Chapter 11 - Introduction to risk modelling

Question 1

Issues in risk quantification

Answer 1

Extreme events - learn from past; updated emerging risk register

Data limitations - internal and external

Interdependence risks

Unquantifiable risk - risk map for simplification

Question 2

Main methods of risk quantification

Answer 2

Enterprise risk - Dynamic financial analysis
All enterprise risks are modelled to produce projected financials

Enterprise risk - Financial Condition Reports
Report into solvency and future development

Underwriting risk - Underwiting modelling or reviews
Financial impact of errors during underwriting

Market risks - VaR, TVaR, Interest Rate Models, Scenario tests
Attempts to value market as whole, individual securities, and relationship between them

Credit risk - Credit risk models
Also consider non-quantitative

Liquidity risk -Asset Liability Modelling
Cash vs short term liabilities

Operational risks -internal/external loss data, scenario analysis, simulations

Question 3

Types of correlation

Answer 3

Linear correlation
Pearson’s rho = Pxy = Cov(X,Y) / Stdx.Stdy

Rank correlation
Spearman’s rho - linear correlation of dist functions
Kendall’s tau - looks at concordant and discordant pairs

Tail correlation

Question 4

Deterministic modelling approaches

Answer 4

Sensitivity analysis
Understand risks faced; insight into dependence of output on subjective assumption; satisfy supervisory req

Scenario analysis
Multiple inputs varied at a time; decide on scenarios to to be tested; Establish impact on risk factors; Review scenarios for relevance; take action based on results

Stress testing
Focus on extreme scenarios or changes in assumptions; Top-down or Bottom-up stress tests; links to BC and CM and bak-testing under Basel

Question 5

Stochastic modelling approaches

Answer 5

Historical simulation (bootstrapping)

Forward looking approaches
Monte Carlo simulation
Data-based VS (focus on modelling key variables, rather than factors that drive them)
Factor-based approach (causal links between variables, described explicitly within model)

Pseudo-random numbers

Market consistency
Check model output against market observations –> may be different due to short-term supply demand effects