Chapter 17 - Use of models in ERM Flashcards
Modelling uncertainty
Stochastic uncertainty: Arises from randomness of finite observations
Model risk: Arise from inappropriate/inaccurate model
- Inappropriate projection of past trends
- Select inappropriate distribution
- Number of parameters chosen
Parameter risk: Arise from inappropriate/inaccurate parameters/assumptions
- Also based on finite set of observations
- Simulate parameters through multivariate normal
- Determine confidence interval for parameters by refitting model to simulated data
Requirements of a good model
APD R3S2 WAC AIN
Appropriate Parameters
Param Dynamic
Adequately Documented
Refine & Develop
Results Clearly Displayed
Reflect Distribution of Business
State Shortcomings
Sufficiently Rigorous
Workings easily Appreciated and Communicable
Allows for features of Business
Independently Verifiable
Not too complex or costly, time consuming
Reasons why an organisation might build a model for ERM decision-making
A3 PEPE CAR
Assessment of:
- strategy, new business, changes
- risk mitigation techniques on profits/results
- economic value of company
Pricing of products
Evaluation of projects
Projection of future capital or solvency requirement
Estimate earnings
CAR - Capital Adequacy Requirements
ERM modelling process
Specify purpose of investigation
Collect data - group/modify
Choose form of model (select param/variables)
Estimate parameters and correlations
Check goodness of fit
Ensure model projects correctly everything required
Run model using selected estimate variables
Output results in appropriate format
Assess sensitivity of deterministic variable values
Corporate decision making using ERM models
Outputs considered relative to risk appetite
External and internal risk preference should be considered
Economic Value Added (EVA) can be compared to average cost of capital - mitigation resulting in positive incremental positive increase in EVA
Models should not be used blindly - qualitative evaluation also important
