Mod 21: Use of models in ERM Flashcards
State the three main uses of models in a practical ERM context ©
Main uses of models in a practical ERM context
1. to enable the actuary or risk manager to give an organisation appropriate advice so it can manage its risks in a sound financial way (avoiding sub-optimal decisions)
2.
to assist in the day-to-day running of the company, eg assist in decision making
3. to help determine suitable checks and controls on the business
List the main uses of models for ERM decision-making ©
Main uses of models for ERM decision-making
1. pricing of products or services
2. assessment of the economic value of the company
3. estimation of the possible volatility of future profits and earnings
4. determination of capital adequacy requirements, including regulatory requirements and internal capital assessments
5. projection of the future capital or solvency position
6. assessment of the effect of risk management and mitigation techniques on both profits and capital requirements
7. assessment of the effect of other strategic decisions, eg changes in investments or new business strategy
8. evaluation of projects ©
Outline three sources of risk when producing mathematical representations of real-world processes
Three sources of risk
1. Model risk (a type of operational risk) is risk arising from the use of an inappropriate or inaccurate model when assessing or managing risks.
2. Parameter risk refers to the use of inappropriate or inaccurate parameters or assumptions within such models.
3. Stochastic uncertainty arises from the randomness of a finite set of observations.
This reduces as the number of observations increases. ©
Describe how parameter uncertainty might be allowed for if least squares regression has been used to determine the parameters
©
Allowing for parameter uncertainty
A multivariate normal distribution for the parameters (using an available covariance matrix) can be used to generate (dynamic) parameters, rather than using a static set of parameters to carry out the simulations.
One approach to determining the confidence intervals for the parameters is estimate a joint distribution for the parameters by using the following process:
1. fit a model using the T data points available simulate
2. T data points using the model
3. re-fit the model to the simulated data points
4. record the parameter values
5. repeat the process a large number of times, starting with the original data set each time.
©
Describe how a ‘wrong’ model might come to be used ©
How a ‘wrong’ model might come to be used
- inappropriate projection of past trends into the future – perhaps due to erroneous, incomplete or heterogeneous data
- the selection of an inappropriate underlying distribution (eg inappropriate skew and/or kurtosis) – perhaps due to insufficient data and/or not enough alternative candidate distributions
- the number of parameters being chosen without reference to:
- the need to avoid over-simplification and the risk of implicit assumptions - the principle of parsimony (where there is a choice, the optimal model selection is the one with the fewest parameters, as this should lead to more stable projections)
©
State the main requirements of a good generic actuarial model
Hint, recall SCARCER FILES from Subject CP1
©
Requirements of a good generic actuarial model
Simple but retains key features
Clear results
Adequately documented
Range of implementation methods
Communicable workings and output
Easy to understand
Refinable and developable
Frequency of cashflows – balance accuracy vs practicality
Independent verification of outputs
Length of run not too long
Expense not too high
Sensible joint behaviour of variables
©
State the additional requirements for a good model in an ERM context ©
Additional requirements for a good model in an ERM context
The modelling scheme overall should:
1. be amenable to an analysis of the impact of parameter uncertainty or incorrectly-specified parameter values
2. exhibit behaviour in simulations that is consistent with the past, but should not exclude plausible future scenarios that might be quite different from historical patterns
3. reflect the dynamics of the organisation, now and as expected to develop in the future, allowing for wider external factors
4. be defined to be comprehensive across all important well-defined risks
5. produce outcomes that are balanced, ie not unduly exaggerated or unduly smooth.
Any shortcomings of the modelling scheme should be clearly stated. ©
Outline the typical steps required to successfully develop and apply a model
Steps in developing and applying a model
1. Specify the purpose of the investigation
2. Collect data and group or modify it if necessary
3. Choose the form of the model, identifying its parameters and variables
4. Select the appropriate time period (reliability of output vs speed)
5. Estimate the required parameters and any correlations between them
6. Check the goodness of fit is acceptable, amend model if not
7. Ensure that the model is able to project all required cashflows and other outputs, including interactions between them
8. Run the model using the estimated variables or stochastic simulations
9. Output the results in an appropriate format (eg summarised) 10. Assess the sensitivity of results to different variable values
Outline factors that should be considered when making corporate decisions using ERM models
©
Other factors in corporate decision making using ERM models
1. inputs: information, data, assumptions, parameters
2. model calculations: cashflows, projections, simulations
3. outputs: deterministic + ranges / sensitivities, stochastic distribution
4. review and discussion
5. corporate risk policy and risk appetite (eg utility function) 6. qualitative factors, judgement and intuition of decision makers
6. whether to focus only on risk preferences of shareholders, ie specific risks diversified away / systematic risk rewarded via a risk premium
7. or whether both external and internal perspectives are important
8. whether to base decisions on economic value = PV of all future shareholder profits and economic value added = increase in economic value (as a return on capital) less weighted average cost of capital
9. or whether to use more than one metric for decision making ©
