18. Models Flashcards
Define a model point
Representative policy in a relatively homogeneous group of policies
What factors would affect number of model points chosen?
- Availability and computing power
- Variability of contracts
- Complexity of in force contracts
- Company age
- Stochastic vs Deterministic model
- Importance of investigation
- Time available
- Sensitivity of results to more/fewer model points
List items you’d expect to see in a model point
- Product type
- Sum assured
- Guaranteed accrued bonus (w/profits)
- Premium
- Term
- Age
- Sex
- Commission level
- Duration in force
- Number of policies
- Distribution channel
- Renewal expenses
- Initial expenses
Why would it be difficult to estimate sales volumes for IP?
- Impact of economic and political factors hard to predict
- Markets under-insured, high risk of getting sales assumption wrong
- Changes in competition can affect sales
- Potential claims mis-handling affecting reputation and sales volume
Name and describe the 4 types of models
- Single policy profit test- projects expected cash and profit flows for a single policy from inception. Model for pricing and product design.
- New business model- Projects expected cash and profit flows from future sales of new business. Used to assess future capital requirements for new business and return on capital achieved on future sales
- Existing business model- Projects expected cash and profit flows from existing in force business. Assess embedded value and test solvency.
- Full model- sum of new and existing. Asses impact of future management decisions on financial development of company
What are the requirements of a model?
- Valid, rigorous enough for purpose and adequately documented.
o Rigorous = produce realistic results under a wide range of circumstances and conditions. - Workings of model should be easy to appreciate and communicate. Results displayed clearly.
- Capable of reflecting the risk profile of the business being modelled.
- Parameters should allow for all significant features of the business being modelled.
- Inputs to the parameter values should be appropriate to the business being modelled. Take into account the business and economic environment in which it is operating.
- Exhibit sensible joint behavior of model variables.
- Outputs of model should be capable of independent verification for reasonableness & communicable to those to whom advice will be given.
- Must not be overly complex or time consuming to run.
- Capable of development and refinement.
What are the basic features of a life model?
- Involves projecting cashflows
- The cost of setting up supervisory reserves and required solvency margins must be allowed for to calculate profit flows.
- Proper allowance must be made for guarantees and options- likely to need stochastic modelling for this.
- Allow for interactions and correlations between variables.
- Frequency of cashflow projection must be short enough to produce reliable results.
List features that can be parameters in a model
- Mortality rates
- Surrender rates
- Rates at which policies are paid up
- Future expenses
- Future expense inflation
- Income component of investment yield
- Capital component of investment yield
- Proportion of profit allocated to ph (w/profits)
- Amounts and mix of new business
- Tax basis
Why might stochastic model be preferred?
- Costing guarantees and options
- Want to see likely distribution of outcomes and not single estimate
- Interaction between variables can be explicitly included, allowing effect of interactions to be assessed
- Need to estimate a probability
Why might deterministic model be preferred?
- Used with sensitivity testing to get an approximation to stochastic result
- If result would be similar or more prudent than stochastic result
- Check on stochastic model
Drawbacks of stochastic models
- Sensitive to chosen parameter values leading to risk of spurious accuracy
- Time and computing constraints
Explain how stochastic models can be calibrated
This refers to setting parameter values.
Risk neutral / market consistent:
* Aim to replicate market prices of financial instruments as closely as possible using an adjusted risk neutral probability measure
* Choose nr of financial instruments of which you know the price. Model is built that can project these cashflows in a range of scenarios.
* Parameters chosen such that average PV of cashflows is sufficiently close to the known market price.
Real world:
* Use assumptions which reflect realistic long-term expectations and which consequently also reflect observable real world probabilities and outcomes.
* Determine model parameters using our expectations of the future.
Explain the financial economic approach to modelling
Assumes market-consistent values for parameters.
Investment return is RFR since potential additional return from more risky investments is cancelled out by their increased risk
Assets valued at market value
Market consistent values of L can be calculates as current market values of risk free assets that match their cashflows.
Other parameters may not have market-consistent indicators for setting the assumptions e.g.:
* Expense assumption may be determined by reference to expense agreements available in the market.
* Difference between current market yields of equivalent fixed-interest and index-linked bonds gives an indication of market’s expectation of future inflation.
* May be actual market valuations for liabilities such as endowments if they are traded.
How can a model be inaccurate?
- Incorrect programming of product structure
- Some products may not be modelled
- Known imminent changes (e.g. fiscal regime) not allowed for
- Ignored policy options
- Incorrect parameters
- Interactions of variables incorrectly dynamic
- Optimistic new business assumptions
- Product changes/launches
- Incorrect data
- Misrepresentative model points
- Used deterministic where stochastic needed
- RDR not making enough allowance for cashflow variance
- Time unit in model too long
- Projection period short
- Miscellaneous software bugs
