29. Risk measurement and reporting Flashcards
Describe how the risk identification ‘brainstorming’ approach can be extended to obtain a subjective assessment of risk exposure
Subjective assessment of risk exposure
- Estimate probability and severity of each risk separately
- Assign a number from the scale 1-5
- The product of probability and severity=> ranked 1-25
- Allows risks to be ranked and prioritised
- Carried out with and without possible risk controls
How can a model be used to assess a risk event/exposure?
- Distribution assigned to both Frequency and severity of a risk event
- Define an event
- Use historic events to calculate a probability distribution for that event
- Alternatively - Frequency of the event defined and used to determine the loss parameter
- Decide stochastic vs deterministic model
- Availability of data=> Influence which model is used
- Important when considering rare events
What the different ways of valuing risk?
SRC
- Scenario analysis
- Stress testing
- Stochastic modelling
- Reverse stress testing
- Combined stress and scenario testing
What steps should be involved in a scenario analysis to evaluate operational risk?
4
- Group risks into categories.
- Develop a plausible adverse scenario of risk events for each group of risks, which is representitive of all risks in the group
- Calculate the consequences/costs of the risk event occurring for each scenario. Involving input from senior staff.
- Calculate the total costs of all risks represented by the scenario
What categories may operational risk be divided into for the purpose of scenarios analysis?
6
- Fraud
- Loss of key personnel
- Mis-selling of financial products
- Calculation error in the computer system
- Loss of business premises
- Loss of company e-mail access
What is stress testing?
- Deterministic method of modelling adverse event over a period of time
- Commonly used to model extreme market events but can be applied to other risks.
- It models the impact of event, but not probability
In relation to market risk:
* It involves subjecting a portfolio to extreme market movements by:
1. by radically changing the underlying assumtions and characteristics
2. Change asset class correlations and volatilities
Designed to:
* Identify weak areas=> investigate effect of localised stress situation
* Different combinations of correlations + volatilities => Impact of major market turmoil=> all parameters
* Ensure consistency between correlations while they are stressed
A provider of unit-linked investment bonds has constructed a model to investigate the stress scenario of a sustained reduction in equity market values.
List factors that would need to be built into this model
The model would need to allow for the impact of the sustained reduction in equity market values on:
* Income received from fund management charges
* Persistency of existing bonds
* New business volumes
* Regulatory capital requirements
* Value of shareholders’ interests
* Probability of any gurantee biting
* Other economic conditions=>interest rates, inflation + investment returns on other assets
What is reverse stress testing?
- Construction of a severe stress scenario
- Only just allows the company to continue
- To fulfil its strategic business plan
- Financial or non-financial
- Although extreme, MUST BE PLAUSIBLE
How can a stochastic model be used to evaluate a particular risk?
- Variables that give rise to risk => treat as RV with probability distributions
- Model must be dynamic- with full interactions and correlations between variables
- Model run to determine amount of capital to just avoid ruin
- With a given probability
A stochastic model used to evaluate risk may become impractical to run.
Outline three approaches to limiting the ideal scope of a stochastic model in order to make the model more practical
- Restrict the time Horizon that the model projects
- Limit the number of variables that the model models stochastically
- Carry out a number of runs with a single different stochastic variable
- Followed by a single deterministic run using all the worst-case scenarios together
Why can the phrase 1 in 200-year event be misleading, in relation to risk and setting riskbased capital requirements?
- Misleading to non-experts
- If the risk event has just occurred=> will occur again in 200 years
- Rare events in practise such as stock market crashes are occurring more frequently than the assumed probability indicates
- 1 in 200-year combined event not the same as combining individual 1 in 200 year events
How does the overall capital requirement relate to the individual capital requirements of a group of risks?
- Fully dependent=> Overall cap = sum of individual risk cap
- Fully independent=> Overall cap< sum of individual risk cap. Under certain assumptions: Overall risk is square root of sum of squares of individual risks
- Partially dependent=> Overall cap requirement< sum of individual cap
- The difference is the diversification benefit
What are examples of likely correlations between risks?
- Inflation risk = expense risk
- Equity markets inverse to inflation rates
- Equity market falls= higher withdrawal rates on unit linked savings products
- Operational risks correlated to other risks
- Longevity risk strongly negatively correlated to mortality risk
What methods exist for aggregating partially dependent risks?
- Stochastic model
- Correlation matrices
- Copulas
What is a correlation matrix and how can it be used in the assessment of risk?
- Specifies the correlations between all pairs of individual risk factors being modelled
What is a copula and how can it be used to model risk?
- F(Marginal CDF)= Joint CDFs
- Method of calculating joint probabilities of risk
- P(RBonds>L,REquity<M)= Joint probability
- Different copulas are used to describe different degrees of dependence between RV
- Including dependence in the tails of the distributions
- USEFUL for modelling tail risk=> Capital requirements for extreme events
How can liability risk be measured?
- Analysis of experience- A/E
- NB consistent classification and measurement of the risk event+ exposure to risk
What is Value at Risk (Var)?
- Maximum possible loss
- On a portfolio
- Over a given time period
- With a given degree of confidence
- Absolute amount or relative to a benchmark
What are the disadvantages/drawbacks of using VaR as a measure of risk?
3
- Calculated assuming a normal distribution of returns. NOT TRUE IN PRACTISE
- VaR can be used with a different distribution. Data is sparse particulalarly in the tails=> difficult to fit accurately
- VaR does not quantify the size of the tail=> loss might be past VaR confidence level
What is tail VaR or TVaR?
- Expected shortfall below a certain level, given that shortfall has occurred
- e.g. if average loss on the worst 5% of possible outcomes is 5 mil, then the TVaR is 5 mil for the 5% tail
What is a risk portfolio and what might it contain?
- Means of categorising the various risks to the company
- Against each risk=> likely severity + probability
- Product=> idea of relative importance of the various risks
- Extended to explain how each risk was dealt with
- Details for mitigated risks=> revised assessment of risk remaining
- Retained risks=> Details of control, risk owner, need for management.
Why is regular risk reporting important within a business?
FRAUD CRIME
- Financing
- Rating agencies
- Attractive investors
- Understand better
- Determine appropriate control systems
- Changes over time
- Regulator
- Interactions
- Monitor effectiveness of controls
- Emerging risk identification
Explain why, if risk is being managed at the enterprise level, it is necessary to have a coherent system of risk reporting across the whole enterprise.
- Each unit given risk exposure allocation
- Benefits of diversification rely on each business unit taking on exposure allocated
- Necessary each unit report on exposure they are taking on
- NOT done in a consistent way= more capital required
What are the two modelling steps of the determining the amount of capital to hold?
- A model used to determine risk event at the required level of probability
- Stochastic model used to determine this
- Second model=> determine the consequences/ costs of the risk event determined â– Deterministic model likely to be used
What are the issues when assessing risk-based capital requirements?
- Ruin P expressed over single year or entire run off of the business?
- Modelling more than 2 varaibles stochastically impractical => method of identifying correlation between varibales is needed = Correlation matrix
- Effect of multiple risk events influenced by diversification and their interactions
- Operation risk is highly subjective => Look beyond past events for risk scenarios
- Past data used with caution => censequences of rare events
Describe the key considerations when designing risk reporting, including typical contents of such reports.
- Risk reporting should be clear and relavant, linking to company’s risk appetite/tolerances in order to allow effective decision-making
- Split according to risk types and operating units, and include:
1. Key Risk Indicators (KRIs)
2. Summaries of main risk areas in tables/graphs, showing likelihood and severity of each risk.
3. Traffic light systems to highlight areas of concern. - Must contain the appropriate level of detail for the audience, balancing completeness with clarity and simplicity.