The Central Bank of the UAE published Model Management Standards and Guidance on 23 December 2022. The Standards ( “MMS”) present mandatory modelling practices to be implemented by banks in the UAE based on international standards while considering local circumstances. The MMS outlines general standards applicable to all models as well as specific requirements for the application of the standards. The Model Management Guidance (“MMG”) expands on technical aspects specific to certain model types.

Objectives

The objectives of the standards and guidance are as follows:

• To ensure models used by UAE banks meet quality standards to adequately support decision-making and reduce model risk
• To improve homogeneity of model management across UAE banks
• To mitigate the risk of underestimating provisions and capital requirements across UAE banks

Scope of models

The MMS  applies to all types of models employed by institutions to support decision-making. Therefore it covers, amongst others, risk models, pricing models and valuation models. The MMS document includes the following non-exhaustive list of models that should be considered in scope:

Model management framework

Banks must have a model management framework that covers the following key areas.

Model governance

Banks are required to develop and maintain appropriate policies and procedures to ensure that model risk is managed effectively. The model governance framework must cover, at a minimum, the definition of model objectives, key steps of the model life-cycle, a model inventory,  model ownership, a list of key stakeholders involved in decision-making, details relating to relationships with third parties involved in model development, an assessment of the adequacy of internal skills, comprehensive model documentation, and a comprehensive reporting framework.

Data governance

Banks are required to have a formal Data Management Framework to ensure the availability of accurate and representative historic data, used in the development and validation of models. The data management framework must cover the identification of relevant data sources, data collection frequency and level of granularity, data quality control, data storage and access control protocols, and policies to ensure robust system infrastructure.

Model development

Banks must have a well-documented, structured process for model development with sequential and logical steps. This process must include the following steps at a minimum, to be performed iteratively as required:

• Data preparation
• Data exploration
• Data transformation
• Sampling considerations
• Selection of suitable model methodology
• Model construction
• Model selection
• Model calibration
• Pre-implementation validation
• Impact assessment

Model implementation

Banks must have a separate framework to govern the implementation of approved models after they have been developed. Implementation of each model must be treated as a project within a suitable project management framework. This framework must include the implementation scope, detailed implementation plan including functional and technical specifications, roles and responsibilities, a roll-back plan and User Acceptance Testing (UAT) with suitable test cases.

Model usage

Models usage must be clearly defined, well-documented, monitored and managed. A model usage framework must consider the identification of model users, the frequency of model usage, the source and nature of model inputs, the destination of outputs within the IT system and operational processes, how output must be interpreted including defining the decisions that they can support, limitations of model output including level of uncertainty, and strict governance on the use of input and output overrides.

Performance Monitoring

Banks must have a process in place to monitor the performance of models in light of changes in market conditions and the economic or business environment. Such changes may impact the stability, performance, assumptions or reliability of models. This process must be well documented and cover roles and responsibilities, frequency of monitoring, appropriate metrics and limits, reporting, and escalation processes.

Independent validation

In addition to performance monitoring, banks must have a robust process to ensure the independent validation of models to ensure that they are performing as intended. These processes must include defined responsibilities, frequency of reviews, appropriate metrics and limits, and reporting. Validation must include both quantitative and qualitative approaches and the outcome must include a clear conclusion of the model’s suitability, and the bank must have a well-defined remediation process in the event that models fail the independent validation.

Model Management Guidance (“MMG”)

As a supplement to the Model Management Standards, The Central Bank of the UAE also published Model Management Guidance (“MMG”) which expands on technical aspects specific to certain model types, based on commonly accepted modelling practices used by practitioners and academics internationally. These models include:

• rating models,
• PD models,
• LGD models,
• Macro models,
• interest rate risk in the banking book (IRRBB) models,
• net present value models (NPV)

Although the provisions in the Guidance are not mandatory, they are recommended and deviations from the MMG must be justified and subject to review by the Central Bank. The Standards and Guidance are available on the Central Bank of the UAE’s website.