18 - Models (1) Flashcards
What is the prime objective of building a life insurance model?
The prime objective is to enable the actuary advising a life insurance company to provide appropriate advice so the company can be run in a sound financial way, assisting in day-to-day operations and providing checks and controls on its business.
What are the key requirements of a good model?
A good model should be:
* Valid, rigorous, and adequately documented
* Capable of reflecting the risk profile of the products
* Inclusive of all significant business features
* Based on appropriate inputs for the economic/business environment
* Easy to understand and communicate, with clear results
* Sensible in joint variable behavior
* Verifiable and communicable outputs
* Not overly complex unless required
* Capable of development and refinement
What does it mean for a model to be ‘rigorous’?
A rigorous model produces realistic and useful results under a wide range of circumstances, ensuring reliability across various scenarios.
What is a ‘model point’ in life insurance modeling?
A model point is a data record representing a policy or group of policies, containing key characteristics, used as input for a modeling program to project results for a representative subset of the business.
Why might policies be grouped into model points?
Policies are grouped to reduce the number of data points to a manageable level, making modeling computationally feasible while ensuring each group is homogeneous enough to produce similar results.
What information would you expect in a model point for an endowment assurance product?
For an endowment assurance (single life):
* Product type coding
* Sum assured
* Guaranteed bonus accrued (if applicable)
* Premium
* Term
* Age
* Sex
* Commission level
* Duration in force
* Number of policies in the group
* Possibly distribution channel and renewal expenses
How is the validity of grouped model points checked?
Validity is checked by comparing model outputs from ungrouped data to grouped data for a block of policies, ensuring the results are acceptably close.
What factors should an actuary consider when assuming new business volumes and mix?
Recent new business production, trends, intended marketing changes, planned product launches, and imminent legislative or fiscal changes.
How does political commitment affect long-term care insurance (LTCI) business?
Political commitment influences LTCI take-up and persistency by determining state-provided alternatives, such as the level of state benefits and eligibility rules for state funding based on personal wealth.
Why is it difficult to estimate sales volumes for income protection business?
Sales depend on unpredictable economic and political factors, markets are often under-insured, competition is keen, and mishandled claims can unpredictably affect reputation and sales.
What are structural components in a life insurance model?
Structural components replicate real-world features like cashflows (premiums, benefits, expenses) and their calculation methods, based on product types, assets held, and the model’s purpose.
When should a parameter be included in a model?
A parameter should be included if the decision reached would change with different values of that parameter; if the decision is unaffected, the parameter is irrelevant and can be excluded.
List at least five parameters that could affect a life insurance company’s financial results.
Examples include:
* Mortality rates
* Surrender rates
* Future expense levels
* Investment yield (income component)
* Taxation basis
What is meant by the ‘basis’ of a model?
The basis is the full set of assumptions (parameter values) used in the model, typically tested with a central basis and alternative bases to assess sensitivity.
Why should a model exhibit sensible joint behavior of variables?
Variables like inflation and interest rates are interdependent in reality, so the model must reflect these relationships to produce realistic outcomes.
How can model outputs be independently verified for reasonableness?
By reconciling with supervisory valuations, past model runs, ratio checks, or a simple ‘back of the envelope’ model for order-of-magnitude checks.
What factors influence the number of model points chosen?
Factors include:
* Computer power availability
* Contract variability/complexity
* Company age
* Model type (stochastic/deterministic)
* Investigation importance
* Time available
* Sensitivity of results to model point count
What are the four types of life insurance models mentioned?
- Single policy profit test model
- New business model
- Existing business model
- Full model office
What is the focus of a single policy profit test model?
It projects expected cash and profit flows from a single policy from its issue date, focusing on annual profit emergence for pricing and product design.
What is the purpose of a new business model?
It projects cash and profit flows from future sales to assess capital requirements and overall return on capital from new business.
What does an existing business model assess?
It projects cash and profit flows from in-force business to evaluate intrinsic value (embedded value) and test solvency.
What is the role of a full model office?
It combines new and existing business projections to assess the impact of future management decisions on the company’s financial development.
What are ‘discretionary benefits’ in a model?
Benefits whose levels depend on company decisions, not fixed policy conditions.
How does a model account for supervisory reserves?
It includes notional cashflows: negative ‘increase in reserves’ when funded from cashflow or capital, and positive ‘release of reserves’ at claim/maturity.
What formula calculates total profit in a life insurance model?
Profit = Premiums + Investment Income - Payments - Commission - Expenses - Tax - (Vt - Vt-1), where Vt is the supervisory reserve at the end of year t.
Why must asset and liability models be consistent?
To reflect real-world interactions, ensuring accurate solvency projections.
What is a dynamic model?
A model where asset and liability parts interact as in reality.
Why are stochastic models vital for guarantees and options?
They assess the impact of variable outcomes by simulating multiple scenarios.
What are the advantages of a stochastic approach over deterministic?
- Assigns probability distributions to parameters
- Calculates positive liabilities for guarantees/options
- Models dynamic variable interactions
- Estimates probabilities (e.g., insolvency)
What are the disadvantages of stochastic modeling?
- Time and computing constraints
- Sensitivity to parameter assumptions risks spurious accuracy
When might a deterministic approach be appropriate?
- When similar to stochastic results
- For quick checks on stochastic results
- In simple cases using formulas
Would you use a deterministic or stochastic model to assess losses from a 1918-level influenza epidemic? Why?
Deterministic, as only one specific scenario’s outcome is needed, not a distribution.
Would you use a deterministic or stochastic model to estimate the cost of a minimum maturity value on a unit-linked product? Why?
Stochastic, as it costs a financial guarantee with variable outcomes.
What is risk-neutral (market-consistent) calibration in stochastic modeling?
It adjusts parameters to replicate market prices of financial instruments using a risk-neutral probability measure.
What is real-world calibration in stochastic modeling?
It uses realistic long-term expectations and observable probabilities to project future asset/liability values.
Why is a monthly projection frequency often chosen?
It balances reliability with practicality, allowing monthly/quarterly solvency checks.
What is the typical projection period for a full company model, and why?
3-5 years, as longer periods enter the ‘tunnel of doubt’ with uncertain new business assumptions.
List five ways a model could be inaccurate.
- Incorrect product structure programming
- Ignoring policy options
- Optimistic new business assumptions
- Misrepresentative model points
- Deterministic used when stochastic needed
What is the financial economic (market-consistent) approach to modeling?
It sets parameter values to be consistent with market values, valuing assets at market value and liabilities to match risk-free asset cashflows.
How does the financial economic approach handle non-financial risks like mortality?
It’s subjective due to no active market; assumptions may use industry statistics or market indicators.
