Chapter 43 - The Risk Management Process (2) Flashcards
Discuss the use of risk registers/portfolios.
It is a way of categorising risks
Will contain:
- Risk name
- Probability of occurrence
- Impact
- Likely importace = Probability x Impact
- How it has been dealt with (mitigated, diversified, retained, transferred + revised assessment of residual risk)
Simple business: Subjective impact figure like 1-5
Complex business: Risk analysis model to determine impact. Some statistical measures available but likely to be subjective
+ Show where management action is needed
+ Expose concentration of risks
Name the risk measures.
- Asset measures
- Historical tracking error
- Forward looking tracking error
- Active money position
- Liability measures
- Analysis of experience
- Like for like analysis
- Value at Risk
- Tail Value at Risk / Expected shortfall
Explain the risk measurement tools used for asset risks.
-> Risks involved with active investment management
- Historical tracking error
- Annualised S.D. of the difference between the fund and the benchmark, based on observed relative performance - Forward looking tracking error
- Estimate of the annualised S.D. of the expected future difference between the fund and the benchmark, if the current fund structure does not change.
- Quantitative modelling techniques
- Assumptions- Correlations
- Likely future volatilities relative to the benchmark
- Active money position
- Sum of 0.5 [Windex-Wfund]
Explain the risk measurement tools used for liability risks.
- Analysis of experience = Actual/expected
- Like-for-like analysis of changes in the population between periods
- Consistency in classification and measurement of risk
- Consistency in exposed to risk
Discuss the use of VaR as a risk measure.
- VaR generalises the likelihood of underperformance by providing a statistical measure of downside risk
- Quantifies the maximum loss expected to occur over a period at a given confidence level
- Can be absolute or relative to a benchmark
Disadvantages
- Implicit assumptions about distribution of returns (assume normality)
- Fatter tailed at extreme values
- Does not quantify the size of the tail
- Can be an understatement -> misleading
- Further into tails -> less data -> more arbitrary the choice of the underlying probability
- Not suitable for portfolios exposed to credit risk, systematic bias or derivatives (stochastic modelling, Monte Carlo simulations)
Discuss the use of ES or TailVaR as a risk measure.
- ES below a certain level
- If L is a specific percentile point it becomes tail VaR
- (1-p) TailVar is the expected shortfall in the pth lower tail
- TailVar = ES given there is a shortfall
+ Easy to understand
- No attention to returns in excess of L
- Focus on less profitable outcomes so may lead to a very prudent value
What are the characteristics of high impact-low probability risks.
- They can only be diversified in a limited way
- They can be transferred to an insurer or reinsurer (Catastrophe, Stop Loss)
- They can be mitigated by using management control procedures e.g. disaster recovery planning
What are the drawbacks of using VaR?
- Implicit assumption about the distribution of returns (normality)
- Size of tail is not quantified
- The further you move into the tail -> less data -> more arbitrary the choice of the probability
- Cannot be used for portfolios exposed to credit risk, systematic bias or derivatives -> stochastic modelling or Monte Carlo simulation
- Fatter tailed at extreme values or may be skewed
- Can be misleading (understatement)
- Results sensitive to assumptions made about parameters
How would the forward looking VaR be calculated to measure the risk of underperformance.
- Will be relative to a benchmark (e.g. R10m or 3%)
1. Assume a specific distribution like normal
2. Estimate parameters (can use quantitative modelling techniques for volatility and correlations)
3. Use Monte Carlo simulation to generate the shortfall relative to the benchmark for several cases, e.g. 1000
4. Find level of underperformance below which there is only a 5% frequency.
What is evaluation of risks and how would we do it?
- Evaluation of risk = determining how much capital to hold against the risk
- Depends on
- Appropriateness of a mathematical model
- Nature of the risk
- Amount of info/data
- Methods
- Scenario analysis
- Stress testing
- Stochastic modelling
Discuss how scenario analysis is used as a risk evaluation technique.
- Scenario analysis is a way of risk evaluation where a mathematical model is not appropriate, since the model will require too many subjective parameters.
- Risk exposure are grouped into broad categories
- They behave the same or are triggered by the same event
- E.g. financial fraud or system errors - A plausible adverse scenario is determined for each group
- Scenario is representative of all the risks in the group
- Scenarios can be interlinked between risk groups - The financial consequences are calculated for each scenario
- Direct an indirect costs
- Redress pay to those affected
- Correcting errors and systems
- Regulatory fines and fees
- Opportunity costs
- Midpoint of a range of values - Total cost is the sum of the consequences of all the scenarios, accounting for interlinked scenarios.
Discuss how stress testing can be used as a risk evaluation tool.
- Stress testing = subject a portfolio of assets or a block of business comprising of assets and liabilities to extreme market movements by radically changing the assumptions and characteristics to determine the portfolio’s sensitivities to predetermined risk factors.
- Used when
- Lots of data available
- Mathematical model is appropriate
- Want a deterministic model of the business
- Types
- To identify weaknesses in the portfolio, you look at the different combinations of correlations and volatilities to investigate the effect of local stress tests.
- Gauge the impact of major market turmoil affecting all parameters, while ensuring consistency between correlations when they are ‘stressed’
- Also
- Consider asset volatilities and correlations since both tend to increase during extreme market movements
- Scenario must apply to specific business -> early intervention & weakness identification
Discuss how stochastic modelling can be used to evaluate risks.
- Extension of stress testing
- All variables that give rise to risk are included as probability distributions
- A full set of dynamic interactions between variables must be specified
- Can determine capital needed to avoid ruin at a given probability level
- Complex to build and running time is long
Limiting the scope
- Restrict the duration
- Limit nr of variables that are modelled stochastically
- Do many runs, each with a different stochastic variable. Then one deterministic run with all worst case scenarios together. This will determine the interaction between variables.
What are the steps in stochastic modelling?
- Specify the purpose
- Collect, group and modify data
- Determine assumptions and density functions + specify correlations
- Assign values to fixed variables
- Determine time period
- Determine model points
- Construct model based on E(CFs)
- Check goodness of fit
- Change model if necessary
- run model many times
- Produce a summary
What are the advantages of stochastic modeling?
+ assess a wider range of economic scenarios
+ can allow for uncertainty
+ can model guarantees and optons
