Emerge's credit and other origination risk management models, like insurance, fraud and pricing models form part of selling new products.
The effective provision of credit to the individuals and businesses is a central pillar of any economy. However, the provision of credit is fraught with risks, particularly in the South African economy where indebtedness is at historic heights. Even though credit bureaux make generic credit scores available, credit providers usually benefit materially from the development of their own bespoke credit scorecards that take the particular features of the credit providers’ operations into account.
Emerge Analytics was tasked with developing a new credit scorecard using machine learning for a well-established financial institution. It was noted that for the specific tranche of debt under study, the financial institution was in fact substantially loss-making over several previous years. The new scorecards developed by Emerge Analytics were shown in back-tests to convert the loss-making tranche of loans into a profit-making tranche. For the tranche of loans in question, the use of the Emerge scorecards converted an embedded value loss of more than R250 million to an embedded value profit of approximately R20 million.
The management of risk in financial services where machine learning can be effective is not limited to credit risk. The methodology can also be used to predict other risks like fraud and even claims likelihood and impact for Insurance companies.