Deep Credit Analytics Lab
Conventional credit rating/scoring by credit agencies is insufficient to meet the expectations of credit analysis in the digital era. We see dynamic credit portfolio analysis in ever-changing economic environments as continuously and consistently aggregating individual credit exposures into a lender's credit portfolio for various application horizons (https://nuscri.org/en/home/).
Moreover, alternative data collected from digital media and other sources can enhance our ability to discern credit risks particularly for MSMEs (micro, small and medium–sized enterprises).
Defaults are rare events. Sharing credit information on MSMEs among lending institutions can significantly improve a model’s quality. However, data privacy must be respected. Our lab in collaboration with CriAT has developed iCASS, a new "federated learning" system that can robustly calibrate credit models over distributed data sites subject to network latency and occasional local site failures.
- Evergrande's credit crisis peaks as repayment pressures pile on and refinancing woes deepen
- Credit risk for Russian banks skyrockets in the face of unabating sanctions
- Rising geopolitical tensions between Russia and Ukraine drive up credit risk for Russian companies
- Major Australian banks are well-positioned to navigate potential headwinds in domestic housing market (PDF)