Modeling of Automobile Insurance under Solvency II
Roland Voggenauer-Graf von Bothmer Group Actuarial, Allianz SE
1 Automobile Insurance in Germany The Market Modeling of Automobile Insurance under Solvency II
Pricing
Reserving Reinsurance
2
Automobile Insurance under Solvency II Solvency I & US Risk Based Capital Solvency II: Standard Model Solvency II: Internal Model
© Allianz SE 2011
Table of contents
Use Test of the Internal Model
2
Premium generated by the German insurance industry
Modeling of Automobile Insurance under Solvency II
Euro 18 x 1010
© Allianz SE 2011
183,000,000,000 € 3
© Allianz SE 2011
€ 60 bn
* Property & Casualty 4
Modeling of Automobile Insurance under Solvency II
… out of that:
… and the Non-Life Market splits up in: € 20 bn
Premium
About 55 million vehicles Property 43%
More than 45 million private passenger cars
Modeling of Automobile Insurance under Solvency II
Total Automobile 33%
Liability 13%
Germany
… which is 55 cars per 100 inhabitants Accident 11%
Reserves Around 30 million vehicles Approximately 30 cars per 100 inhabitants Only a quarter of the vehicles are insured
Total Automobile 41%
Property 24%
Liability 27%
© Allianz SE 2011
Mexico
Accident 9%
5
Automobile Insurance in Germany Lines of Business Motor Third Party Liability (MTPL): Mandatory since 1939
Modeling of Automobile Insurance under Solvency II
Motor Own Damage (MOD): • Voluntary cover, split up in: - Partial coverage = fire, theft & nat cat
- Full coverage = partial + accidental damage Accidentcover for passengers and the driver Other additional covers, like
© Allianz SE 2011
• Assistance, Mobility • Extended Warranty, Gap •… 6
Automobile Insurance in Germany Share of MTPL and MOD 3 out of 4 cars have an MOD cover
Motor Own Damage 4%
Premium (€ 8 bn + € 12 bn)
Reserves
MTPL 61%
© Allianz SE 2011
Motor Own Damage 39%
Modeling of Automobile Insurance under Solvency II
MTPL 96%
7
Automobile Insurance in Germany Legal requirements for MTPL • Bodily injury (BI):
€ 7.5 mn
• Property damage:
€ 1.0 mn
• Other damages:
€ 50,000
Modeling of Automobile Insurance under Solvency II
Minimum coverage
Standard coverage • Used to be “unlimited”
© Allianz SE 2011
• Now: € 50-100 mn, with BI limited to € 8-15 mn per person Close co-operation between association of insurance companies, Vehicle registration offices, and the police 8
Automobile Insurance in Germany Cumulative market shares of the 14 (out of about 100) largest players 100%
80%
72%
82%
67%
70%
62% 57%
60%
52% 46%
50% 39%
40% 28%
30% 17%
© Allianz SE 2011
20%
76%
79%
85% 86%
Modeling of Automobile Insurance under Solvency II
90%
10% 0% 9
Automobile Insurance in Germany Consequences of a strongly fragmented market Medium and small size insurers need support in their pricing by Modeling of Automobile Insurance under Solvency II
• German Insurance Association (GDV) • Pools Very competitive market (small margin – if any) Focus on distribution channels
Innovations in tariff structures and pricing techniques
© Allianz SE 2011
Cyclical market
10
90%
100%
MTPL
MOD
80%
70%
11
Modeling of Automobile Insurance under Solvency II
60% © Allianz SE 2011
2010
2008
2006
2004
2002
2000
1998
1996
1994
1992
1990
1988
1986
1984
1982
1980
Automobile Insurance in Germany The cycle: Claims(!)-ratios
120%
110%
Automobile Insurance in Germany Market shares by distribution channels
Broker 25%
Modeling of Automobile Insurance under Solvency II
Direct 15%
Agents 60%
Aggregators are gaining market share Across all distribution channels, yet ... ... mainly in Direct
© Allianz SE 2011
• 25% of new business and • >80% of this by one comparison website
12
Pricing of automobile insurance Evolution of rating criteria Type of vehicle, Type of occupation, Region, Bonus / Malus
Today:
On average almost 20 factors, such as:
© Allianz SE 2011
Modeling of Automobile Insurance under Solvency II
Prior to 1994:
13
Criteria to model
b
Identification of risk criteria
e.g. nationality
b
Selection of usable criteria
e.g. origin of driving license
1
Rating criteria 1
Grouping
Region
2
Rating criteria 2
Grouping
Occupation
3
Rating criteria 3
Grouping
Age*Gender
n
Rating criteria n Cube-like structure
© Allianz SE 2011
...
Modeling of Automobile Insurance under Solvency II
e.g. claims frequency
a
Occupation
Interaction of criteria
Pricing of automobile insurance
Region Age*Gender 14
Pricing of automobile insurance Cube-like structure of e.g. 3 dimensions
Munich
Modeling of Automobile Insurance under Solvency II
Actuary, male, 35-40, living in Munich
Occupation
Region
Actuary
Actuary:
x
Munich:
y
male & 35-40: z
Male, 35-40
Gender*Age
© Allianz SE 2011
Tariff calculation uses factors:
… and derives the premium P as: P = Base Rate * x * y * z
15
Pricing of automobile insurance Multiplicative rating structure
with classes i є (1, ... , n) and k ε (1, ... , l) respectively
producing a cube Q of tariff clusters (i,k) ε ((1, ... , n) x (1, ... , l)) =: Q Modeling of Automobile Insurance under Solvency II
Let I and K be rating criteria ...
Let v i,k be the volume and S i,k the amount of losses in (i,k), then: The pure premium in cluster (i,k) should equal Ε(S i,k / v i,k) =: P i,k In a multiplicative structure there are parameters μ, αi, βk such that for all (i,k) ε Q
© Allianz SE 2011
P i,k = μ αi βk
μ being a base rate and αi, βk (normalized) rating factors 16
Reserving in automobile insurance Reserving under local German GAAP requested to be prudent! Usually we see patterns like this … 150%
“Long tail”
Own Damage Known Share of Ultimate
Incurred
Incurred
Third Party Liability Modeling of Automobile Insurance under Solvency II
“Short tail” Known Share of Ultimate
100% Paid Paid
Claims Development in Years
Claims Development in Years 1
2
3
4
5
6
© Allianz SE 2011
50%
Thus run-off losses do occur on portfolio level, but are exceptional with regard to single known claims 17
Reserving in automobile insurance Development Factor Methods
Historical Data
Modeling of Automobile Insurance under Solvency II
(e.g. Chain-Ladder, Bornhuetter/Ferguson)
Future Development
Historical claims development is used to predict future development
© Allianz SE 2011
Apply “Tail Factor” – if need be … which most often is the case
… subject to: data are homogeneous, no systematic changes, etc. 18
Reserving in automobile insurance DF-Methods are applied to both Incurred and Paid
Incurred
100%
Third Party Liability
Modeling of Automobile Insurance under Solvency II
In theory we would expect the two projections to converge at one “Best Estimate”
Paid
In reality this seldomly works out
0% 1
2
3
4
5
6
7
Munich Chain Ladder (by Th. Mack & G. Quarg, 2004) Analyzing correlations between paid and incurred, we often observe
© Allianz SE 2011
• after a low “paid to incurred ratio” higher than average paid factors • after a high “paid to incurred ratio” lower than average paid factors
Making use of this can reduce the gap between paid and incurred projections 19
Reserving of automobile insurance Some of the challenges we face
Duration of BI-claims increases in MTPL (less people die after an accident) … but in MOD claims are settled much faster (better claims management) Court decisions change, mostly in favor of the insured or the claimant (annuity vs lumb sum payment) … many more 20
© Allianz SE 2011
… while the average cost of claims is increasing (increasing values, cost of medical treatment)
Modeling of Automobile Insurance under Solvency II
The claims frequency in MTPL is declining over time (safety standards)
Reinsurance in automobile insurance Typical programs would be
Modeling of Automobile Insurance under Solvency II
Quota Share Risk-based Excess of Loss, e.g.: € 4 mn xs € 1 mn € 15 mn xs € 5 mn € 80 mn xs € 20 mn
MOD Quota Share Event-based Excess of Loss, e.g. € 25 mn xs € 25 mn € 50 mn xs € 50 mn Stop Loss for nat-cat perils Facultative Reinsurance
© Allianz SE 2011
MTPL
21
2 1
Automobile Insurance in Germany The Market Pricing Modeling of Automobile Insurance under Solvency II
Reserving Reinsurance
2 Automobile Insurance under Solvency II Solvency I & US Risk Based Capital Solvency II: Standard Model
© Allianz SE 2011
Solvency II: Internal Model
Use Test of the Internal Model
22
Solvency I – The old world Local German Regulation for Solvency I in Non-Life
Premium index
Modeling of Automobile Insurance under Solvency II
Solvency Margin = max (Premium index; Claims index) = 18% * Premium of up to € 50 mn * Self Retention + 16% * Premium above € 50 mn * Self Retention = 26% * Incurred Claims of up to € 35 mn * Self Retention + 23% * Incurred Claims above € 35 mn * Self Retention
© Allianz SE 2011
Claims index
23
US Risk Based Capital for P/C (NAIC) The first “Factor-Model” in insurance Define:
R0 := Asset Risk - subsidiary insurance companies Modeling of Automobile Insurance under Solvency II
R1 := Asset Risk - fixed income investment R2 := Asset Risk - Equity R3 := Asset/Credit risk – Recoverables, Reinsurance
R4 := Reserve Risk
and put: Total Risk Based Capital :=
R0
2 1
R
2 2
R
2 3
R
R4
2
© Allianz SE 2011
R5 := Premium Risk
R52 24
Europe wants „A new drug ...“ To achieve ... consumer protection focus on risk-management & risk-steering
... in 2009 the Directive stipulated Principle based approach to supervision Market consistent approach for valuing assets and liabilities Capital requirements linked to the company„s risk profile
... to come into force in 2013!? 25
© Allianz SE 2011
Modeling of Automobile Insurance under Solvency II
higher transparency on underlying risks
Solvency II Directive A three-pillar structure
Pillar 1 Risk Quantification
Pillar 2 Risk Management
Pillar 3 Risk Disclosure
Model
Governance
Supervisory and public reporting
Technical provisions
Use Test
Capital required
ORSA
Modeling of Automobile Insurance under Solvency II
Solvency II
© Allianz SE 2011
Transparency
Originally focused on adequate risk management systems 26
Solvency II Directive Pillar 1: Quantitative requirements Limit worst case (financial ruin) within one year to a 0.5% probability
Modeling of Automobile Insurance under Solvency II
Availability of free own funds to cover losses of current business (premium risk) and runoff losses (reserve risk) Market Value Balance Sheet (“Fair Value”) • Minimum Capital Required (MCR) similar to Solvency I
• Use of approved internal models to evaluate the Solvency Capital Required (SCR) - Reduction in required capital is estimated to be 20%, yet … - …only a minority of insurers will apply for the usage of an internal model • Individual evaluation of risks with standard formula, allowing for diversification
© Allianz SE 2011
• Stepwise intervention of the regulator in case MCR < Own Funds < SCR
27
Solvency II Capital Requirement
Modeling of Automobile Insurance under Solvency II
Based on Monte-Carlo simulations
© Allianz SE 2011
Expected Value
28
Solvency II Directive Pillar 2: Governance & Risk management Adequate and transparent assessment of all risks Principle based approach to allow for individual implementations at company level Principle of “Proportionality”: Medium and small sized insurers should not be overburdened Extensive audits and evaluations by the supervisor, e.g. of strategies, processes, governance systems etc. Far-reaching authorization of the supervisor, e.g. in case of outsourcing
© Allianz SE 2011
Germany: Introduction of “MaRisk” in 2009 already anticipates much of that
Modeling of Automobile Insurance under Solvency II
Risk oriented approach: All material risks need to be included
29
Solvency II Directive Pillar 3: Disclosure & transparency Disclosure of information regarding the risk situation – both public and to the supervisor only Modeling of Automobile Insurance under Solvency II
Uniform supervisory reporting within the EEC Public disclosure of the solvency situation following the „Solvency and Financial Condition Report“ including e.g.
• Business policy, corporate structure, market environment, strategies, … • Governance structure and compliance statement • Principles of evaluation of assets and liabilities
© Allianz SE 2011
• Internal governance • Required MCR / SCR
• Disclosure and justification in case the capital requirements are not meet 30
The Standard Model Solvency Capital Required (SCR)
Intangable Assets
Health Business
Modeling of Automobile Insurance under Solvency II
Life Business
Non-Life
Non-Life Business
Market Risk
© Allianz SE 2011
Credit Risk
Credit Risk
Life
Intangibles
Health Market 31
31
The Standard Model – Non-Life Risk
Premium Risk
Modeling of Automobile Insurance under Solvency II
Non-Life Business Non-Life
Reserve Risk
© Allianz SE 2011
Nat Cat Risk
32
The Standard Model Correlation:
Premium Risk
Modeling of Automobile Insurance under Solvency II
How likely is it to have a major hurricane and to increase reserves for prior years at the same time?
Nat Cat Risk Reserve Risk
Measurement of the Risk Exposure:
© Allianz SE 2011
Volume * Factor
Based on volatility & 99,5% percentile For reserves and future business
Allow for diversification by line of business and country
33
The Standard Model
Modeling of Automobile Insurance under Solvency II
Concept for Nat Cat Scenario based
Geographical exposure and insured volume Add manual Cat if needed
© Allianz SE 2011
Allow for diversification
34
Solvency II – Internal model for P/C business Ultimate premium risk (non-cat) Gross Model (Sub-LoB Level) 2
Distribution fitting Large Losses
Attritional Losses
Exposure adjustment
Inflation Adjustment
Historical data (Premium/Exposure/Losses)
IBNR/IBNER adjustment
Individual Large Losses
Best Estimate Ultimate by AY
Inflation Indices
Dependencies
Sub-LoB aggregation
Frequency & Severity
Premium Cycle
Frequency
Frequency
Planning data (Premium/Expenses/ Exposure/Losses)
4
3
Adjusted Data
Raw Data
Data Adjustment
Modeling of Automobile Insurance under Solvency II
Data Input
Severity
Severity
Net Model (LoB Level) 5 Reinsurance
6 Levels of reinsurance 1 - QS & Surplus 2 – Risk XoL 3 – Event XoL 4 – Multi-Line XoL 5 – Stop Loss 6 – Net QS
Severity
UW Result (LoB & LE Level) 6
Dependencies
LoB aggregation
Low/Medium/High
7
UW result gross/net
© Allianz SE 2011
1
35
35
Solvency II – Internal model for P/C business Assessing the reserve risk by bootstrapping techniques A Parameterisation
(1) Data triangle
(2) Fit a CL model to your data
What would have been the historical data given the latest loss information and under the assumption the CL model is the true model?
(3) Fitted triangle
- Residuals can be used for sampling with replacement
- Pearson residuals - for incremental losses (ODP) - for individual development factors (Mack) - deviation from fitted to original triangle - need to be standardised
- Fitted triangle plus different sets of residuals result in pseudo triangles
(4) Calculate residuals
B Simulation Pseudo Reserves
- Process error is included when forecasting
0,016
0,014
Probability Density
0,012
0,010
0,008
Modeling of Automobile Insurance under Solvency II
Reserve Estimates
© Allianz SE 2011
0,006
0,004
0,002
0,000 750.000
800.000
850.000
900.000
950.000
1.000.000
1.050.000
Reserve
… (5) Generate pseudo triangles
… (6) Refit the same CL model
1.100.000
1.150.000
1.200.000
1.250.000
1.300.000
1.350.000
- Also provides stochastic cash flow
(7) Distribution of reserves 36
Solvency II – Use Test In order to prove the quality of the internal model
© Allianz SE 2011
Modeling of Automobile Insurance under Solvency II
Solvency II requires it to be used for daily business decisions!
37
Solvency II – Use Test Example The large loss model & hence the purchase of reinsurance Large losses very much depend on the individual insurer Modeling of Automobile Insurance under Solvency II
No standard model is able to reflect this appropriately The internal model is simulating empirical large losses
The reinsurance program should provide appropriate protection against them
Hence:
© Allianz SE 2011
The large loss model should match the reinsurance program
38
Solvency II – Use Test Linking Risk Models and Business Management Business Management
Technical Pricing
NatCat Limit Controlling Reinsurance Optimization
Modeling of Automobile Insurance under Solvency II
Capital Assessment
Strategic Planning
ALM Underwriting
RC Calculation
Reporting Reserving MVM Calculation
© Allianz SE 2011
Claims Analysis
Plan Year‟s Exposure A. Loss Ratio A.
39
Future Development
Incurred
Historical Data Occupation
Paid
Pillar 2
Pillar 1 Third Party Liability
100%
Model 0% 2
3
4
5
Pillar 3
6
Risk Manage ment 7
Region
Reporti 90% ng
Age*Gender76%
80%
82%
85% 86%
72%
67%
70%
62% 57%
60%
52% 46%
50% 39%
40% 28%
30% 20%
79%
© Allianz SE 2011
Solvency II
100%
1
Modeling of Automobile Insurance under Solvency II
Any Questions?
17%
10% 0%
40
Thank you for your attention!
Roland Voggenauer-Graf von Bothmer
[email protected] +49 (0) 89 3800 13480