Model Risk

    There is a growing trend for the use of various statistical models in the decision-making process in the financial sector. Such models are widely used in lending, financial accounting and risk management. The use of mathematical and statistical models in retail sector lending is especially growing. The models based on machine learning and artificial intelligence algorithms are now used more often. This increases the effectiveness of the model, but also complicates its analysis and risk assessment.

     

     

    To address this challenge, the National Bank developed the Regulation on Managing Risks for Data-based Statistical, Artificial Intelligence and Machine Learning Models.The purpose of the Regulation is to promote effective risk management of the model. It sets out the basic principles of model building, validation and application. Principles and standards for model risk management are based on current practices, challenges and advanced international supervisory experience in the financial sector. This framework will smooth the wider and more efficient use of models in the financial sector.

     

     

    All new important models of the financial institutions are now required to be constructed in accordance with the Regulation, while the currently available models must comply with the Regulation by September 2021. Models for accounting Expected Credit Losses in financial institutions are being introduced (in connection with the introduction of the IFRS 9 standard) at the National Bank. These models are high risk carriers due to the complexity of the model and the impact on the financial stance of the bank. The National Bank of Georgia will use its resources to construct a benchmark model? saorientacio to analyze these risks. The results of the analysis will serve as one of the tools to evaluate the model of financial institutions.