As credit unions continue to embrace digital transformation, the use of machine learning (ML) and artificial intelligence (AI) models has become increasingly relevant. However, managing the risks associated with these models remains a critical challenge. To gain insight into the current state of Model Risk Management (MRM) within credit unions, we conducted a survey that revealed key findings about the adoption of AI/ML models, the development of MRM frameworks, and the challenges faced by these institutions.

Adoption of AI/ML Models: A Growing Presence

The survey results indicate that AI/ML models are becoming more central to credit union operations, with 63.64% of respondents planning to develop or enhance their AI/ML strategies within the next 12 months. The most commonly used models include Member Behavior Models (e.g., segmentation and churn forecasting), which were reported by 54.55% of respondents. Other significant areas of adoption include Credit Risk Models, Financial Forecasting Models, and Marketing Models, each used by 45.45% of respondents. However, 36.36% of credit unions still do not use any AI/ML models, highlighting an area of potential growth.

Development Stages of MRM Frameworks: Early Days for Many

The maturity of MRM frameworks varies widely among credit unions. A notable 45.45% of respondents are still in the early stages of developing their MRM frameworks, while 18.18% have moderately developed frameworks that could benefit from further enhancements. Unfortunately, 36.36% of respondents indicated that they do not have any MRM framework in place, underscoring the need for more comprehensive strategies to manage model risk effectively.

Challenges in Model Risk Management: Governance and Monitoring Issues Prevail

Credit unions face several challenges in managing model risk, with the most prominent issues being a lack of clear governance and oversight (54.55%) and challenges with ongoing monitoring and reporting (also 54.55%). Additionally, 45.45% of respondents cited insufficient risk assessment and mitigation as a significant hurdle, while 36.36% struggle with maintaining an up-to-date model inventory. These challenges highlight the complexity of managing AI/ML models and the critical need for robust governance structures.

Responsibility and Strategic Planning: The Need for Clear Accountability

Interestingly, more than half of the respondents (54.55%) reported that no one within their credit union is specifically responsible for managing model risk, which could impede the development of effective MRM practices. Only 36.36% identified their Data & Analytics or IT departments as responsible for these activities. This lack of clear accountability could hinder the effectiveness of MRM efforts and highlights the importance of defining roles and responsibilities within the organization.

Key Takeaways

The survey results provide a snapshot of where credit unions currently stand in their journey toward effective Model Risk Management. While there is significant interest in leveraging AI/ML models, the development of robust MRM frameworks is still in its early stages for many institutions. Addressing the challenges of governance, risk assessment, and accountability will be crucial in ensuring that credit unions can manage the risks associated with their data models effectively.

As the financial services industry continues to evolve, it is imperative that credit unions prioritize the development of comprehensive MRM strategies that not only protect their operations but also enable them to fully harness the potential of AI/ML technologies. By doing so, they can better serve their members and navigate the complexities of the modern financial landscape.