative definition of dependent variables were introduced in the models. Chow tests are usually used to look out for structural change or shift in paradigms in case of time series data. The researcher has used Chow tests to investigate whether there exists any structural change in terms of Bangladesh based commercial bank‘s profitability determinants, profitability-asset quality relationship, and profitability-capital adequacy relationship. As already mentioned, Chow tests look out for structural changes in time series data, but similar techniques are applicable in panel dataset as well. It was established through the Chow test that there exists structural change in the dataset in the pre-Basel and post-Basel regime. It was evident that profitability parameters, profit-capital adequacy relationship and profitability-asset quality relationship has changed significantly during the pre-Basel and post-Basel period. Basel accord was phase-wise implemented in Bangladesh based financial sector. In order to better understand the influence of this regulation on bank‘s profitability, the researcher has used Basel dummy variable. It is evident that from the regression results that on an average bank profitability is lower during the post-Basel era than the case with pre-Basel era. The regression parameter associated with Basel dummy is negative and statistically significant. The sign of the regression coefficient makes total sense since extra equity caution was naturally supposed to depress profit numbers. It was further evident through interaction ix effects that the negative profitability-capital adequacy relationship is stronger during the post-Basel era than the case with pre-Basel era. Similarly, it was found that the positive profitability-asset quality relationship is more-stronger during the post-Basel era than the case with pre-Basel era. The substantial features and contribution of this research are provided below: i) Based on the PCA results, it was concluded that 'dimension reduction' would not be an appropriate approach to understanding the profitability determinants of Bangladesh-based commercial banks since the first two principal components can together explain only 45% of the total variation. ii) The researcher designed three regression models for the profitability of Bangladesh based commercial banks. The profitability of commercial banks was defined through three perspectives: accounting profit (measured through ROA), economic profit (measured through residual income), and the market's perception of profit (measured through CAPE). It was evident that several bank-specific, macroeconomic, and industry-specific variables influence commercial banks' profitability. iii) The regression results showed that factors such as business size, activity mix, cost management, interest rate, GDP growth rate, asset quality, and net interest margin positively impacted commercial banks' accounting profitability. Conversely, capital adequacy and inflation rate were found to negatively affect accounting profitability. iv) It was documented that factors such as business size, activity mix, cost management, interest rates, GDP growth rate, asset quality, and net interest margin had a positive impact on the economic profitability of commercial banks. In addition, capital adequacy and the inflation rate were found to negatively impact economic profitability. x v) Similarly, business size, activity mix, cost management, interest rates, GDP growth rate, asset quality, and net interest margin were also observed to positively influence market based profitability measures for commercial banks, while capital adequacy and inflation rate had a negative effect. vi) The estimated effects are robust to ‗Omitted variable bias‘, and ‗Reverse causality‘ concerns. It was observed that the estimated effects are robust to model preference as the regression sign does not flip and the level of significance does not change when the researcher uses a random effects model. Likewise, similar types of results can be extracted if MLE or GMM estimation techniques are introduced instead of the OLS framework. The study documented that the estimated effects are robust to out-of-the-sample contexts [using a 70%-30% training-testing split]. The researcher did not observe any signs of cross sectional heterogeneity in the estimated effects. vii) Chow test results documented a shift in paradigm in the profitability-asset quality and profitability-capital adequacy relationship when pre-Basel and post-Basel regimes are compared. It is evident from the regression result that on average banks‘ profitability is lower during the post-Basel era than the case with the pre-Basel era. It was further evident through interaction effects that the negative profitability-capital adequacy relationship was stronger during the post-Basel era than the case with the pre-Basel era. Similarly, it was found that the positive profitability-asset quality relationship was stronger during the post Basel era than the case with the pre-Basel era. Further study can be undertaken in order to better understand the nexus among bank profitability, asset quality, and capital adequacy with respect to other regulatory shocks like Basel implementation.