Conventional Credit Assessment Method for Bank Loans
Prime Minister's OfficeSpeakers
Summary
This question concerns whether conventional bank loan assessment methodologies will be updated for changing business models, as raised by MP Zainal Sapari. Senior Minister Tharman Shanmugaratnam, on behalf of the Prime Minister, replied that while big data and AI improve risk management, core credit principles remain fundamental. He noted that unconventional data helps underserved segments but poses risks regarding data accuracy and algorithmic transparency. To address this, MAS co-developed the FEAT Principles to promote fairness and ethics, requiring firms to maintain robust governance for auditable models. Senior Minister Tharman Shanmugaratnam affirmed that MAS will adapt its supervisory approach to support innovation while safeguarding public confidence in the financial sector.
Transcript
64 Mr Zainal Sapari asked the Prime Minister whether the policy and guiding principles for bank loans based on conventional credit assessment methodology will be reviewed and updated to be relevant to changing business models.
Mr Tharman Shanmugaratnam (for the Prime Minister): MAS' regulations and guidelines pertaining to banks seek to ensure that they make sound lending decisions, which support resilient and sustainable financing of our economy and the markets that they serve abroad.
Big data analytics, machine learning and artificial intelligence (AI) are opening up new avenues for lenders to improve their risk management, as well as improve outcomes for customers across a range of business activities.
The availability of unconventional data sources, such as customers' near real-time cashflows, transaction and payments data patterns, could potentially lead to better insights on customer preferences, behaviour, and credit worthiness. Such insights could benefit customers themselves, by allowing banks to provide better tailored and useful services. For example, they could lead to more differentiated pricing benefiting customers with strong credit standing, or greater availability of credit to previously under-served segments of the market.
But the core principles of credit risk management that underpin MAS' regulations remain relevant. Proper governance and oversight, robust risk models and methodologies, independent credit assessments and reviews are some of these fundamental supervisory requirements. However, new lending practices that rely heavily on unconventional data sources or artificial intelligence do come with specific risks that banks have to address. The use of inaccurate or non-representative data could result in wrong lending decisions or pricing and unintentional biases. AI algorithms that determine lending decisions may also be harder to validate for accuracy relative to more established lending processes.
To promote the responsible use of AI and data analytics, MAS has co-developed with the financial industry a set of principles to promote fairness, ethics, accountability and transparency in the provision of financial products and services (FEAT Principles). Through these FEAT Principles, MAS is working with financial firms to strengthen internal governance around data management and use. MAS expects financial firms to put in place robust data and model governance frameworks and processes that support transparency and auditability, so that data-driven or AI models can be explained, understood and monitored, and be subject to regular reviews and validation. Financial firms are also expected to ensure that the use of AI and data analytics continues to be aligned with their corporate ethical standards and codes of conduct.
We are closely monitoring these developments in technology and business models. MAS will continue to proactively refine and adapt its regulatory and supervisory approach in a way that promotes innovation while safeguarding public confidence in our financial sector.