Predicting business failure in a South African business bank

Abstract

The aim of the study is to understand which factors cause business failure in a South African business bank and how can business banks successfully retain business banking customers with a probability of business failure by using a customer retention strategy and a predictive model. Business failure has been a topic for research projects across different industries such as hospitality, fishery, mining and mobile companies. Only a few studies have focused on business failure in South Africa relating to the failure of business banking customers and how can business banks effectively assist their customers by offering them services to help their business needs through the use of big data, customer value management, customer life cycle and analytical tools. There have been discussions on how to use analytical tools and statistical methodologies to help predict and detect business failure across different industries in order to help retain businesses that show a probability of business failure. With the availability of big data and analytical tools, there is also the challenge of data quality, data integrity and data access. As business banks’ data generally are situated across different servers and warehouses, it requires the data to be merged from different warehouses and be put into a sensible format, which is a complex process. A logistic regression model is used in the study to help predict business failure; it uses a methodology that has a dichotomous binary dependent variable that is recorded as either a zero or one, where one is true for business failure and zero is false for business failure. In a South African business bank, whenever a business banking customer’s business fails, the loss of time, cost and effort in managing that customer is absorbed by the bank. This then affects the country’s GDP target and the National Development Plan, which is to develop entrepreneurs and to grow the economy. Business failure increases the unemployment rate of the country, as employees will be retrenched because the business would not be sustained. Through a customer retention strategy, business banking customers will be provided with products that meet their business needs, be advised on their business’s financial positioning and be given support through the bank’s entrepreneurship programme, which is generally given to business banking customers at no cost. The study will show the effectiveness of the use of data analytics and statistical tools in solving banking problems and deriving solutions based on informed decisions through strategic data usage. Contributions of the study are as follows: • Some business failure factors could be determined using business banking customer data. • A logistic regression model can be used to predict business failure. • A customer retention strategy is proposed to help retain business customers that show signs of business failure. • Text mining was used in the study to determine the industry the customer is in, as some of the business banks standard industry classification codes were incorrect, therefore, text mining was used to confirm the industry of the business customer using the business name.