On 18 Jul 2025, the HKMA issued revised guidance CA-G-4 Validating Risk Rating Systems under the IRB Approach (V.8), superseding the previous version dated 17.05.18. The guidance establishes the HKMA's expectations for Als to validate their risk rating systems for IRB approach eligibility, requiring regular validation of discriminatory power and calibration for PD, LGD, and EAD. The document emphasizes robust data management, independent validation processes, and specific methodologies for low-default portfolios where traditional validation approaches are not feasible.
This article was generated using SAMS, an AI technology by Timothy Loh LLP.
Increased Transparency at Heart of Hong Kong Regulatory Changes
Hong Kong has continued to maintain its status as a leading financial centre throughout 2016 and into 2017, and there have been a number of regulatory developme
OTC Derivatives: Proposed Hong Kong Reporting & Record Keeping Requirements
Amendments in 2014 to the Securities and Futures Ordinance introduced a new statutory regime for OTC derivative transactions. The regime establishes a framework
SPOTLIGHT ON
FAMILY OFFICES
Introduction
On 18 Jul 2025, the Hong Kong Monetary Authority (HKMA) issued the revised Supervisory Policy Manual module CA-G-4 Validating Risk Rating Systems under the IRB Approach (V.8), superseding the previous version dated 17.05.18. This statutory guideline sets out the HKMA's approach to the validation of authorized institutions' (Als) risk rating systems for the purpose of using the Internal Ratings-Based (IRB) approach to calculate credit risk for capital adequacy purposes.
HKMA's Validation Approach
The HKMA's validation approach adheres to the Basel Committee's IRB validation principles, emphasizing that validation is fundamentally about assessing the predictive ability of risk estimates and the use of ratings in credit processes. The HKMA recognizes that validation is an iterative process requiring both quantitative and qualitative elements, with the bank having primary responsibility for validation. The HKMA expects Als to conduct regular validation of their rating systems, including both in-sample and out-of-sample validation, and to establish internal tolerance limits for deviations between estimated and realized credit risk components.
Key Requirements for Rating Systems
Als must demonstrate that their rating systems exhibit economic plausibility, with risk factors well-grounded in economic and financial theory. The HKMA requires Als to validate the discriminatory power and calibration of their rating systems at least annually, using generally accepted quantitative techniques. For PD validation, Als must employ methodologies such as Cumulative Accuracy Profile (CAP), Receiver Operating Characteristic (ROC), and Bayesian error rate (BER). For LGD and EAD, Als must validate their estimation processes, including stability analysis, comparisons between internal estimates and realized outcomes, and risk differentiation analysis.
Data Management and Governance
The HKMA places substantial emphasis on data management, requiring Als to have effective systems for collecting, storing, processing, and utilizing data on obligor and facility characteristics. Als must maintain comprehensive data for rating system development, implementation, and validation, with appropriate data quality assessment conducted at least annually. The document also outlines detailed requirements for corporate governance, including Board and senior management oversight, independent validation processes, and transparency in documentation of rating systems and their operations.
Special Considerations for Low-Default Portfolios
For low-default portfolios (LDPs), where Als lack sufficient default and loss data, the HKMA requires specific risk quantification and validation methodologies. Als should consider data-enhancing techniques such as pooling data with other financial institutions, combining portfolio segments with similar risk characteristics, or using multi-year PD calculations. When back-testing is not feasible due to insufficient data, the HKMA expects Als to use benchmarking tools to validate their rating systems and credit risk component estimates, with appropriate internal tolerance limits established for differences between internal estimates and benchmarks.
View the full article:Source
