Solvency II – The Final Test Israel April 5 th , 2011 Prepared by Marc Beckers of Aon Benfield Analytics
Solvency II – The Final Test
Israel
April 5th, 2011
Prepared by Marc Beckers of Aon Benfield Analytics
Solvency II: Executive Summary
Harmonizes insurance regulation across Europe, replacing Solvency I (est. 1973) and local regimes (e.g. UK ICAS)
Start date January 2013 + transitional period of up to 5-10 years
Capital charges under Solvency II will be finalised in Dec 2011
Risk-based approach to required capital that demands insurers to develop robust risk management practices
Risk and capital mitigation effect of reinsurance only allowed if the reinsurance counterparty has a Solvency II ratio above 100% or a rating of at least BBB
The average non-life solvency ratio expected to decrease from about 200% under Solvency I to 165% under Solvency II
Captives and mono-lines are at a significant disadvantage as the standard formula penalises those writing less diverse portfolios – both geographically and in terms of class of business.
The capital charges for counterparty default risk may drive a flight to quality for reinsurance counterparties (rating A or better)
Catastrophe risk is a key concern within the Solvency II Standard Formula
The local catastrophe scenarios under QIS 5 often lead to a materially higher result than the commercial catastrophe models
2Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Topics for discussion
Section 1 Summary Conclusions QIS 5
Section 2 Life, Non-Life and Health within QIS 5
Section 3 Partial Internal Models in QIS 5
3Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Section 1: Summary Conclusions QIS 5
Solvency Ratio from 310% in Solvency I to 165% under QIS 5 ! Diversification for composite solo submissions 32% and 46% for groups Quality of QIS 5 submission debatable Significant issues with the QIS 5 requirements for some key risks Time is running out
Summary: Surplus falling under QIS 5 but still acceptable on average
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 5
Surplus: Solvency I to QIS 5
Overall 2009 level of surplus for European industry has decreased by 30% representing a €120bn reduction (from €480m to €360m)
The margin under the MCR has increased by €200bn
Industry wide solvency ratio falls from 310% to 165% under QIS 5
SR = 310% 165% 466%
Valuation Adjustment Effect +99%
Analysis of Surplus Movement
Market consistent valuation of assets under QIS 5 reduces solvency ratio by 14%
Technical provisions assessed at fair value increases solvency ratio by 107%
Risk based capital requirements are the key driver for the reduction in solvency ratio
Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Are diversification and loss absorbing mechanisms too large ?
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 6
Source: EIOPA QIS 5 report
32% diversification for solo composites and 46% for groups versus a diversification benefit by S&P of below 10% !
Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Groups – Key Observations
Under Standard Formula, average group surplus reduces by 43% vs Solvency I, but using internal model surplus increases by 6%
Average group diversification 20% under accounting consolidation
Significant adjustment for loss absorbency and deferred taxes
Equal balance between life and non-life insurance risk
Market risk is substantial at 57% of BSCR
– Equity risk and spread risk are significant
– High interest rate charge suggests poor ALM
Counterparty charge large relative to other risk categories
Diversified Group SCR
7Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
UK: Acceptable quality except for USPs and SCR Underwriting Risk !
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 8
Source: FSA QIS 5 report
Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
What issues transpired in QIS 5 ?
Ireland difficulties:
1. Complexity, especially counterparty default risk module and non-life cat risk
2. Non-Life underwriting risk calibration
3. Life contract boundaries and Expected Profits in Future Premiums (EPIFP) France difficulties:
1. Asset risk fair value very different from today
2. Complexity SCR, especially non-life cat and non-proportional reinsurance
3. Calculating technical provisions, eg. contract boundaries and illiquidity premium
4. EPIFP Spain:
1. Complexity counterparty default risk (need to run the model twice !)
2. Catastrophe risk for non-life and life, eg. consorcio not taken into account in life Cat
Other observations: poor data “quality control”
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 9Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Solvency II implementation to cost in excess of €3m on average
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 10
Source: FSA QIS 5 report
Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Was QIS 5 “only” a test ?
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 11Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Solvency II implementation by 2013
2005 2006 2007 2008 2009 2010 2011 2012 2013
Directive Development(Commission)
Directive Adoption(Council & Parliament)
Level 1 Implementation in member states
Solvency II Start Date1 Jan 2013
Level 2Implementation Measures (CP’s)
Level 3Guidelines by
Regulators
Proposal expected in June 2011Final measures by end 2011
Expected before March 2012
(depends on Level 2)
QIS
Omnibus II
QIS 1 QIS 2 QIS 3 QIS 4 QIS 5 QIS5 results expected by March 2011
Transitional Phase?
QIS = Quantitative Impact Studies
Key Dates for 2011 July 2011 Sept 2011 Q3 2011 Nov 2011 Dec 2011
EC finalise technical standards for Solvency II
Pan European Stress Test
EIOPA Advice on assessment of Switzerland, Bermuda, Japan
3rd country assessment methodology
EC confirm transitional measures for Solvency II
12Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
?
Transitional measures under “Draft” Omnibus II
2013 2014 2015 2016 2017 2018 2019 2020 2021 2022
Valuation of Assets & Liabilities
Requirement for fair value assessment of assets and liabilities delayed for up to 10 years Existing member state rules as at 31 December 2012 will apply during transitional period
Calculating Technical Provisions
Requirement for calculation of technical provisions delayed for up to 10 years No requirement to be as best estimate with discounting and risk margin, Existing Solvency I directives will apply during transitional period
Tiering of Own Funds
Requirement for criteria by which funds classified into various tiers delayed for up to 10 years Existing Solvency I directives for classification of own fund items will apply during transitional period
Public Reporting
Delayed for up to 3 years Higher level reporting in Solvency II
directive required from 2013
SCR
Requirement for SCR calculation delayed for up to 10 years Calculation of ‘transitional SCR’ may include modifications to standard formula calculations Insurers must comply with a transitional SCR that is no higher than the SCR and no lower than the
sum of the MCR and 50% of the difference between the SCR and the MCR.
Governance Delayed for up to 3 years Existing Solvency I directives will apply
during transitional period
Draft Omnibus II directive published by European Commission on 19 January 2011
Solvency II start date confirmed as 1 January 2013
Omnibus II provides maximum transitional periods
EC will publish delegated acts in 3rd quarter 2011 confirming actual transitional measures – likely to be shorter than the maximum time periodsEquivalence
of third countries
Criteria for equivalence delayed up to 5 years Decisions would be made on case by case basis under delegated
acts of law
Supervisory Reporting
and Governance
RSR is delayed for up to 3-5 years (including supporting systems and governance)
Existing Solvency I reporting applies
13Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Section 2: Life, Non-Life and Health in QIS 5
QIS 5 shows that key challenges remain before the implementation of Solvency II
Underwriting risk is the key SCR component for non-life insurers
Non-Life Solo SCR
Average capital requirement for non-life insurers is higher than life insurers under Solvency II
– This is driven by the loss absorbency of life mathematical provisions Greater diversification in the BSCR for non-life insurers Underwriting risk is the key component for non-life insurers Market risk is largest component of life insurance capital requirement
* Adjustment for tax is estimated for non-life SCR based on group value provided in EIOPA report on QIS 5. ** Diversification is estimated using QIS 5 correlations and non-life / life individual risk SCR’s post-diversification
Life Solo SCR
15Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
QIS 5: Life and Health Cat Risk 11% of U/W risk pre-diversification
LIFE CAT RISK 1.5 per mille on capital at risk
HEALTH CAT RISK Arena risk (50% of stadium) + Concentration risk (100% + 300m around) + Pandemic risk (0.075 per mille)
Accidental Death 10%
Permanent total disability 1.5%
Long term disability 5%
Short term disability 13.5%
Medical / Injuries 30%
16Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
QIS 5 results show the importance of EQ risk (net of reinsurance)
Overall reinsurance is a key aid to reduce the exposure to NatCat risks
Overall (net) EQ exposure within a local Cypriot non-life insurer would be about 25% of the total capital requirement
“The CTF recommends a more accurate and appropriate estimation of the undertaking's catastrophe risk through the use of a partial internal model”
17Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
NatCat: Earthquake
)()( ZoneZoneZoneZoneCountryEarthquake
CountryEartquake TIVFTIVFAGGQCAT
Calculate the gross 1/200 OEP per country
Total Insured Value per Cresta
Vulnerability factor (quake)
“Aggregation” Matrix (quake)
1 in 200 OEP factor
Parameters-non-life-catastrophe-risk_en.xlsEQ_CRESTA_CY
Provided by company
18Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Evaluating catastrophe models – components
Industry Exposure database
Industry Exposure database
Geocoding ResolutionGeocoding Resolution
Occupancy Type
Occupancy Type
Vulnerability Regions
Vulnerability Regions
Construction Type
Construction Type
Post-Loss Amplification
Post-Loss Amplification
Loss by Geographical RegionLoss by Geographical Region
DeductiblesDeductibles
Losses by LoB / Coverage
Losses by LoB / Coverage
Regional CorrelationRegional
CorrelationBusiness InterruptionBusiness
Interruption
Line of BusinessLine of Business
Hazard EP Curves / RPsHazard EP
Curves / RPs
Several steps along the catastrophe modelling chain where we can evaluate and compare models
19Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Potentially significant difference in output between Standard Formula approach and cat models Scenario testing vs cat model by client Negative percentage = cat model lower
than scenario For over 80% of clients the standard
formula for natural cat is unacceptable
Standard Formula Nat Cat – Cat Models vs Scenario
France
Comparison with Cat Models
Damage by Cresta zone is going back more than 15 years in time Munich Re / Swiss Re approach prior to
cat models? Almost all clients have data more
granular than Cresta All commercial cat models have
finer granularity Data quality is ignored
No impact for location granularity No differentiation by occupancy No impact for construction, age, height Single damage function
No differentiation by buildings, contents, BI
No application of limits, excess, original deductibles
Issues with Standard Formula
Our Recommendation: Catastrophe Partial Internal Model
-60%
-40%
-20%
0%
20%
40%
60%
80%
100%
120%
140%
A B C D E F G H I J K L M N O P Q R S T U V W X Y Z AA AB AC AD AE AF AG AH AI AJ
Per
cen
tag
e
Cedants
PML QIS 5 > PML vendor models
PML QIS 5 < PML vendor models
20Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Non-Life Solos – Key Observations
Non-life cat risk has significant issues with calibration under Standard Formula
Health care component relatively small but complex to compute
– Segmentation into SLT and non-SLT of workers comp difficult
– Similar to life (SLT) only 2.5% of BSCR for non-life firms
• Is the complexity justified? Life risk close to nil Market risk smaller for non-life
companies reflecting lower risk investment strategies
– Equity risk largest component
Diversified Non-Life Solo BSCR
21Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Non-life SCR (Solos)
Manmade Cat
Natural CatM
etho
d 1
28%
Premium risk largest driver of non-life capital
– Most entities did not use USP’s or NP adjustment Cat risk split fairly evenly between Method 1 and 2
– Non-EEA cat risk carries significant capital charge Manmade cat charge similar to Nat cat
– Does this make sense for a risk not currently explicitly modelled?
Lapse assumed nil by most companies
Non-Life SCR (Solos)
22Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Health SCR (Solos)
Health Cat
Similar to Life (SLT) Disability is the largest risk within SLT (76%)
– Income protection disability charge too penal Lapse risk also significant, many reported difficulties to identify
+ve and –ve surrender strains Health cat does not factor in medical expenses for pandemics
– Also difficult to calculate and inconsistent with life cat Cat scenarios too generic for local markets and insurers
Health SCR (Solos)
39%
11%
23Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Life SCR (Solos)
Lapse risk and longevity risk are most material life risks
– Feedback on longevity risk critical of it only being a shock on level, rather than accounting for trend risk
– Lapse risk calculation was criticised as too onerous on a policy by policy basis Revision risk is nil
– Surprising result: possible error in report? (in appendix to EIOPA report = 2.9%)
– Or indicator that calculation is too complex for most companies?
Life SCR (Solos)
24Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Market Risk SCR (Solos)
Credit Quality
Equity and spread risk are most significant
– Both criticised as too penal
– Spread risk excludes Sovereign risk
• Note EIOPA stress test includes it Currency risk: counterintuitive to hold assets in
reporting currency instead of the liabilities Counterparty risk: too complicated and cash at
bank too penal relative to equivalent spread risk
Return Generating Assets
Market Risk SCR (Solos)
Other Equity forms part of Equity charge (exact % unspecified in EIOPA report)
25Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Non-life: Undertaking Specific Parameters and Non-Proportional Adjustment
Approximate 20% reduction in underwriting volatility under non-proportional adjustment for liability and fire
Significant area of feedback:– Too complex, most undertakings did not complete– Not suitable for different types of non-proportional
reinsurance– Data requirements onerous
Undertaking Specific Parameters
Sample Size Sample Size
Non-Proportional Adjustment
USP’s submitted to supervisors under QIS 5 generally provide a significant reduction in volatility Especially for liability classes
Some concern about cherry-picking and USP’s not becoming an “internal model lite”
26Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Section 3: Partial Internal Models for Catastrophe Risk
Who is planning an internal model for catastrophe risk? Is an Internal Model worth it? Do not underestimate the complexity ! Internal models could be the answer, but will they be?
Weighted average benefit from Internal Model 20%
Aon Benfield | Israeli Market presentation Proprietary & Confidential | 5 April 2011 28
Very high standard deviation of results: between 25% and 250% on 27 German insurers
Weighted average SCR via Internal Model 140% of Standard Formula result in Spain The average life firm’s internal model SCR was almost equal to their standard formula
SCR in the UK UK Internal Models:
Source: FSA QIS 5 report
Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Internal Models – too complex and worth the effort ?
For a model to be approved the following need to be satisfied:
– Use test: have to show that the model is used as a decision tool in daily risk management work
– Statistical quality standards
– Calibration standards
– Profit and Loss attribution
– Validation standards
– Documentation standards
– Internal Model governance
– Integration of external models needs to be understood In general: what regulators want to see is a controlled process around the internal
model, acknowledged and used by management Most popular use test example: reinsurance !
Do regulators have the capabilities or will they rely on outside consultants (like they are doing in Switzerland for the SST)
29Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
An example of a pre-application timetable (FSA)
Quarterly face to face meetings
Reviews and assessments of internal model
FSA Review
Application submission
SII requirements and contents of application
meetings
Completion of self assessment
Meetings to agree work-plan
Monthly reporting by company
2010 2011 2012
Scoping and planning
Experience shows that it takes at least 2 years from kickoff to approval (expect over 50 on-site visits from regulator)
Hundreds of documents Thousands of pages About 200 meetings About 100 employees involved (60% quantitative people) About 15 departments involved
30Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Catastrophe Partial Internal Models
Area Considerations
Internal model approval process is currently very onerous required detailed documentation and validation of the science behind the catastrophe model
If the client cannot answer questions about the internal model (including the underlying external models), the model will not be approved
Potential issue for most commercial models, which are largely black boxes One regulator indicated preference for Impact Forecasting due to transparency
Clients fix the “model boundaries” at the outset If model changes fall within the boundaries, no need for renewed model approval
One regulator indicated that they prefer a multi-model approach Advantage: a change in one of the models would only partially impact the results No need for equal weighting of different models: is this ensuring best practice?
Some regulators indicated a pragmatic approach to facilitate the use of external commercial cat models and have the clients benefit from their data quality
Possible solution: simplified internal model approval process for cat (proposal submitted in February 2011 to EIOPA by Aon Benfield)
A simplified Internal Model approval process would be beneficial to insurers and to regulators and lower significantly the barrier for internal models as well as reduce the workload (and cost) for regulators
Focus for a simplified approval process should be on input data requirements and the process to correctly apply the catastrophe models
Model Approval Process
Model Change
Simplified Approval Process
31Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
QIS 5 as a precursor of Solvency II: The good, the bad and the ugly
Sufficient surplus capital on average
Better awareness of risk and the value of data quality
Quality of QIS 5 was quite poor in many aspects
Many results from the Standard Formula were wrong
Standard Formula overall much too complex
Resource requirements will come at a cost !
32Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
Advise from one of the EU regulators
Take the time for implementation !
Get used to market value and the corresponding volatility !
33Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011
For more into, contact us locally or globally
Gareth Haslipt: +44 20 7522 8137e: [email protected]
Tessa Moultont: +44 20 7522 7137e: [email protected]
Jürgen Wielandtst: +32 2 661 7164e: [email protected]
Marc Beckerst: +44 7931 472 999e: [email protected]
34Aon Benfield | Israeli Market presentationProprietary & Confidential | 5 April 2011