Skip to Content
👋 Hey there! Welcome to the Next.js Starter. Check it out
ICSNEWLevel 1 and Level 2 texts7. Capital requirement – Other Methods7.3.3 Criteria for internal model approval

7.3.3 Criteria for internal model approval

  • L1-158. The IAIG is responsible to ensure and provide evidence that the model complies with all the requirements in this section and the GWS should be satisfied that these requirements are met.

7.3.3.1 Statistical quality test

7.3.3.1.1 Scope, theory and structure of the internal model
  • L1-159. The scope of the internal model is complete by including all material quantifiable risks in assets, liabilities, legal entities and lines of business within its intended scope. The methods used in the internal model are based upon current and credible information and realistic assumptions.
  • L1-160. The data used is accurate, complete and appropriate. The parameterisation approach is well justified, consistently implemented, tested and documented. The documentation of the scope of the internal model covers how materiality has been assessed.
  • L2-383. The base quantitative methodology of the internal model is in line with generally accepted market practice and robust actuarial and statistical theory. More specifically, the methodology and technology implementation used for this should be perceived as an industry standard or better.
  • L2-384. The structure of the model is clear, logical and consistent with how business is managed.
  • L2-385. The IAIG explains the appropriateness of the chosen methodology taking into consideration the nature, scale and complexity of the risks.
  • L2-386. In addition, for everything within the scope of the internal model, the IAIG has demonstrated an approach to differentiate between the materiality of the risks being taken by the IAIG. The internal model allows for changes in the risk profile over time adapt to foreseeable risks, which can be achieved by parameterisation or model changes. If future management actions are considered, they follow the principles below:
    • a) Be objective, realistic and verifiable;
    • b) They cannot be contrary to the IAIG’s obligations to policyholders or legal provisions applicable to the IAIG;
    • c) Are consistent with the IAIG’s current business practice and business strategy unless the GWS is satisfied that there is sufficient evidence that the IAIG will change its practices or strategy;
    • d) Can reasonably be expected to be carried out under the specific circumstances to which they apply; and
    • e) The assumptions about future management actions take into account the time needed to implement them, as well as any resulting incremental expenses.
  • L1-161. Risk mitigation techniques may be recognised in the ICS risk charges provided they meet the requirements specified in the Level 2 text.
  • L2-387. In order to be recognised, a risk-mitigation technique should meet the following criteria:
    • a) The risk mitigation technique is effective and legally enforceable in all relevant jurisdictions and results in an effective transfer of risk to a third party;
    • b) The contractual arrangement has been reviewed to ensure that the risk transfer is clearly defined;
    • c) The calculation of the ICS risk charges allows for the effects of risk mitigation techniques through a reduction of the risk charge commensurate with the extent of risk mitigation. That reduction makes a reasonable allowance for any basis risk effects due to changes in risk mitigation assumptions and relationships including during tail events, and there is an appropriate treatment for any corresponding risk embedded in the use of risk mitigation techniques (eg counterparty default). These two effects are treated separately;
    • d) There is no double-counting of mitigation effects;
    • e) The documentation for the arrangement sets out a direct claim on the IAIG’s counterparty in the event of its default, insolvency, bankruptcy or other credit events; and
    • f) Consideration of the credit quality of the providers of the risk mitigation is made within the internal model.
  • L2-388. The methodology to calculate the ICS capital requirement are consistent with the methods to calculate the ICS balance sheet. The initial balance sheet of the internal model reconciles with the ICS balance sheet.
  • L2-389. The IAIG decides the best way to consolidate and account for the risks to the whole of its business.
  • L2-390. The determination of overall regulatory capital requirements with the internal model considers dependencies within and across risk categories. Key variables leading to dependencies must be identified and modelled, tail dependencies and non-linear effects must be adequately captured. Where the internal model allows for diversification effects, the IAIG justifies its allowance for diversification effects and demonstrate that it has considered whether dependencies may increase under stressed circumstances.
7.3.3.1.2 Assumptions
  • L2-391. The main assumptions of the model have a sound theoretical and empirical justification, including the circumstances under which the assumptions would be considered false and provide a rationale for not using alternatives.
  • L2-392. Where simplifying assumptions are used, this is clearly defined and documented.
7.3.3.1.3 Data
  • L2-393. The length of data has sufficient historical information to assess the characteristics of the underlying risks, and no relevant data is excluded from the use in the model without justification. Where historical data is deemed insufficient, this may be complemented with expert judgment.
  • L2-394. The data is accurate and consistent, free of material errors and recorded consistently over time, and is part of the validation process.
  • L2-395. The data is appropriate, fit for its purposes and reflect the relevant risks to which the IAIG is exposed.
  • L2-396. Nonetheless, the model still allows for outcomes not directly observed in past data but to which the IAIG may be exposed in the future, eg new business classes or tail risks.
  • L2-397. In addition, data extensions, capping or modification is documented and justified. The process for dealing with outliers and data-smoothing is performed in a way that does not materially underestimate the actual volatility of the risk.
  • L2-398. The data used for estimations is up-to-date, recorded on time and updated at least annually, and the IAIG minimises the gap between the end-of sample data and the calibration date.
  • L2-399. The IAIG has a data policy that is consistent with the provisions above and demonstrate that they comply with it.
7.3.3.1.4 Parametrisation
  • L2-400. The parameterisation methods are in line with appropriate actuarial and statistical practices. The latter includes an explanation for the input and the reasoning for the selection among several candidates.
  • L2-401. When parameterisation is made using expert judgment, this is in compliance with the IAIG’s policy on expert judgment.
  • L2-402. Parameterisation results are reasonable, up-to-date and validated. The IAIG defines minimum fitting requirements, such as R2, chi-square statistics, q-q plots, residual analysis, whenever possible. Whenever appropriate, the IAIG performs stress testing and sensitivity analysis when choosing model parameters. Parameter risks are taken into consideration whenever material.
  • L2-403. The parameterisation is reviewed at least once a year. In the event of material differences in the parameters between exercises, this is explained and justified.

7.3.3.2 Calibration test

  • L1-162. The IAIG may use a different confidence level, risk measure or time horizon than Value-at- Risk (VaR) at 99.5% confidence level over the one-year time horizon for internal modelling purposes as long as the outputs of the internal model can be used by the IAIG to calculate the ICS Requirement in a manner that provides policyholders and beneficiaries with an equivalent or higher level of protection.
  • L2-404. Where the IAIG uses a different confidence level, risk measure or time horizon than the one set out for the ICS standard method, capital requirement calculations provide evidence on how the model outcomes compare to the ICS target criterion (ie VaR 99.5% over a one-year time horizon).
  • L2-405. Where the IAIG cannot derive the ICS capital requirement directly from the probability distribution forecast generated by the internal model, the GWS may allow approximations to be used in the process to calculate the ICS capital requirement, as long the IAIG can demonstrate to the GWS that policyholders are provided with a level of protection at least equivalent to VaR at 99.5% confidence level over the one-year time horizon.

7.3.3.3 Use test and governance

  • L1-163. The IAIG explains the different uses of their internal model and how they ensure consistency between the different outputs where the internal model is used for different purposes.

  • L1-164. The internal model is adequately fit to the way business and risks are managed:

    • a) The modelling approaches reflect the nature, scale and complexity of the risks inherent in the activities of the IAIG, which are within the scope of the internal model;
    • b) The outputs of the internal model and the content of the internal and external reporting of the IAIG are consistent; and
    • c) The internal model is capable of producing outputs that are sufficiently granular to play an important role in the relevant management decisions of the IAIG. Furthermore, IAIG are able to run the model in a promptly and timely manner, in case model outputs are need to decision-making outside form the usual reporting timeline.

  • The model is widely used and plays an important role in the system of governance of the IAIG and how business and risk are managed.

  • L2-406. The model is used, but not limited to, for example in:

    • a) Strategic and business planning and business management;
    • b) Risk management, economic capital calculation, and own risk and solvency assessment;
    • c) Risk appetite setting;
    • d) Measurement of risk-adjusted performance and performance management;
    • e) Reporting, decision making, communication and feedback loop development; and
    • f) The internal model is sufficiently used in profit and loss (P&L) attribution.

  • L2-407. Profit and loss (P&L) attribution compares the internal model’s predicted profit and loss with the actual profit and loss incurred.

  • L2-408. The IAIG should demonstrate that the model is fully embedded in its operational and organisational structure and confirm that it remains valid and is applied consistently over time.

  • L2-409. The Board and Executive Management of the IAIG are able to demonstrate an overall understanding of the internal model including their major assumptions, strengths and limitations.

  • L2-410. The persons who effectively run (Executive and Senior Management) the IAIG and staff who run the internal model are able to demonstrate a sufficiently detailed understanding of the parts of the internal model used in the area for which they are responsible.

  • L2-411. The Board, executive and senior management receives appropriate training on the internal model.

7.3.3.3.1 Model governance
  • L1-165. Model governance structure, roles, and responsibilities are clearly defined by the IAIG and considered appropriate by the GWS.
  • L2-412. The IAIG Board is responsible for ensuring 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.
  • L1-166. The Board may delegate its responsibility for the ongoing maintenance, use, application and validation of the internal model to the risk management function, or senior management and staff, possessing relevant expertise and no conflicts of interest. The model’s governance structure is clearly defined and documented, and this includes reporting lines, allocation of responsibilities and escalation paths.
  • L2-413. The IAIG has documented procedures for appropriate segregation of duties between those responsible for building, operating and maintaining the model and those responsible for making decisions based on the model’s output.
  • L1-167. Mechanisms to prevent conflicts of interest are in place and addressed in the model’s governance framework.
  • L2-414. The resources that operate the model are adequate with respect to the nature, scale and complexity of the risks modelled.
  • L2-415. There are adequate and effective controls in place in relation to the operation and maintenance of the internal model.

7.3.3.4 Validation standards

  • L1-168. The regular validation process aims to ensure 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.
  • L2-416. The internal model validation process specifies the following:
    • a) Scope of validation;
    • b) Processes, methods, and tools to be used;
    • c) Frequency of validation; and
    • d) Persons involved, roles, reporting lines and escalation paths.
  • L1-169. The model validation process applies to all internal model parts and cover all requirements.
  • L2-417. The IAIG demonstrates that the model has been validated independently (externally or internally) from those who develop, change, update, run, and use the model.
  • L2-418. Internal model validation requires the IAIG to have model validation reports covering model components and the entire model. The reports document the validation process and conclude on the adequacy of the model component or model being validated and the appropriateness of the resulting capital charge for the regulatory capital.
  • L2-419. There is a clearly defined remediation and follow-up process for model validation findings, an action plan, and implementation monitoring.
  • L2-420. As part of the validation process, the IAIG performs at a minimum and obtain reasonable results and insights from inter alia:
    • a) Sensitivity testing, stress testing and scenario analysis, including reverse stress testing;
    • b) Back-testing;
    • c) Stability testing;
    • d) Profit and loss attribution; and
    • e) Other validation tools, such as benchmarking or alternative methods/models.

7.3.3.5 Documentation standard

  • L1-170. Model documentation allows a knowledgeable third party to understand the design and details of the internal model and form a sound judgment as to its compliance with regulatory requirements.
  • L2-421. The documentation provides a detailed description of the structure, scope, theory, data, assumptions, expert judgment, parameterisation, results, validation, model changes, model governance and model policies. Furthermore, the documentation details all key software, external models (including their customisation) and data and the reasons for their use.
  • L2-422. The model documentation, processes including roles and responsibilities are covered by the model governance. The documentation is appropriately structured and may include an inventory of all the documents forming the internal model documentation.
  • L2-423. Finally, the documentation identifies the main limitations and weaknesses of the internal model and conditions under which the model may not adequately determine the IAIG’s capital requirement.

Last updated on:
7.3 Internal models7.3.4 Additional considerations