The Agency for Regulation and Development of the Financial Market (ARDFR) announced that it has begun implementing automated supervisory tools for banks based on machine learning and artificial intelligence algorithms. This is stated in the regulator's first report on the assessment of model risk management in Kazakhstani banks. This was reported by Qazaqyia.kz citing Kursiv Media.
All this is aimed at building a full-fledged model risk management system, which will require significant organizational and financial resources from banks.
According to the report, the new requirements are planned to be introduced in stages. This approach, according to the regulator, will create a sustainable system that will identify, assess and limit model risks at all stages of bank model operations.
The monitoring covers all key models used by banks, including tools for assessing credit, market, operational and liquidity risks. Models used in making lending decisions are also subject to control.
As explained by the agency, the new system should help identify cases where models operate incorrectly, are used without proper control, or lose accuracy over time. If violations are detected, the regulator will be able to respond promptly.
The ARDFR noted that the first supervisory assessment showed that model risk management in Kazakhstani banks is still at the formation stage and requires further development. At the same time, the regulator emphasizes that this is a relatively new area for the country's financial sector.
According to the agency's assessment, model risk directly affects the financial stability of banks, the volume of reserves and capital adequacy. Therefore, managing such risks is considered one of the key elements of risk-based supervision.
In the future, the results of the assessment will be used in planning new bank inspections, developing methodological approaches and improving regulation in the field of model risk management.
Model Risk Management (MRM) is a system of control to ensure that mathematical, financial, scoring, forecasting and other models do not lead the company to erroneous decisions.
Simply put: if a bank, insurance company or fintech uses a model to make decisions, there is always a risk that the model is wrong. Model risk management is needed to identify, measure and mitigate this risk.
Earlier, Kursiv reported that the ARDFR updated the deadlines for the mortgage refinancing program for Kazakhstani citizens.