Internal Modelling of Operational Risk Under Solvency II

Febrabans F b b International I t ti lO Operational ti l Ri Risk k Conference, 12-14 May 2010

www.abioric.com

Agenda

Operational risk under Solvency II Challenges of internal modelling How ORIC fits in

Solvency II: principle-based regulatory themes

SOLVENCY II
PILLAR 1 Quantitative requirements Risk-based capital requirements MCR SCR: standard approach versus internal modelling PILLAR 2 Supervisory activities and internal risk governance Identification of risks not captured under Pillar 1 Supervisory review ORSA should cover risk identification, measurement, management, reporting and monitoring PILLAR 3 Reporting and disclosure Improve private and public risk disclosure Market-consistent valuation More transparency (firms risk profile & management practices)

2010
Level 2 implementing measures

2011

2012

2009 June 10 ECs draft proposals for level 2

October - November Adoption of L2 by EC

H2 2011 Final discussions in Council and Parliament

Quantitative Impact Study (QIS) 5

Feb - Mar Draft QIS 5

Early May Deadline for comments

July - October QIS 5

Q1 2011 QIS results

UKs IMAP process

April - October Entry into FSAs pre-application process April 2010 H2 2011 p IM application process runs

H2 2011 - H2 2012 Application review

Operational risk in Solvency II

Operational risk included in the SCR (arts 101, 103) Definition: inadequate or failed internal processes, or from personnel and systems, or from external events (art 13) (art. Includes legal but excludes strategic and reputational risks (art (art. 101) Outsourcing viewed as a potentially major source of operational hazards (art. ti lh d ( t 49)

Operational risk SCR

Standard formula Risk modelling

Operational risk capital charge

Full vs Partial internal modelling for determining firm firms s SCR

Standard Formula approach

9 Key features of the SII Level 1 text
SCR for operational risk depends on volume, Also, Capital charge capped at a % of Basic SCR 100% correlation with other risks (no divers. benefits) Technical provisions p Earned premiums Expenses related to unit-linked business

Standard SCR

PROVISIONS
MAX of

Life Non-life Health

Capped at X% of BSCR

Expenses unit-linked business

Other loading factors

PREMIUMS

Standard SCR: technical provisions

Standard SCR: premiums

Standard SCR: other factors

Internal modelling approach

9 FSAs Internal Model Approval Process (IMAP)
Sh t t Short-term actions ti Firms who aim at IM encouraged to do QIS5 Meet Pre Pre-Application Application Qualifying Criteria assessment (PAQC) Optional process; note not pre-approval process If IM is rejected, firm reverts to standard SCR

IM is tool to better understand firms risk profile criteria: Model scope scope, architecture and development plans Timeline, methodology, tests, statistical quality, documentation Model governance, change policy (for full and partial models) Group issues (for complex groups)

Pre-application timeline

Internal model SCR

Use test

P&L attribution

Statistical quality and data

Validation

Tests and standards for IMA

Calibration

Model g governance

Documentation

Challenges of oprisk modelling

Internal loss data collection understand key drivers of costs and benefits (awareness (awareness, culture) Risk categorisation event classification boundary issues (oprisk often less well developed risk area) Data consistency and quality (e.g., measurement) Scaling of external losses / benchmarking Root-cause analysis (key learning tool)

OpRisk M&M
Internal loss data

Reporting, KRI etc. KRIs, t

External loss data

ORM
Selfassessments Scenarios

Root cause analysis

Scenarios forward-looking element

Identifies and characterises (33) ( ) top-priority pp y scenarios for oprisk in insurance Provides industry guidelines f frequency for f and d severity it assessments Explains the key stages of a sound scenarios programme Analyses y behavioural biases in SA workshops

Loss data and scenario assessment

Robust internal modelling

Less reliable modelling

Purely internal losses

Publicly available (and some proprietary) losses

More reliable modelling

Consortium-based loss events

Evidence on the need for scaling

Analyses robust techniques for scaling the size and number of external loss events Illustrates sound approaches and identifies early trends in the ORIC database Size of the insurer is correlated with the severity and number of operational losses But great deal of the variability in observed variability remains unexplained by models room for improvement p

Key exposures for benchmarking/scaling

Written Premiums Technical Provisions/Insurance Liabilities Non-financial Assets Number of FTEs Operating Expenses

Illustrative ORIC benchmark report (WIP)

Quarterly benchmark report for oprisk losses

Example of scaling
Internal losses only
0.6

Internal and scaled external losses

0.6

0.5

0.5

0.4

0.4

Probability

Probability
9 10 0 11 1 12 2 13 3 14 4 15 5 16 6 17 7 18 8

0.3

0.3

02 0.2

0.2

0.1

0.1

0.0

10

12

14

16

Loss amount (in natural logarithm)

Loss amount (in natural logarithm)

Input

BetaGeneral

Input

Lognorm

18

Thank you!
For more information
Mariano Selvaggi, Head of ORIC +44 (0)20 7216 7387 Mariano.Selvaggi@abi.org.uk i l i bi k

www.abioric.com bi i