Skip to Content
👋 Hey there! Welcome to the Next.js Starter. Check it out
ICSNEWLevel 1 and Level 2 texts7. Capital requirement – Other Methods7.3.4 Additional considerations

7.3.4 Additional considerations

7.3.4.1 Model change policy

  • L1-171. The internal model may need to change over time, particularly when it no longer captures the underlying risk in its entirety.
  • L1-172. There is a model change policy that sets out the governance requirements in relation to changes to the internal model, including internal approval, internal communication, documentation and validation of changes and implementation.
  • L1-173. The policy addresses the following:
    • a) the definition of major and minor changes;
    • b) how these are assessed; and
    • c) the governance and decision-making process for such changes.
  • L2-424. The model change policy defines how minor model changes may be aggregated over time and when a combination of minor changes is considered a major change.
  • L2-425. Major changes to the internal model, as well as changes to that policy, are always subject to prior supervisory approval.
  • L2-426. Minor changes to the internal model are not subject to prior supervisory approval insofar as they are developed in accordance with that policy.
  • L2-427. There are formal criteria and processes to assess model changes and ensure these are appropriate, these include:
    • a) impact on the ICS capital requirement and on the relevant risk charges and diversification benefits, changes in model scope;
    • b) structure and theory/methodology;
    • c) data sources and external model changes and changes in the system of governance;
    • d) validation framework;
    • e) model policies; and
    • f) model use.
  • L2-428. Model changes are appropriately documented, implemented and communicated to the GWS.

7.3.4.2 Expert judgement

  • L1-174. The IAIG has a thorough governance process concerning all inputs that are subjected to expert judgment.
  • L2-429. The IAIG grades all inputs that are subjected to expert judgement by the level of uncertainty and impact in the model, and demonstrate that they comply with it.
  • L2-430. Where expert judgment is employed, this is documented. The supporting documentation describes the assumptions used, their materiality, the rationale for the opinion, the experts involved, the qualification of the experts, the appropriateness of the judgment being made, the intended use and the period of validity.
  • L2-431. Expert judgment is approved at levels of sufficient seniority according to their materiality. To assess the materiality of expert judgment, quantitative and qualitative indicators may be considered.
  • L2-432. The process and the tools for validating the assumptions and using expert judgment are documented and in compliance with the expert judgment governance.

7.3.4.3 External models and data

  • L1-175. External models used in parts of the internal model adhere to the same standards as the internally developed parts of the internal model.
  • L1-176. The GWS ensures there is an appropriate understanding of the validation process for and results of external models within the IAIG.
  • L2-433. Where external data is used, the IAIG demonstrates an appropriate understanding of its limitation and adherence to the risk profile, and it is in compliance with the data and expert judgment governance.
  • L2-434. External data and expert judgement are in the scope of model validation.

7.3.4.4 Ongoing appropriateness of the model

  • L1-177. The IAIG ensures the ongoing appropriateness of the design and operations of the internal model and that the internal model continues to appropriately reflect the risk profile of the IAIG.
  • L2-435. The IAIG develops benchmarking studies that allow them to monitor the evolution of the model overtime, defining indicators that allow understanding of how the model evolved in comparison to its risks and the cumulative effect of model changes. Approximations for the cumulative effect can lead to appropriate results such that it may not be necessary for the IAIG to calculate the capital charges using different development stages of the internal model. The indicators are not limited to ICS standard method.
  • L2-436. The persons who effectively run (Executive and Senior Management) the IAIG and staff who run the internal model receive appropriate training on the IM periodically. They are able to demonstrate a sufficiently detailed understanding of the parts of the internal model used in the area for which they are responsible.
  • L2-437. The IAIG ensures that the teams involved with the IM are adequately staffed and that the model knowledge is maintained over time.
  • L2-438. The rigour of the validation process is maintained over time and ensures that the standards under which the IM was approved are maintained. The validation process includes regular internal model reviews post-approval to assess whether the internal model deviates from the assumptions, portfolio characteristics, structure or parameterisation used in previously approved versions, and ad- hoc reviews in case such deviations have been identified. In addition, the validation process ensures that the IM continues to be perceived as an industry standard or perceived best practice. The IAIG informs the GWS in due time if it becomes aware that the internal model may not be longer compliant with the requirements imposed by the GWS.

7.3.4.5 Non-insurance entities

  • L2-439. For financial non-insurance entities with a sectoral capital requirement, the capital requirement should be consistent with the equivalent requirements under the ICS standard method (L2-336). For other non-insurance non-financial entities, the IAIG can use an internal model if it can demonstrate that the risks being run are covered and validated within it or, failing that, the IAIG can use a partial internal model with the approach specified under the standard method for non-financial entities.

Last updated on:
7.3.3 Criteria for internal model approval7.3.5 General provisions on the use of partial internal models (PIM)