Concept Review: VAR, Risk Management

VAR is an important risk measurement tool for any corporation with an enterprise risk management system. In conjunction with other measures, it provides companies a way of estimating the probability of a loss over a specific time period. Alternatively, it can also measure the probable maximum loss. For example: 5% VAR is $35,000 over the next 3 months. This would indicate that the company faces a 5% probability of incuring AT LEAST a $35,000 loss over the next 3 months. Alternatively: 95% VAR $35,000. This would indicate that the company is 95% confident the loss will not exceed $35,000 over that time period. There are 3 methods of calculating VAR: #1. Analytical VAR - Assumes normal distribution - Uses expected returns and variance to determine probable loss over a time period Advantages of Analytical VAR - Can be applied to future time periods - Considers the correlation of risk factors - Allows a company to see the impact of each business unit and maximize return for a given level of VAR. Disadvantages of Analytical VAR - Assumes normal distribution, and as a result, will have a tendency to underestimate probable loss, particularly in areas where stale pricing or illiquidity is an issues. Stale pricing results biased standard deviation downwards - Difficult to incorporate when dealing with many risk factors - Any others??? Calculated As : [(Rp - (z x s.d))] x Vp #2. Historical VAR - Uses historical values to determine VAR by ranking returns from highest to lowest, and identify the lowest x % of returns. The highest of these would be the 1-day or 1-month x % VAR (Depends on time period). Advantages of Historical VAR - Easy to calculate and understood - Does not require assumption of Normal Distribution - Can be applied to different time periods Disadvantage of Historical VAR - Since it’s based on historical data, assumes that characteristics of these historical values will continue in the future. #3. Monte Carlo VAR - Uses the same calculation procedure as Analytical VAR - Identifies a risk factor model where various risk factors that impact VAR and included in a model a distribution of expected returns and variance is generated. Advantages of Monte Carlo VAR - Incorporates correlation of risk factors - Incorporates any return distribution Disadvantages of Monte Carlo VAR - Assumes risk sensitivities do not change over time - Must make thousands of assumptions about the inputs and correlations Overall, the advantage of using VAR in a ERM system is that it allows a firm to maximize return for a given level of VAR and to see how each business unit adds to VAR and effects total firm risk. The disadvantage is that it does not account for outliers ie: the magnitude of the losses. In addition, there are many factors, both financial and non-financial that VAR may not be able to incorporate, particularly in situations of illiquidity where VAR should be adjusted for liquidity since illiquid assets have a tendency to underestimate variance due to stale pricing and as a result, VAR will be understated. To compensate, firms should use other measures in conjunction with VAR such as: 1. Incremental Var - how the addition of a risk factor will impact overall VAR 2. Maximization Loss Optimization - involves identifying the risk factors that would have the largest potential impact 3. Worst Case - Take each risk factor to the maximum loss position to see the overall impact 4. Cash Flow At Risk & Earnings At Risk (similar to VAR measures) 5. Tail Value At Risk (TVAR) - evaluates the magnitude of the potential loss ie: accounts for outliers (Value in the tails of the distribution).

stress testing and scenario analysis missing

Yes… forgot that part… equally important but pretty straight forward concepts…

One other advantage of analytical VAR is that it can be interpreted the same regardless of the asset being measured

ChiTownShane Wrote: ------------------------------------------------------- > One other advantage of analytical VAR is that it > can be interpreted the same regardless of the > asset being measured I believe that’s a general advantage of VAR, not just analytical. You can compare performance across business units even with different assets.

You can’t do peer comaprison with VAR Historical method - non-parametric, no need to assume anu distribution

atpr Wrote: ------------------------------------------------------- > You can’t do peer comaprison with VAR According to Schweser, one of the benefits of VAR is that you can calculate the measurement for each business unit regardless of assets used for comparative purposes. Unless I’ve missed something here.

^Yes a firm can b/c they will use the same method. What ATPR is referring to is, Firm A can’t compare its VAR to Firm B and Firm C’s VAR b/c more than likely all 3 calculated it differently.