2VALUE AT RISK VALUE AT RISK Definition : The expected maximum loss ( or worst loss ) over a target horizon within a given confidence intervalEvolved through the requirements of SEC to provide a minimum capital requirement in provision for risks taken by financial institutions
3Models to estimate VaR Historical Market Data assumption is that historical market data is our best estimator for future changes and thatasset returns in the future will have the same distribution as they had in the pastVariance – Covariance (VCV)assuming that risk factor returns are always (jointly) normally distributed and that the change in portfolio value is linearly dependent on all risk factor returnsMonte Carlo Simulationfuture asset returns are more or less randomly simulated
4Historical Market Data involves running the current portfolio across a set of historical price changes to yield a distribution of changes in portfolio value, and computing a percentile (the VaR).BenefitsSimple to implementdoes not assume a normal distribution of asset returnsDrawbacksrequires a large market databasecomputationally intensive calculation.
5VALUE AT RISK (VAR) Market Value x Risk Variability Confidence Level = Worst Loss
6DEFINITION : Market Value : Risk variability : Current market value of the respective transaction to be managedMark to market at the end of time horizonRisk variability :Usually the standard deviation of the risk to be managedThe higher the variability the higher is VAR
7DEFINITION : Time Horizon : Confidence Level : Time period to be considered correspond to time required for corrective actions as losses starts to developAnnualizedThe longer the time horizon the higher is VARConfidence Level :Confidence level of loss occurringThe higher the confidence level the higher is VAR
8Example :Current market value of transaction USD 100,000 paid by 3 months usance L/CStandard deviation of Rp/USD is 10%Time horizon is 3 monthsConfidence level is 95%VAR = 100,000 X 0.10 X X 0.95= USD 4, Maximum expected loss
9WEAKNESSES OF HISTORICAL SIMULATION Past is not prologue, history does not always repeat it selvesTrends in the data, since all data are treated equal, despite the fact that some periods might experience higher volatilityFor new assets or market risks no historical data available
10MODIFICATIONS OF HISTORICAL SIMULATION Weighting the recent past more, assuming that the recent past is a better predictor of the immediate future than the distant past ( can be using indexes to adjust each return based on its timeline)Combining historical simulation with time series models, by fitting a time series models through the historical dataVolatility updating, by comparing past volatility with recent volatility and then adjusting the past return accordingly e.g. past volatility = 0.5, recent volatility 0.75, past return 10%, recent predicted return is 0.75/0.5* 10%=15%
11Variance – Covariance (VCV) Basic assumption is that the risk factors are normally distributedHence are the returns also normally distributed
12Variance – Covariance (VCV) VAR formula for VCV :σ = standard deviation of the riskSigma for 95% confidence level is 1.645Sigma for 99% confidence level is 2.33
13Exchange rate (USD/Rp) EXAMPLE :Exchange rate (USD/Rp)Loss / Gain (p)Probability (r)d2 = r * p29.600-4000.0254.0009.700-3002.2509.800-2000.0502.0009.900-1000.20010.0000.40010.10010010.20020010.30030010.400400Average =S =d2 =d =With 95% confidence interval the VAR = * Rp = Rp The maximum potential loss for exchange rate risk is Rp
14ASSESMENT OF V-CV METHOD Strength :simple to compute after making assumptions about the distribution of returns and inputted the means, variances and covariances or returnsWeaknesses :Wrong distributional assumptions, if it turns out that the returns a re not normally distributed and the outliers are higher, computed VaR can be lower that actual VaRInput error, if data used to calculate are for example based on historical data, which is not reflecting the current situationNonstationary variables, happens if the underlying assumed correlation does not hold anymore, e.g. interest rate is adjusted by FED
15Monte Carlo Simulation Decide on N, the number of iterations to perform.For each iteration:Generate a random scenario of market moves using some market model.Revalue the portfolio under the simulated market scenario.Compute the portfolio profit or loss (PnL) under the simulated scenario. (i.e., subtract the current market value of the portfolio from the market value of the portfolio computed in the previous step).Sort the resulting PnLs to give us the simulated PnL distribution for the portfolio.VaR at a particular confidence level is calculated using the percentile function.For example, if we computed 5000 simulations, our estimate of the 95% percentile would correspond to the 250th largest loss; i.e., ( ) * 5000.
16STRENGTH OF MONTE CARLO SIMULATION Does not need to rely on historical data, those historical data can still be used as benchmark and then adjust accordinglyDoes not need to assume normal distribution for the returnsCan be used for any type of portfolio including options or option like securities
17WEAKNESSES OF MONTE CARLO SIMULATION Needs to estimate the probability distribution for all the market risk variables that we want to considerNumber of simulations that need to be run on the model will be substantially large
18MODIFICATION ON MONTE CARLO SIMULATION Scenario simulation, only likely combination are run through the modelMonte Carlo simulation with Variance-Covariance method modification, assuming normal distribution for the returns
19Indications on method to use For Value at Risk for portfolios, that do not include options, over very short time periods (a day or a week) and normality can be assumed, the variance-covariance approach does a reasonably good job.If the risk source is stable and there is substantial historical data (commodity prices, for instance), historical simulations provide good estimates.In the most general case of computing VaR for nonlinear portfolios (which include options) over longer time periods, where the historical data is volatile and non-stationary and the normality assumption is questionable, Monte Carlo simulations do best.
20LIMITATIONS OF VaR :Return distributions cannot always be correctly predictedHistory may not be a good predictorNonstationary correlationsOnly looking at the downside risk (negative side of risk)Best for calculating short term riskDifficult to use for comparing different investments
21VaR can lead to suboptimal decision Overexposure to risk, managers will tend to be more bold in making risky investments while actually the rest of 5-10% probability of incurring risk might be hugeAgency problem, because VaR usually uses past data, managers who knows about irregularity in the volatility can misuse them for his own advantage.
22VAR is best used for financial institutions, since they are dealing with short term assets which are related to the common market risksVAR for nonfinancial institutions should be used as a secondary measure, unless it is for short term assets.