Chapter 18 Flashcards
What is the prime objective of building a model?
To enable the actuary advising a LIC to give appropriate advice so that it can run in a sound financial way.
What are the requirements of a good model?
VARIABLE CRISP CARDS
What is a model point?
A data record fed into a computer as an input for the modelling program.
It will represent either a policy or a group of policies, containing data on the most important characteristics of the policy.
Tell me more about model points.
The underlying business modelled will comprise a wide range of different policies and these may need to be brought together into a manageable number of relatively homogeneous groups.
* Groupings need to be sufficiently homogeneous, such that each policy in a group is expected to produce a similar result when the model is run.
* It is sufficient for a representative policy in each group to be run through the model, result obtained and scaled up to give result of the total set of policies in the group.
* The representative single policy in each group is termed a model point.
* A number of model points can be used to represent the whole of the underlying business.
What needs to be considered when constructing model points?
- The validity of the groupings needs to be checked - compare the model outputs from the ungrouped and grouped data.
- If the model includes NB, actuary must decide on NB volume and mix for each product. Look at most recent NB production, trends, intended market changes, planned new product launches and any imminent and pertinent legislative/fiscal changes.
What factors influence sales volumes and persistency of a LIC?
- Economic morale
- Government provision of welfare and tax
What influences the number of model points used in a model?
- Computer power and availability
- Variability of contracts sold
- Complexity of contracts IF
- Age of IC
- Whether the model is stochastic or deterministic
- Importance of the investigation
- Time available
- Sensitivity of result to using more/less model points
What should the components of a model allow for?
All features of business being modelled that could significantly affect advice being given.
What are the structural components of a model?
Real-world features of a model we are trying to replicate, e.g. CFs generated by products being modelled and how to calculate them. Also refers to the parameters to include.
When should a parameter be included into a model?
If the decision reached would alter for different values of parameter.
What are some examples of parameters to include in a life office model?
- Mortality/morbidity rates
- Surrender rates
- Paid-up rates
- Future expense levels/inflation
- Investment yield
- NB volumes and mix
- Surplus allocated to PHs
- Tax basis
What is EV?
It represents the value of the future profit stream from the company’s EB together with the value of any net assets separately attributable to SHs.
What are some general outputs from a model?
- All future revenue account and income statement entries (premiums, expenses, investment income and claims)
- All BS entries (supervisory reserves, total assets, required solvency margins)
- Various info on policy portfolio (number of IF policies, amounts of NB, withdrawals)
How can you check a model against data independent of the model?
- Reconcile results with those from last supervisory valuation
- Reconcile with results from the last time the model was run
- Perform simple ration checks on future years’ results
- Use a simple model to compare the full model’s results with
What is an important check if you are updating rather than creating a model?
Compare last set of model projections with actual experience.
What are the different types of models?
- Profit test model
Projects the expected cash and profit flows from a single policy from the date of issue
Key model for pricing and product design - NB model
Projects all expected cash and profit flows arising from future NB sales
Useful for assessing future capital requirements for NB and overall capital return - EB model
Projects all expected cash and profit flows from all EB a company has in force at a particular time
Important for testing intrinsic value of EB and also testing the solvency position of company’s EB - Full model office
Projects all expected cash and profit flows from both EB and NB
Fundamental importance in assessing impact of future management decisions on future financial development of company
What should be considered when constructing a model?
- Frequency and total projection period
More frequent: more reliable output (dangers of spurious accuracy)
Less frequent: model can run and obtain results faster - Interactions
CFs need to allow for interactions/dynamic links
Particularly when assets and liabilities are modelled together
Assets and liabilities should be consistent: model should be dynamic (A+L parts should interact as they do in reality)
Dynamism: company’s investment strategy should respond to changing conditions - Guarantees and options: likely a stochastic modelling facility will be necessary
- Supervisory reserves and solvency requirements
Cost of setting up supervisory reserves and required solvency margins needs to be allowed for to calculate profit flows
Notional CFs: increase and release in reserves
Actual CFs: premiums, claim payments, expenses, investment income, tax
Real and notional CFs combine to form profit flow - investment income from supervisory reserve will be included as a positive in the profit flow
What are the advantages of using a stochastic model?
- Good quality of results
- Tests a wider range of economic scenarios
- Allows a PDF to be assigned to one or more of the unknown future parameters
- Future parameters may be assumed o vary together as a dynamic set, which is particularly useful for modelling with-profit business
What are the advantages of using a deterministic model?
- Easily explainable to non-technical audience
- Clearer what economic scenarios have been tested
- Easier to design and quicker to run
What are the disadvantages of using a stochastic model?
- Time and computing constraints
- Sensitivity of results to assumed values for parameters
- Risk of spurious accuracy
- Increased difficulty in interpreting and communicating results
What are the disadvantages of using a deterministic model?
- Requires thought as to the range of economic scenarios to be tested
- Danger that certain scenarios, which could be detrimental to the company, are not identified
- Subjectivity required in choosing the scenarios that should be tested and the probability assigned to different outcomes
When is a stochastic model preferred?
- To cost guarantees and options
- Wish to see distribution of outcomes, not a single estimate
- Interaction between variables can be explicitly included
- Need to estimate a probability
- Identifying potentially high-risk future scenarios
When is a deterministic model preferred?
- If only a single answer is required
- With sensitivity testing in order to get an approximation to a stochastic result
- If result obtained would be very similar to/more prudent than a stochastic model’s result
- As a check on a stochastic model
- Within a stochastic projection to avoid having a nested stochastic model
What approaches are there for setting parameters when the economic assumptions vary?
Risk-neutral market consistent calibration
- Used for valuation purposes
- Parameters chosen to replicate known market prices of financial instruments, using risk-neutral probabilities
- Used for valuation of business - especially options and guarantees
Real world calibration
- Use LT realistic assumptions and known real world probabilities
- Typically used to project into the future
