Chapter 15 Introduction to risk modelling

Question 1

Potential issues when quantifying risks LUIE

Answer 1

• Limitations on data
• Interdependencies of risks
• Unquantifiable risk
• Extreme events

Question 2

Merits of Pearson’s rho WEEN

Answer 2

• Widely used
• Easy to calculate – you just need observations
• Elliptical joint distributions of marginal distributions assumed
• Normal and t distribution can directly use the results

Question 3

Considerations of Correlations and Volatility CADOS TACOS

Answer 3

• Concentration measured
• Aggregation of risks considers correlation
• Diversification of risks assessed
• Optimise portfolio by considering correlation
• NB: Changes in correlation in stress scenarios
• Tolerance setting (max deviation from VAR)
• Assessment of risk (VAR)
• Controls (risk limits)
• Optimization of risk (MVPT)
•

Question 4

Merits of sensitivity analysis DUCS

Answer 4

• Dependence of the model on assumptions detected
• Understanding of variables’ effect on a model outcome
• Concentration risk detected
• Supervisory requirements met

Question 5

Merits of scenario analysis and stress testing CHES B CHEP

Answer 5

• Catastrophe management done
• High impact low likelihood scenarios can be assessed
• Evaluate impact of plausible risk events
• Supplement traditional statistical models
• Business continuity management and Back-testing
• Complex process
• Hypothetical events must be plausible
• Exhaustive list of scenarios may not be found
• Probabilities not assigned to the scenarios

Question 6

Merits of bootstrap modelling PALP ORIA

Answer 6

• Probability distribution not specified
• Applicable to various situations
• Large amounts of data not required
• Past data represented without parameterisation
• Outcomes limited to past data
• Reliance on past data
• Indicative nature of past data assumed
• Autocorrelation of past data not considered

Question 7

Merits of Monte Carlo simulation WACSI FT

Answer 7

• Widely available computer packages can do MC simulation
• Accuracy can be increased by repeated simulation
• Complex financial instruments can be modelled
• Simple math – easily understood
• Interdependence of risks can be simulated
• Full range of possibilities may not be represented
• Time consuming to do many simulations

Question 8

Features of rank correlation measures BINC

Answer 8

• Binary measure – 1 if ranks match, 0 if not
• Independent of marginal distributions
• Non-linear strictly increasing transformations retains value (co-monotonic functions)
• Combined with copulas in applications

Question 9

Limits of linear correlation HENI

Answer 9

• Heavy tailed distribution cannot use it
• Elliptical joint distributions required
• Non-linear relationships cannot be measured
• Incompatible with marginal distributions – joint distribution cannot be created from it