Brehm1 Flashcards
Enterprise Risk Management definition
ERM is process of systematically and comprehensively identifying critical risks, quanitfying their impacts and implementing integrated strategies to maximize enterprise value
key aspects of ERM to keep in mind
- program should be regular process, not one-time project; should be dynamic and ready to respond to changing conditions
- risks should be considered enterprise-wide, focusing on those with most significant impact to firm’s value; should consider risks other than insurance risk (insurance hazard risk, financial risk, operational risk, and strategic risk); this ensures efficient use of resources
- should quantify risks where possible and incorporate correlations between risk
- to max firm value, risk management strategies are evaluated for trade-off between risk and return
- strategies must be implemented to avoid, mitigate or exploit risks
4 steps of ERM process
- diagnose - conduct high level risk assessment to identify risks posing most significant threats to firm’s value
- analyze - model critical risks where possible and incorporate dependencies between risks; select and calculate risk metrics
- implement - implement activities to manage risk such as avoiding risk, reducing its occurrence, or mitigating its effects
- monitor - monitor plan compared to expectations and update/improve continually
Insurers face the following risks
- Insurance hazard
- Financial (Asset)
- Operational
- Strategic
Insurance hazard risk
Insurance hazard: risk assumed by insurer in exchange for premium
U/W: risk due to non-CAT losses from current exposures
Accumulation/CAT: risk due to CAT losses from current exposures
Reserve deterioration: risk due to losses from past exposures
*reserving risk will be greater for long-tailed lines because threat of reserve deterioration is much greater
*CAT modeling uncertainty will be larger for Home lines etc
Financial (Asset) risk
risk in insurer’s asset portfolio related to volatility in interest rates, foreign exchange rates, equity prices, credit quality, and liquidity
AKA asser risks due to market, liquidity and credit risks
Operational risk
risk associated with execution of company’s business (execution of IT systems, policy service systems, etc)
Strategic risk
risk associated with making the wrong or right strategic choices (risk of choosing the wrong plan give current and expected market conditions)
first step of ERM is to diagnose risks that pose greatest potential threats to insurer, for different insurers
some of these risks will be common but others may be firm specific relating to the type of business written, their liabilities, and firm-specific operations
different types of strategic decisions Enterprise risk model can help insurer with
- determine capital requirements to support its risk or maintain credit rating
- decide between different reinsurance programs to manage risk
- identify risk sources that significantly contribute to most adverse outcomes and cost of capital to support them
- planning growth
- managing asset mix
- valuing companies for M&A
key elements that differentiate the quality of model
- should reflect relative important of different risks and reflect dependencies between them
- modelers should have deep knowledge of fundamentals of the risks
- modelers should have trusted relationship with senior managment
- includes math. Techniques to reflect relationships among risks
- shows balance between risk and reward for diff strategies
- reflects uncertainty of output of other models being incorporated (CAT or macroeconomic models)
4 aspects of parameter risk
- estimation risk
- projection risk
- event risk
- systematic risk
estimation risk
data is used to estimate form and parameters of distributions; estimation risk is risk that form of distribution and parameters don’t reflect true form and parameters
projection risk
there are changes over time (ie trends) and projection risk is added uncertainty of projecting changes from time of data into the future as well as uncertainty in loss development
event risk
event risk is added uncertainty to loss due to large unpredicted events outside of company’s control
systematic risks
impact a large number of policies and can’t be diversified away, such as macroeconomic factors like inflation
this adds uncertainty
multivariate normal distribution for ERM
had low tail dependency between risks so combined results from ERM will be unrealistically stable
tail dependencies the ERM should incorporate
inflation would impact both UW losses and loss reserve development
extreme events would cause large losses for both Home and Auto
if high tail dependency between 2 risk, should use
copula with greater joint probabilities in tail would be appropriate
ERMs help insurer to find optimal level of capital that balances efficiency and prudence
common approaches for setting requirements:
- holding enough so that probability of default is remote (conservative and protects mainly policyholder)
- holding enough to max insurer’s franchise value (protects both policy and shareholder)
- holding enough to continue service renewals (since renewals tend to be more profitable)
- holding enough so that insurer not only survives major CAT but thrives in aftermath
problem with holding enough so that probability of default is remote
Default is unlikely outcome in far right tail of distribution and ERM model is least reliable in this portion of the distribution
Mainly protects policyholders
Shareholders are impacted prior to default point and capital requirements should consider protecting shareholders
holding enough so that insurer not only survives major CAT but thrives in aftermath: example for setting capital requirement
To withstand CAT and continue afterwards, assume minimal capital requirement is 6x 95th TVaR, this ensures that average 1-in-20 yr event only depletes 1/6 of capital, using this company would have enough capital to continue following CAT
Setting capital requirements
capital must be sufficient to: sustain current U/W, provide for adverse reserve changes, provide for declines in assets, support growth, and satisfy regulators, rating agencies and shareholders
-ERMs help insurer to find optimal level of capital that balances efficiency and prudence
Essential elements of mathematical enterprise model are
U/W risk
reserving risk
asset risk
dependencies/correlations
U/W risk
loss freq, loss severity, pricing risk, parameter risk, CAT modeling uncertainty
- pricing risk: instability in U/W results arises from variations in prems as well as losses; U/W cycle contributes heavily to pricing risk and needs to be modeled over mult periods
- parameter risk: estimation risk, projection risk, event risk, systematic risk
- CAT modeling uncertainty: CAT exposure is incorporated into ERM by incorporating proprietary CAT models; these models include considerable uncertainty relating to prob of various events and amount of damaged caused by event; ERMs need to incorporate this uncertainty into CAT model results
Reserving risk
risk of reserves developing other than anticipated; reserve uncertainty affects both the amount of required capital and time for which capital must be held; traditional tech are deterministic, but stochastic is becoming more popular and is important for understanding reserve variability
Asset risk
important asset classes to model include bonds, equities, real estate, exchange rates
key aspect=modeling scenarios consistent with historical patterns
good ERM can help balance asset and U/W risk since risk profiles on liability and asset sides of balance sheet are often different in terms of duration so can optimize use of capital by offsetting insurance risks with investment risks (duration matching)
dependencies/correlations
sources of dependency: inflation rates, interest rates, equity values are correlated and should be modeled as such; U/W cycles, insurance loss trends, reserve developments are correlated across LOBs and with each other; CATs and other kinds of event risk are often correlated across LOBs
modeling dependency
modeling tail dependency in extreme events is crucial when developing ERM; tail dependency is often modeling using copulas
