14-26 Risk Management Flashcards
Contrast decentralized and centralized risk governance systems
Decentralized - places responsibility for execution within each unit of the firm
-risk mgmt is in the hands of those closest to the risk
Centralized - aka enterprise risk mgmt - places responsibility for execution within one central unit of the firm
- better view of overall risk borne by firm and considers correlations of risk between units within firm
- offers economies of scale
4 Qualities of high quality Risk Governance
- transparent
- establish clear accountability
- cost efficient in use of resources
- effective in achieving desired outcomes
3 types of financial risks
Explain
- market risk - usually largest component of risk - related to changes in i rates, exchange rates, equity prices, commodity prices, etc (are tied to supply and demand)
- credit risk - second largest risk component - risk of loss caused by counterparty’s default
- liquidity risk - risk of loss caused by inability to liquidate a position quickly at fair price (can be estimated by bid-ask spread - narrow vs wide - and average trade volume)
7 types of non-financial risks
Explain
- operational risk - risk of loss caused by firm’s systems failure or disaster events like terrorism (internal or external events)
- settlement risk - aka Herstatt risk - is risk that I paid my promised amount (e.g. in a swap) while counterparty defaults and doesn’t pay his amount.
- risk is reduced by using exchanges (netting and reduce credit risk) vs OTC transactions or continuously linked settlements (CLS) where settlements are made in defined window of time - model risk - risk that poor quality of inputs and assumptions give poor quality results (think option pricing models)
- sovereign risk - is a form of credit risk - that depends on willingness and ability of government to pay
- regulatory risk - risk that regulations adversely change or uncertainty of how transaction will be regulated
- accounting and legal risk - uncertainty of how rules can change and adversely impact transactions (think derivatives which are hugely subject to changing rules)
- Political risk - changes in government that lead to any of the above risks
3 types of other risks
- ESG (environmental, social and governance) risk - is risk that company harms the env’t, human resources or corporate governance which leads to decline in firm value
- performance netting risk
- settlement netting risk
How is market risk measured?
standard deviation for any asset price or asset’s excess returns (aka active risk)
beta for stocks
duration for bonds
delta for options
What is VAR?
5% VAR of $1,000
measure of minimum expected loss or maximum expected loss (dual interpretation)
5% probability that expected minimum loss is $1,000 (no less than)
95% probability that expected maximum loss is $1,000 (no greater than)
Think 5% on the left tail of return distribution
3 methods of computing VAR?
- analytical method aka variance-covariance method
- historical method
- monte carlo method
4 advantages and disadvantages of analytical VAR
advantages
- standard dev and correlations needed are readily available from standard sources i.e. Ibbotson
- easy to compute unlike MC, so don’t need a lot of computer power
- allows modeling the correlations of risk
- can be applied to shorter or longer time periods
disadvantages
- need to assume normal distribution of returns so can’t perfectly model leptokurtotic returns (fat tails) OR skewed returns from options (modification can be done by delta-normal method but results are less than satisfying)
- with large portfolios with many assets, hard to estimate standard deviations and need a LOT of correlation pairs
- problem with standard sources is that they provide LT inputs which can’t be easily adjusted to calc daily or weekly VAR
- still working with historical data which is not forward looking, e.g. correlations can change quickly
- can’t easily use for assets with non-linear risks
assumptions behind analytical method
- returns are normally distributed
- returns are serially independent
- return distributions can be calculated using modern portfolio theory (market is in equilibrium, risk averse investors, borrow and lend at Rf, etc)
4 advantages and disadvantages of historical VAR method
advantage
- easy to calculate and explain to clients
- does not assume normal distribution
- can be applied to different time periods
- no need to assume serial independence, unlike analytical method
- no need to assume modern portfolio theory holds, unlike analytical method
disadvantage
- assumes that historical pattern will repeat in future (not forward looking). This can be huge mistake for assets that change characteristics quickly overtime, such as options and bonds
- requires huge database of historical returns data which is costly to maintain
- not flexible enough to take portfolio changes into account on a real-time basis
- doesn’t allow sensitivity tests to determine how VAR is impacted by assumed changes in st dev, correlations, asset weights, etc.
5% and 1% VAR is how many standard deviations below the mean?
1.645 and 2.33
How to transfigure monthly VAR into annual VAR?
weekly VAR and daily VAR into annual VAR?
annual VAR into monthly and weekly and daily VAR?
E(R)
- multiply 12 or divide 12 (monthly - annual)
- multiply 52 or divide 52 (weekly - annual)
- multiply 250 or divide 250 (trading days - annual)
- multiply 22 or divide 22 (trading days - month)
standard deviation
-multiply or divide by sq root
formula for VAR
Z = [ R(var) - E(R) ] / standard dev
Solve for R(var) which is in %VAR
Then multiply by portfolio value to get $ VAR
Z = standard deviations below mean, based upon alpha provided. e.g. 1.645 for 5% alpha
3 advantages and disadvantages of Monte Carlo VAR method
advantages
- does not need to assume normal distribution of returns so can analyze non-linear risk and options with skewed returns
- more flexible than other methods bc analyst can specify inputs and probability distributions
- more likely to generate outlier possibilities than historical method (potential disadvantage is analyst may ignore them thinking they are unlikely to happen)
disadvantages
- output is only as good as quality of inputs. e.g. analysts get different VARs on same portfolio
- requires a great deal of mathematical modeling capability
- more variables mean more simulations that have to be run
Why would two analysts get different VAR estimates on a portfolio using Monte Carlo VAR method?
using different inputs and assumptions
- 5% vs 1% alpha (or significance level)
- different time horizon (daily VAR vs monthly VAR)
- etc
4 overall advantage and disadvantage of VAR
advantages
- industry standard for risk management and required by many regulators
- easy to understand because it is one number
- can be applied to capital allocation decisions
- can compare operating performance of different assets with different risk characteristics
disadvantages
-some VAR methods are difficult and costly
-doesn’t show maximum possible loss so can give false sense of security
-looks at the absolute $ risk in portfolio but doesn’t measure trade-off between return and risk so not useful to rank investment alternatives where E(R) and risk rankings are different (can only rank investments whose E(R) are equal)
-no one VAR method and different methods give different VARs for same portfolio
-can’t capture non-market risk
-sensitive to significance level chosen to represent worse case return
-can’t easily determine VAR if using a lot of structured products that are not actively traded and whose price histories aren’t known
generally more tedious to calculate as number of assets in portfolio increase
other measures that complement VAR
- Incremental VAR
- cash flow at risk
- earnings at risk
- tail value at risk
- credit VAR
- stress testing
- back testing (comparing actual results across time with projections)
- sharpe ratio
define stress testing and two forms of it
complement to VAR which measures impact of unusual events that might not be reflected in VAR
-2 types are scenario analysis and stressing models
List types of scenario analysis and stressing models
Contrast scenario analysis and stressing models
scenario analysis types: stylized scenarios, actual extreme events and hypothetical events
stressing model types: factor push analysis, maximum loss optimization, worst-case scenario
Contrast:
- scenario analysis measures impact of simultaneous movements in multiple inputs (or factor) on portfolio value
- stressing model is an extension of scenario analysis that focus on adverse outcomes
- suffers from same problems like incorrect inputs, inability to accurately measure by-products of major factor movements, and user bias
define the 3 types of stressing models
factor push analysis - simple stress test where analyst pushes factors to worst combination of possible factors and measures impact on portfolio
maximum loss optimization - uses more sophisticated mathematical and computer modeling to find this worst combination of factors
worst-case scenario - is the worst case the analyst thinks is likely to occur (i.e., worst outcome of a combination of factors)