Small and medium-sized enterprise (SME) financing is a vital component of economic growth across the globe, and the need for access to capital is especially important in developing countries. However, lending markets in these countries are also the least developed, and financial institutions are often reluctant to lend money to companies without any or limited credit history. This has led to the industry naming these organizations as “thin-file” customers. The resulting credit gap that formal SMEs face is about $1.5 trillion.

This problem is especially acute in the Asia Pacific region in which 40% of the gap originates, totaling $600 billion. The Asian Development Bank reports that 74% of rejected trade finance transactions come from SMEs, and at least 36% of rejected trade finance may be fundable. This suggests that there is tremendous opportunity to address the underserved SME market, but how?

In this white paper we discuss how artificial intelligence (AI) and machine learning (ML) can unlock value from the treasure trove of data trapped in the databases of traders, banks, logistics companies, and others that could – in combination with alternative data sources – be algorithmically predictive in guiding risk management to unlock SME finance. With machine learning models trained using hundreds of billions of transaction data, Flowcast delivers a game-changing approach to credit decisioning for thin-file customers, empowering financial institutions to close the significant SME finance gap.

Flowcast’s machine learning model takes a similar approach to Tensorflow’s Object Detection API. Analogous to deploying trained models capable of identifying multiple objects in a single image, Flowcast can accommodate the varying data available across companies. And, even with limited data, Flowcast’s technology can borrow from the behavioral patterns of other companies to predict outcomes, albeit with lower confidence levels.