Advanced (Economic) Capital Topics Cliquez pour modifier le style des sous-titres du masque
Michel Rochette, MBA, FSA 2009 Valuation Actuary Symposium Anaheim, CA September 25th 2009
Topics General topics: Purpose and principles of capital model development Major components of a capital model Uses of a capital model Specialized topics: Approaches to building a capital model Validation considerations Calibration considerations Correlation in the tail: diversification Emerging risk considerations 25/09/2009
Enterprise Risk Advisory LL
Purpose ¨ Risk management system of an insurer for the
analysis of the overall risk situation of the insurance undertaking, to quantify risks and determine the capital requirement on the basis of the company specific risk profile¨ CEA Groupe Consultatif Required capital is assessed in light of: available capital & other financial resources enterprise risk management processes strategic goals & risk appetite regulatory requirements 25/09/2009
Enterprise Risk Advisory LL
Principles All material risks should be covered: links to ERM
and emerging risks Models must be appropriate for the scale and complexity of the firm Models must be dynamic and flexible Models must be embedded in the financial, strategic and operational processes: Use Test in Solvency II Governance of models development: Board/top management oversight and involvement documentation of models, limitations & changes
internal controls over development: auditableEnterprise Risk Advisory LL
25/09/2009
Major components Exposure models of key risks: financial risks & underwriting risks: assets and liabilities models cash flows non financial risks: operational and business models strategic risks: strategic models Risk drivers models: ESG, catastrophic, scenarios,
stochastic, EVT, competitor, behaviour, management actions Aggregation approaches: correlation with var/cov, copulas, none Time horizon: short-term view versus run-offEnterprise Risk Advisory LL 25/09/2009
Uses
Investment decisions: existing and new Product development Strategic decisions Corporate finance decisions: financial leverage Hedging strategies External events and emerging risks Regulatory proposals: CP 37 & CP 56 in Solvency II “…widely used and plays an important role in the course
of conducting an insurer's regular business, particularly in risk management. "
25/09/2009
Enterprise Risk Advisory LL
Approaches
Top down economic and business scenarios: obtain an overall EC estimates for all risks combined Stress tests: judgmental and test specific risks and impact on capital Stochastic: random scenarios and obtain distribution of risks Insurance approach: Frequency, severity, recovery Factor based: used for regulatory Others: Enterprise Risk Advisory LL 25/09/2009 Regression, neural networks, Bayesian, fuzzy logic, EVT
Validation principles Integrates both qualitative and quantitative elements Provides that the models were designed, work as
planned and are implemented correctly – quality assurance Analyses the predictive properties of the models: testing against experience, backtesting Iterative process to assess that assumptions & data are appropriate with a certain degree of confidence: regular cycle Need independence of validation to satisfy basic risk Enterprise Risk Advisory LL management principles: internal and/or external 25/09/2009
Validation elements Model development, design, implementation and
operations: similar to IT systems controls in place like COBIT Review of models inputs: assumptions & key risks continuous appropriate mathematics and
methodologies data accuracy
Review of basic functioning of the models: gaps to internal standards and best industry practices model replication with a different set of random Enterprise Risk Advisory LL 25/09/2009
Validation elements Historical performance: back testing to external sources: industry studies, academic papers, regulatory and rating agencies’ capital Profit and loss attribution: comparison of actual
results to risk drivers predicted by the models. Idem to a source of earnings analysis Management oversight: has management been using the models? has management put in place processes to obtain
assurance that the models are still appropriate
Documentation and independent validationEnterprise Risk Advisory LL 25/09/2009
Calibration principles For each risk drivers, should aim to calibrate four
elements:
level of the risk factor and its uncertainty trend of the risk inherent volatility calamity/catastrophic/tail
Market conditions : impact on pro/counter cyclicality Frequency of calibration: at least annually and
probably more often for financial risks Should be performed before hedging Enterprise Risk Advisory LL 25/09/2009 Should be based on best assumption. No margin
Calibration by risk
Interest rate risk: take into account the parallel , twists, inversion of the term structure QIS4 tail up shocks: 94% at 1yr – low - to 40% multipliers at 10yr interest rate volatility: usually set separately: * 1.5 Equity risk: use different calibrations for publicly-traded, private equity, hedge funds, emerging markets for publicly-traded: tail risk decline of 40% at 99.5% for hedge funds: recent decline around 20% Enterprise Risk Advisory LL 25/09/2009
Calibration by risk Credit, counterparty & asset risk: in a total return context, spread risk anticipates future defaults and migration. No need for an explicit default model spread risk varies by type of assets, rating and currency in Q1S4, spread volatility around 30% and shocks of about 90 bps to treasuries. Probably too low given recent experience concentration risk must be assessed for default risk: recovery assumption crucial in the 30% to 40% range 25/09/2009
Enterprise Risk Advisory LL
Calibration by risk Life underwriting risk: QIS4 mortality rate increased by 15% permanent with a 2.5 additional per mille mortality catastrophe shock – debate in light of potential pandemic lapse shock depends on impact. Can go as high as 100% multiplicative longevity rate increased by a permanent 25% Operational risk: must move beyond the factor based
approach to modelling explicitly and map to insurance coverage and other internal controls Liquidity risk: can be modeled and not simply Enterprise Risk Advisory LL 25/09/2009 managed
Correlation in the tail Correlations exist at different levels: within a risk category: Market Risk
Interest rate
Equity
Interest rate
1
Equity
75%
1
FX
25%
25%
FX
1
between risk categories within an entity between legal entities: should probably be zero because
Enterprise Risk Advisory LL of the non-fungibility of capital and the non recognition
25/09/2009
Correlation in the tail ď‚— Recent experience seems to indicate otherwise ď‚— According to a recent Pimco study:
25/09/2009
Correlation to S & P 500
Early 90s
Early 2008 2008 Meltdown % yearly loss
S & P 500
1
1
37%
High-Yield Bonds
20% -30%
80%
26%
International stocks
30% -40%
70%
45% - 55%
Real Estate
30%
60% -70%
37%
Commodities
0%
-20% -30% 37%
Enterprise Risk Advisory LL
Correlation in the tail: lessons Correlations are unstable in the tail and this what EC
is trying to determine Independent risks become dependent in extreme times: subprime business practices – operational risks › enhanced defaults - credit risk ›
market losses on securitized investments – market risk › capital problems at many FIs – liquidity risk › bankruptcies of many FIs – systemic risk › lawsuits by investors and regulators – legal risk ›
enhanced regulations – regulatory risk ›
25/09/2009
Enterprise Risk Advisory LL
Correlation in the tail: lessons “When people start buying an asset, the act of them
diversifying ultimately makes the asset less of a diversifier .“ Pimco’s Head of analytics Rule: total diversification benefit should not be above 30% One potential approach is to use Clayton copulas which measure non-linear dependency This is difficult as we trying to assess 1 in 200 year events 25/09/2009
Enterprise Risk Advisory LL
Emerging risk EC must be a forward looking process , tied to ERM
and thus must anticipate emerging risks Risk issues and impact on EC – mostly Solvency II
liquidity premium: not allowed in the calculation of the
market consistent value of liabilities discount rate: most likely the risk-free not swap rates group support: not allowed and impact on diversification assumptions in EC calculation MVM: currently set at 6% with no diversification benefit 25/09/2009
Enterprise Risk Advisory LL
Emerging risk
Environmental risks – US based: Fiduciary Responsibility: Legal and Practical Aspects of Integrating ESG Issues into Institutional Investment – UNEP FI NAIC is requiring insurance companies with at least 500 million in annual premiums to start estimating and publishing an Insurer Climate Risk Disclosure Survey starting in May 2010. NAIC seeks to determine "how insurers are altering their risk-management and catastrophe-risk modeling in light of the challenges posed by climate change. “ › direct Enterprise Risk Advisory LL EC implications 25/09/2009
CONTACT
Michel Rochette, MBA, FSA Enterprise Risk Advisory, LLC 954-607-6969 michel.rochette@enterprise-risk-advisory.com
SOCIETY OF ACTUARIES
Valuation Actuary Symposium (September 2009) Session Topic:
Advanced Economic Capital Session:
Overall Rating
All Sessions
29 PD
Expected Attendance
3,643
72
Actual Attendance
3,285
52
Number of responses
1,925
23
Return rate (# of resp./actual att.)
59%
44%
Overall rating of this session
3.69
3.05
Provided you with practical technical information
4.01
3.52
Will enable you to make better business decisions
3.88
3.43
Prepared you to impact industry-wide changes
3.69
3.38
Knowledge of Subject
4.14
3.39
Effectiveness of Delivery
3.68
2.77
1
0
Knowledge of Subject
4.14
3.83
Effectiveness of Delivery
3.68
3.05
1
1
Knowledge of Subject
4.14
3.71
Effectiveness of Delivery
3.68
2.50
1
1
3.82
3.42
4.38
3.07
All Sessions
1
Learning Experience
2
Indicate your level of agreement with the following. This session:
Michel Rochette, FSA
Number of participants indicating presenter included commercial promotion in presentation
Presenter Effectiveness1
This Session
Mark J. Scanlon, FSA, FIA, MAAA, CERA
Number of participants indicating presenter included commercial promotion in presentation
3.05 3.69 3.52 4.01 3.43 3.88 3.38 3.69 3.39 4.14 2.77 3.68 0 1 3.83 4.14 3.05 3.68 1 1
Raj Guttha
Number of participants indicating presenter included commercial promotion in presentation
Mark J. Scanlon, FSA, FIA, MAAA, CERA
Moderator Effectiveness 1 : Rate the moderator's skills in managing this session
Rate the level of audience interaction for this session (7=high, 1=low)
3.71 4.14 2.50 3.68 1 1 3.42 3.82 3.07 4.38
1
The rating scale used: Excellent (5), Very Good (4), Good (3), Fair (2), Poor (1), and N/A (no value).
2
The rating scale used: Strongly Agree (5), Agree (4), Neither Agree nor Disagree (3), Disagree (2), Strongly Disagree (1), and N/A (no value).
Evaluation Tips to keep in mind when reviewing the responses: Numerical evaluations tend to give you a pretty good feeling for how well the attendees responded to the session as a whole. Scores in the range of 3 to 5 are considered successful programs. Written comments come from people who may have a strong opinion, therefore they tend to be very good or very bad. Repetitive comments that point to the same theme could be an indication of an area you may want to capitalize on in the future or work on for future presentations. Perception Solutions, Inc.
www.perceptionsolutions.com
10/9/2009