Identified gaps and developed a model risk policy to draft an MMS policy document
- Model Governance:
- Appropriate processes for identification, measurement, monitoring, mitigation and reporting of model risk
- Integration of Model risk with enterprise risk management, as per the Bank’s appetite for model uncertainty
- Roles and responsibilities of stakeholders across the model life-cycle are appropriately defined
- Model documentation, including model grouping and the maintenance of a model inventory, is appropriate
- Data Management processes, around identification of sources, collection frequency, data quality review, data storage and access, data infrastructure, etc.
- Model Development processes, and whether they ensure the development of the most appropriate models
- Model Implementation processes, and whether they ensure the alignment of planning, funding, and timing, UAT and other testing processes, record-keeping and documentation, suitability and integration with technological infrastructure, etc.
- Model Usage processes, around the definition and documentation of roles and access, expected usage, control of usage, governance around the override of model inputs/outputs, and user feedback management.
- Review of Performance Monitoring Processes: Assessing the processes created by the Bank for performance monitoring of models and whether they include the definition of responsibilities, frequency of monitoring, metrics and limits developed/defined to assess performance, and reporting and decision-making processes
- Independent Model Validation, and processes around ensuring the independence of model validators, vendor assessment and management, validation scope definition, and correction of identified deficiencies.
- Qualitative Validation of Models: A review of processes that assess the maintain the conceptual soundness, design and suitability of the models, and whether they include assessment of factors such as conceptual soundness and scope, mathematical construction choices, output suitability in terms of economic intuition and business sense, and statistical modelling factors (choice of variables, sampling, etc.)
- Quantitative Validation of Models: Review of processes to numerically assess the suitability of the model output with respect to the objective initially assigned to the model. The processes that we will review include those for the assessment of model performance, and whether they appropriately consider the appropriateness of factors such as accuracy, conservatism, stability, and controlled sensitivity
- Training and Awareness policies
- Record Keeping and Reporting
- Model Risk Management Technology and Tools
RAROC Model Validation
- Review of Governance and decision-making process and whether they for transparency, definition of responsibilities, and compliance with CBUAE requirements.
- Verification of the theoretical foundations and assumptions of the RAROC models
- Evaluation of risk-adjusted return calculation algorithms, verifying that the model framework accurately integrates various risk measures and financial variables.
- Assessment of whether calculations are consistent with economic fundamentals, reflecting reasonable economic behaviour, and aligning with banks’ business strategies and objectives.
- Review of Model performance assessment procedures, including Accuracy, Conservatism, Stability and Robustness & Controlled Sensitivity
- A thorough review of the model's risk measurement techniques, capital allocation frameworks and overall risk management strategies to verify that they comply with the most recent guidelines and requirements set by the CBUAE.
- Review the Bank's RAROC model documentation to assess the coverage, comprehensiveness and adequacy, etc.
- Review and assess the suitability and relevance of the approach & methodology, inputs and assumptions used in the RAROC model.
- Assess the model construct for reasonableness, alignment to industry practices, and sensitivity of key assumptions.
- Assess and recommend if RAROC thresholds can be established at business segment level.
Model Risk Management, Climate Risk Stress Testing, RAROC Model Validation, and ESG Scorecard Validation for the UAE branch of a large Bahrain-based Bank (2/2)
Climate Risk Stress Testing
Transition Risk Stress Testing – Platform Methodology
- NGFS Scenarios: The Model employed three scenarios and Integrated Assessment Models (IAMs) provided by the Network for Greening the Financial System (NGFS) that derive the impacts of different policy ambitions on the energy and transition-relevant sectors (transportation, buildings, etc.), emissions, and land use:
- Current Policies (+3.0 °C): Baseline scenario, with no or little change of current policies to combat climate change, causing high physical risks but minimal transition risk with global temperature increase by more than 3 degrees
- Net Zero 2050 (+1.5 °C): Coordinated global policy implementation (Paris Agreement) to combat climate change and limit the global temperature increase to 1.5 degrees, implying moderate transition risks;
- Delayed Transition (+2.0 °C): Delayed policy implementation creates a “Minsky Moment” with high transition risks.
- Stress Testing Approach: The Model employed both Top-Down and Bottom-Up approaches:
- Top-down analyses propose and use a top-down portfolio-level analysis at the economic sector level. No distinction is made between assets or issuers within the same sector
- Bottom-up analyses are based on issuer-specific data. The target is to process a detailed analysis of securities within pre-defined economic sectors that are subject to high transition risk.
- Outputs:
- Change in PD / Value for individual securities under each scenario
- Change in the value of the selected security portfolio
- Comparison of value changes across portfolio groups
- Portfolio value change for sectors within each of the selected regions
- Physical Risk Stress Testing – Platform Methodology
- Exposure Identification: Analysed the geographical location of assets or properties and historical records of extreme weather events using hazard indicators.
- Hazard Measurement:
- Utilised event-based hazard modelling to determine probabilities associated with various event intensities.
- Assigned "High," "Medium," or "Low" risk indicators based on hazard intensity, followed by determining the probability associated with each indicator. These thresholds are calibrated using historical hazard data for a geographical region. Probabilities are then determined for each risk indicator and calibrated with historical data to refine hazard assessment.
- Vulnerability Assessment:
- Evaluated the vulnerability of individual property exposures. This also involves assessing the impact on property value indices, which affects Probability of Default (PD) and Loss Given Default (LGD), calculated using historical property adjustment factors.
ESG Scorecard Validation
Reviewed and assessed the below using a sample of ESG scorecards:
- Studied the existing scorecard to understand its structure, parameter coverage, scoring methodology, and data sources
- Parameter Coverage: Ensured the ESG Scorecard is sufficiently exhaustive in terms of the parameters covered for each of the below-mentioned factors:
- Environmental, such as carbon footprint, energy use and efficiency, water and waste management, pollution control, environmental policies, biodiversity and land use, etc.
- Social, such as labour practices, health and safety policies, diversity and inclusion, customer satisfaction, social impact, etc.
- Governance, such as corporate governance policies, ethics, etc.
- Industry-specific factors
- Incorporated climate risk parameters in alignment with CB-UAE requirements.
- Studied the existing sc Weightage and Scoring Methodologyorecard to understand its structure, parameter coverage, scoring methodology, and data sources
- Evaluated the weighing system, considering the relevance of each parameter for the Bank’s ESG objectives and local regulations, and whether they are consistently applied across all assessment
- Evaluated the scoring system,
- by reviewing the scoring scale for each parameter to ensure it provides sufficient scoring granularity
- Mathematically assessing scoring aggregation methods used
- Assessed data sources, including the efficacy of software systems for automated data collection in case of listed entities
- Evaluated data quality evaluation for completeness and consistency
- Reviewed data management processes
- Ensured regulatory compliance and alignment of the ESG scorecards with CBUAE requirements