Consumers Financial Distress: Prediction and Prescription Using Machine Learning

This paper shows how transactional bank account data can be used to predict and to prevent financial distress in consumers. Machine learning methods were used to understand what are the most significant transactional behaviours that cause financial distress. We show that Gradian Boosting outperforms the other machine learning models when predicting the financial distress of a consumer. We also obtain that, Fees, Uncategorised transactions, Other Income, Transfer, Groceries, and Interest paid were sub-categories of transactions which highly impacted the risk to be financially distressed. The study also proposes prescriptions that can be communicated to the client to help the individual make better financial decisions and improve their financial wellbeing by not entering a state of financial distress. This research used data from a major south African bank and the study was limited to credit card clients.