CH:18 Modelling Flashcards
How do actuaries use models?
(6)
- Set premiums of charges for insurance products
- To determine the financing strategy for a benefit scheme
- Aid risk management
- Determine the capital requirement
- Valuing options and guarantees
- Understanding potential variability of experience
What is a deterministic model?
A deterministic model is one where the parameter values are fixed at the outset, so the result of running the model is a single outcome
Q
What are the advantages and disadvantages of a deterministic model? (4/1)
++ More readily explainable to a non-technical audience
++ Clearer what economic scenarios have been tested
++ Usually easier to design and quicker to run
++ Users can be blinded by too complex models
- Requires thought as to what economic scenarios are to be tested
What are the steps of running a deterministic model
(10)
- Specify the purpose of the investigation
- Collect, group and modify the data
- Choose form of model
- Identify assumptions and assign values to these assumptions using past experience and appropriate estimation techniques (basis of model)
- Construct a model on expected cashflows
- Check that the goodness of fit is acceptable
- Try to fit a different model if the first model doesn’t fit well
- Run the model
- Perform sensitivity tests
- Summarise the results
What is a stochastic model?
A stochastic model estimates parameters by assigning a probability distribution to them. The model then models the output through simulations which results in a distribution of likely outcomes
What are the advantages and disadvantages of a stochastic model? (2/2)
++ Test a wider range of scenarios
++ Assess more complex problems such as guarantees, options and investment mismatching
*- More complex programming
*- Longer to time design, test and build the model
What are the steps of running a stochastic model
(10)
- Specify the purpose of the investigation
- Collect, group and modify the data
- Choose suitable density functions for stochastic variables
- Specify correlation between variables
- Ascribe values to assumptions that are not stochastic in nature
- Construct a model on expected cashflows
- Check that the goodness of fit is acceptable
- Try to fit a different model if the first model doesn’t fit well
- Run the model multiple times using a random sample from the chosen density functions
- Summarise the results showing their distribution
What are they requirements of a good model?
VARIABLE CRISPS CARD
V - Valid
A - Adequately documented (e.g., key assumptions should someone take it up later)
R - Rigorous (i.e., produce realistic output under a range of situations)
I - Inputs to parameter values appropriate
A - Arbitrage-free (i.e., take account of business and economic environment)
B - Behaviour reasonable
L - Length of time to run/ cost not too high
E - Easy to understand
C - Communicable workings and output
R - Reflects risk profile of purpose
I - Independent verification of outputs
S - Sensible joint behaviour of variables
P - Parameters allow for all significant features
S - Simple but retain key features
C - Clear results
A - A range of implementation methods
R - Refineable
D - Developable
What factors will affect the approach in which a model is obtained?
(5)
- Level of accuracy required
- in-house expertise
- Number of times a model is used
- Desired flexibility in model
- Cost associated
What are the operational issues when building a model?
SCARCER FILES
S - Simple but retain key features of the model
C - Clear results
A - Adequate documented
R - Range of implementation methods
C - Communicable workings and output
E - Easy to understand
R - Refineable and developable
F - Frequency of cashflows vs run time required
I - Independently verifiable
L - Length of run times not too long
E - expenses not too high
S - Sensible joint behaviour of variables (where variables or assumptions are linked)
What does the risk dicount rate allow for
2
- return required by the company
- level of statistical risk attached to the cashflows under the particular contract