Examining the effectiveness of loan portfolio management models on the performance of commercial banks in Zimbabwe

Abstract

The United States of America experienced phenomenal bank failures in the 1920s, which spilled over to other continents such as Africa and Asia. In Africa, Kenya, Nigeria, South Africa and Zimbabwe were some of the countries hardest hit by the bank failures. The main cause identified in all the instances, was an increased number of non-performing loans. Research in Zimbabwe has shown this problem became acute after the country obtained independence in 1980, persisting even though the Central Bank of Zimbabwe has tried to introduce several interventions on lending policy. At the centre of the lending systems of Commercial Banks are Loan Portfolio Management Models such as the Credit Score Model; the Z-Score Model and the Asset-Based Model, whose effectiveness have been examined in this study. The overarching study aim is to establish how effective these loan models are in ensuring sustainable performance of Commercial Banks. To investigate these issues, a mixed method, sequential explanatory design was employed in investigating loan model effectiveness. The initial quantitative phase of data collection (survey) made use of a structured questionnaire with a 5-point Likert scale. Using a stratified sampling technique, data were gathered from 406 participants employed by 14 Commercial Banking Institutions in Zimbabwe. This captured data were analysed in SPSS version 24.0 and Analysis of Moment Structures (AMOS) v 24.0, to yield descriptive and inferential statistics. In the qualitative phase of the research approach, a structured interview was then employed to appreciate better and seek clarification of outcomes from the survey, by interviewing both ex-bankers and current chief risk officers. Structural Equation Modelling provided estimates of the strength of all hypothesised relationships, where necessary, whilst Novel was used to analyse qualitative data. The findings revealed the Credit Score model has been adopted by most banks, whilst the Z-score model was the least used model. Results also showed, in some instances, the features of a loan model impact effectiveness throughout the loan cycle. Furthermore, nonperforming loans happen because of deficiencies of loan model parameters; characterised by outdated models, poor markets consequently insignificant cash flows generations, as well as inappropriate use of a loan model. This work has therefore, brought to light how loan models are perceived in the Zimbabwean banking sector. A negative relationship was found between credit risk theories and some loan models. The implications of the findings to Commercial Banks include that loan models should always be reviewed to address credit risks brought by the prevailing business environment. Additionally, use of current data must be emphasised, as well as employing modern models, such as neutral networks, which have proven effective in the developed world. The study’s newly developed model, with an 87.87 percent accuracy rate, can assist banks to eliminate potential bad debts at onset and improve loan quality in the banking sector. The regulator is also expected to share important credit data, which impacts performance of loan models, as well as offering technical assistance to capacitate banks with lending skills. Moreover, the regulator should ensure pronounced policies are aligned to the macro-economic environment and long-term, in order that loan models are not paralysed. The government may motivate those banks with quality loans through tax breaks and ensuring GDP is strengthened, which will improve performance in all sectors, thus, mitigating NPLs. An area that should be further investigated is whether a relationship exists between a loan model’s effectiveness and the number of years the bank has been in existence. The key contribution to knowledge made by this study is, to the best knowledge of the author, the first study to simultaneously test the impact of the three most used models by Commercial Banks and determine that the Credit Score model has the capacity to meet a bank’s requirements in identifying credit risks, particularly in a developing nation. Further to this, the study identified that parameter attributes of a loan model are key in measuring its effectiveness. The study successfully developed a loan model that can be used by a commercial bank through only applying relevant variables, which influence prediction of loan performance. In addition, structural equation modelling was validated as a most robust statistical technique for use in testing hypothesis skewed to loan models. Moreover, a key notable contribution is that credit risk theories only account for borrower characteristics, hence, there is a need to modify them by incorporating external factors, especially macroeconomic factors.