Summary
Small and Medium-sized Enterprises make up over 90% of businesses worldwide, yet access to sufficient financing options remains limited and challenging, necessitating innovative lending solutions better-tailored to SME data and needs.
This research proposes a novel approach to cashflow analysis that leverages Open Banking data to support SME financing mechanisms. The methodology integrates embedding and clustering techniques to create dynamic, real-time, and forward-looking assessment tools for SME cash positions.
The approach serves as an early warning system for potential financial distress and facilitates proactive interventions and informed data-driven decision-making for lenders. By employing sentence transformers for bank transaction description embedding and clustering for monthly transaction segmentation, early results uncover clusters of distinct spending behaviors with forecasting approaches that outperform initial baselines on empirical SME bank transaction data.
Authors: Brandi Jess (SME Capital, University of Warwick), Pietro Alessandro Aluffi (SME Capital, University of Warwick), Marya Bazzi (sea.dev, University of Warwick, The Alan Turing Institute), Matt Arderne (sea.dev), Daniel Rodrigues (SME Capital), Kate Kennedy (SME Capital), Martin Lotz (University of Warwick)