25. Validation

Institutions should apply the validation principles in designing, testing and maintaining IRC models. Validation should include evaluation of conceptual soundness, on-going monitoring that covers process verification and benchmarking, and outcomes analysis.

The validation process of IRC models should include at least the following principles:

• i. Liquidity horizons should reflect actual practice and experience during periods of both systematic and idiosyncratic stresses;
• ii. The IRC model for measuring default and migration risk over the liquidity horizon should take into account objective data over the relevant horizon and include a comparison of risk estimates for a rebalanced portfolio with that of a portfolio with fixed positions;
• iii. The correlation assumptions in the IRC model, including the structure of stochastic dependencies and correlations/copulas, as well as the number and weight of systematic risk factors, must be supported by analysis of 20 objective data in a conceptually sound framework. In particular, the impact of different copula assumptions should be analysed, for example by testing the impact of different distributional assumptions. The default and migration behaviour predicted by the model should be validated against actual default and migration experience of traded debt portfolios.
• iv. The validation of an IRC model should rely on a variety of stress tests and sensitivity and scenario analyses, to assess its qualitative and quantitative reasonableness, particularly with regard to the model‟s treatment of concentration. Such a test should cover both historical and hypothetical events;
• v. The validation should also cover the assessment of calibration of PD and LGD risk parameters;
• vi. An institution should strive to develop relevant internal modelling benchmarks to assess the overall accuracy of its IRC models;
• vii. The validation should assess the transparency and adequacy of documentation of the IRC modelling approach.