Workstream F: Creating Value Through Internal Models What About Op Risk?

Transcription

1 Internal Model Industry Forum (IMIF) Workstream F: Creating Value Through Internal Models What About Op Risk? Michael Sicsic Paul Morrish 2 February

2 Operational Risk Modelling Operational Risk defined in Solvency II text as: the risk of loss arising from inadequate or failed internal processes, personnel or systems, or from external events. Leading the risk profession through delivery of education and life long learning Covers a wide variety of risks, which can potentially interact with each other, as well as the other risk categories in Solvency II (underwriting/credit/market/liquidity) Wide-ranging impacts combined with comparative lack of data poses a specific set of challenges for modelling and subsequent validation of approaches. Dedicated working group in place to review the end-to-end process of Op Risk capital modelling. ORIC International collaborated with Oliver Wyman at the end of 2014 to survey its members on their Operational Risk Frameworks. Topics included their specific approaches to capital calculations under Solvency II, modelling approaches and Validation.

3 Challenges and Benefits from Op Risk Internal Data Driven Model? Challenges in Modelling Operational Risks Data-driven approaches Not applicable on internal data only, due to scarcity External events: emerging trends, still scaling issues, data pool is growing Use of scenarios: the challenge of uncertainty Scenario-driven approaches Very broad range of events: how to consider them all? How to assess potential major events scenarios? How to correlate scenarios? Data driven Internal Model Benefits Reduce risk! More accurate identification and assessment, better risk culture will eventually reduce major risks Optimised capital compared to standard formula Risk-based approach Scenarios

4 No. of respondents Operational Risk Capital Capital Calculation Method for Operational Risk* Operational Risk Capital (Diversified)* Preliminary Results Standard Formula Internal Model % of total SCR * Of respondents subject to Solvency II requirements from 2016 Value: Getting a risk based approach in order to manage and monitor Operational Risk.

5 No. of Respondents Operational Risk Modelling Approaches 12 Types of Model* Preliminary Results Granularity of Classification 10 Risk Level Scenario 7 Risk Categories 14 0 Loss Data based Scenario based Hybrid Scenario/Loss Data Single loss estimate Other Modelling Approach * Of respondents subject to Solvency II requirements from 2016 Value: Development of an Internal Model integrated with the whole Operational Risk Framework as an indicator of increased risk management maturity.

6 Operational Risk Quantification Fit Curve Frequency Poisson (12) Negative Binomial (3) Preliminary Results Distributions Severity Lognormal (10) Weibull (3) Generalised Pareto (5) Empirical Distribution (2) Aggregation VaR/CoVaR (2) Gaussian Copula (7) T-Copula (2) * Respondents running an Internal Model for Operational Risk Value: Ensure that a higher degree of sophistication does not prevent buy-in from the business.

7 Internal Model Validation Frequency of review: Overwhelming consensus Reviews occur at set intervals Almost always on an Annual basis Preliminary Results Who performs the validation? Generally within Risk Function (but Independent of model owners) Some external support, generally less frequently than internal validation process Validation Techniques Plausibility Checks (25) Sensitivity Analysis (24) Back-testing (20) Model Replication (3) Expert review (against industry practices) (20) Value: Provides assurance that model outputs sensible and the model is fit for purpose; as well as building on understanding to allow continuous improvement.

8 Business Use/Value Preliminary Results The use of Operational Risk tools in decision making: Product Approvals (26) Insurance Purchasing (20) Project Approvals (25) Strategic Planning and Budgeting Process (24) Outsourcing Agreements (26) Performance Management (14) M&A (22) Remuneration Packaging (11) Value: Senior management confidence that model outputs are appropriate for use around the business.

9 Additional Key Points Spurious Complexity Actuarial sciences vs Op Risk Management PRA Engagement Area of concern for most firms ICAS vs. Solvency II Insurance as a risk mitigant Lack of maturity area requiring further work Board Engagement Best Practice identified Getting NED to oversee a specific part of the model

10 Next Steps Complete the survey analysis and report out Additional working group meetings covering: Expert Judgment Distributions Correlations Aggregation Key principles around Validation to be formalised Best Practice on the use of external data PRA Engagement

11 Questions?

12 One last word. Everything simple is false. Everything complex is unusable. Paul Valéry

13 Tel: +44(0) Fax: +44(0) Institute of Risk Management 6 Lloyd s Avenue London EC3N 3AX United Kingdom