VaR I PDF

Valuations and Risk Models – I
© EduPristine – www.edupristine.com
© EduPristine For VaR-I (2016)
VaR Methods
© EduPristine For VaR-I (2016) 1

Introduction to Risk
 Risk can be broadly defined as the degree of uncertainty about future net returns
• Credit risk relates to the potential loss due to the inability of a counterpart to meet its obligation
• Operational risk takes into account the errors that can be made in instructing payments or settling
transactions
• Liquidity risk is caused by an unexpected large and stressful negative cash flow over a short period
• Market risk estimates the uncertainty of future earnings, due to the changes in market conditions
 Broadly the standard deviation of the variable measures the degree of risk inherent in the variable
 Say the standard deviation of returns from the assets owned by you is 50% and the standard
deviation of returns from assets I own is 0%. We can say that risk of my assets is zero
I own risk-less assets as My assets are very risky

the standard deviation of returns as the standard deviation of
of my assets is 0% returns of my assets is 50%.

Value at Risk (VaR)
 Value at Risk (VaR) has become the standard measure that financial analysts use to quantify
this risk.
 VAR represents maximum potential loss in value of a portfolio of financial instruments with a given
probability over a certain time horizon.
 In simpler words, it is a number that indicates how much a financial institution can lose with
probability (p) over a given time horizon (T).
 Say the 95% daily VAR of your assets is $120, then it means that out of those 100 days there would
be 95 days when your daily loss would be less than $120. This implies that during 5 days you may
lose more than $120 daily.
There may be a day out of 100 when your loss is $5000,

which means VAR doesn’t tell anything about the extent to which we can lose

Visualizing VAR
Confidence (x%) ZX%

0.45 Probability
90% 1.28
0.4
95% 1.65
0.35
97.5% 1.96
0.3
99% 2.32
0.25
95% daily-VAR
0.2
0.15
0.1
0.05 Z values
0
-4 -3 -2 -1 0 1 2 3 4
Mean = 0
 The colored area of the normal curve constitutes 5% of the total area under the curve.
 There is 5% probability that the losses will lie in the colored area i.e. more than the VAR number.

Measuring Value-at-Risk (VAR)
0.45
0.4
0.35
0.3
0.25
0.2
VAR X % (in %)  Z X % *
0.15
0.1
0.05
0
-4 -2 0 2 4
Mean = 0
 ZX% : the normal distribution value for the given probability (x%) (normal distribution has mean as
0 and standard deviation as 1)
 σ : standard deviation (volatility) of the asset (or portfolio)
 VAR in absolute terms is given as the product of VAR in % and Asset Value:
VAR  VARX % (in %) * AssetValue
 This can also be written as:
VAR  Z X % * * AssetValue

Measuring Value-at-Risk (VAR)
 VAR for n days can be calculated from daily VAR as:
VaR (n days) (in %)  VaR (daily VaR) (in %) * n
 This comes from the known fact that the n-period volatility equals 1-period volatility multiplied by
the square root of number of periods(n).
VaR (n days) (in %)  ZX% * * Asset Value * n
 As the volatility of the portfolio can be calculated from the following expression:
 portfolio  wa2  a2  w 2b  b2  2wa w b *  a *  b * ab
 The above written expression can also be extended to the calculation of VAR:
VaR portfolio (in %)  wa2 (%VAR a ) 2  w 2b (%VAR b ) 2  2wa w b * (%VAR a ) * (%VAR b ) * ab

Question 1
 Asset daily standard deviation is 1.6%
 Market Value is USD 10 mn
 What is VaR (%) at 99% confidence?

Solution
 Daily VaR = 0.016 x 10 x 2.33 = 0.3728 mn

Question 2
 What is the VaR value for 10 day VaR in the earlier case?

Solution
 10 day VaR = 0.3728 x (10)^0.5 = 1.1789

Question 3
 What is the daily portfolio VaR at 97.5% confidence level?

• Investment in asset A is $40 mn
• Investment in asset B is $60 mn
• Volatility of asset A is 5.5% and asset B is 4.25%
• Portfolio VaR if correlation between A and B is 20% ?

Solution
 VaR(A)(in %) = 5.5 x 1.96 = 10.78%; VaR(B)(in %) = 4.25 x 1.96 = 8.33%;
 Portfolio VaR = [(40 x 0.1078)2 + (60 x 0.0833)2 + 2x0.1078x0.833x40x60x0.20]0.5 = 7.22 mn

Extended Question 3.1
 Portfolio VaR if
• If correlation between A and B is Zero?
• What if correlation is 1?
• Or if -1?
 What are the implications?

Question 4
 Market Value of asset $10 mn

 Daily variance is 0.0005
 What is the annual VaR at 95% confidence with 250 trading days in a year?

Solution
 Daily VaR = 10 x (0.0005)0.5 x 1.65 = 0.36895 mn
 Annual VaR = 0.36895 x (250)0.5 = 5.834 mn

Question 5
 For an uncorrelated portfolio what is the VaR if:

• VaR asset A is $10 mn
• VaR asset B is $20 mn

Solution
 The VaR comes out to be 22.36 mn

Value-at-Risk Measurement Methods
VaR
Linear Valuation Method Full Revaluation Method
Delta Normal Historical Monte Carlo

Method Simulation Simulation

Delta Normal VAR: Linear and Non-Linear Assets
 Linear: When the value of the delta is constant for any change in the underlying
• Primarily in the case of forwards and futures we have linear assets
• The method to calculate VAR for linear assets is called Delta Normal method
• Delta Normal method assumes that the variables are normally distributed
 Non Linear: When the value of the delta keeps on changing with the change in the underlying
asset
• Options are non-linear assets, where delta-normal method cannot be used as they assume the linear payoff
of the assets
• To calculate the VAR for non-linear assets, full revaluation of the portfolio needs to be done
• Monte Carlo methods or Historical Simulation are commonly used to fully revaluate the portfolio

Delta Normal VAR: VaR for Linear Derivatives
 Linear Derivatives: Payoff diagrams that are linear or almost linear:

• Forwards, futures
linear
Payoff
0
k Share/asset price
Long position
 Delta of Derivative: Change in price of Derivative to change in underlying asset
 For example:
• The permitted lot size of S&P CNX Nifty futures contracts is 200 and multiples thereof.
• So VAR of Nifty Futures contract is 200 * VAR of Nifty Index
VARLinear Derivative   *VARUnderlying Risk Factor

Delta Normal VAR: VaR for Non-Linear Derivatives
 Main reason for difference is the shape of the
payoff curve Option
Price value
 For Delta Normal VAR
• A linear approximation is created
• Approximation is an imperfect proxy for the portfolio
• Computationally easy but may be less accurate.
• The delta-normal approach (generally) does not work for Slope = delta
B
portfolios of nonlinear securities.
• E.g.; Options VAR = Delta of Option * (VaR at Zx%) A Asset/Stock price
 Consider a portfolio of options dependent on a
single stock price, S. Define: f ' ' ( x0)( x  x0) 2
f ( x)  f ( x0)  f ' ( x0)( x  x0) 
2
S ΔP
x  Delta ( ) 
S ΔS
• Approximately: P   S  S x
• For Many Underlying variables: P  S

i
i i  xi

Question 6 (Linear Assets)
 If the daily VaR at 5%of Nikkei is USD 0.8 mn and you have 100 lots of Nikkei contract.
 Calculate annual VaR at 95% confidence for your portfolio assuming 250 days?

Solution
 Solution = 0.8 x 100 x (250)0.5 = USD 1264.911 mn
 Here, delta = 100, because for every 1 unit change in the Index Nikkei, the futures price will change
by 100 units because the lot size is 100

Question 7 (Non-Linear Assets)
 If the value of stock is 100 and the value of the put option at 110 is 20. 10 units change in the
underlying brings in change of 4 units change in the option premium. If the annual volatility is 0.25.
Calculate daily VaR at 97.5% assuming 250 days?

Solution
 Solution: Daily volatility = 0.25/(250)0.5 = 0.0158; Daily VaR = 100 x 0.0158 x 1.96 = 3.099;
 Daily VaR of option = 0.4 x 3.099 = 1.239

Quantifying Volatility in VaR Model

Quantifying Volatility in VAR Model: Fat Tails
 There are two explanations for the Fat Tails
• Conditional Mean: Mean changing over the period of time 0.45
0.4
• Conditional Volatility: Volatility changing over the period of 0.35
time 0.3
95%
daily-VAR
 Market conditions may cause the mean and variances 0.25
0.2
to change over the period of time, which leads to 0.15
fat-tailed distributions 0.1
0.05
0
-4 -2 0 2 4
 The fat-tailed unconditional distribution can be broken down into two conditional distributions,
either with similar means and different variances or similar variances and different means.
 Many a times when we observe marked differences between the estimated and actual volatilities,
it’s a result of regime switching which means that the average volatility in the market has now
changed too much when compared to the previous estimate
Risk Management is all about understanding tails of distribution

Regime Switching
High Vol Estimate
1 n
   (R i  E ( R))2
2
n i 1 Actual volatility
30bp/day
18bp/day
Low Vol Estimate
Unconditional
1 n
volatility    (R i  E ( R))2
2
n i 1
10bp/day

Alternative Measure
 In this case we observe distinct periods of high and low volatility.

 A more accurate measure of risk would be to measure the risk or volatility in these high and low
risk areas separately, and then base our forecast depending on whether or not we are currently in
a high or low volatility period.
 This implies, we would have the problem of assessing whether we were currently in a high or low
risk period.

Cyclical Volatility Of Financial Markets
 Volatility
• Time Varying
• Sticky
• Predictable
• Magnitude of recent changes is informative
• More information in recent past than distant data

Estimation of Volatility
 Let xi be the continuously compounded return during day i (between the end of day “i-1” and end
of day “I”).
 Let σn be the volatility of the return on day n as estimated at the end of day n-1.
 Variance estimate for next day is usually calculated as:
• Variance = average squared deviation from average return over last ‘n’ days
 x 
n
2
i x
Variance  i 1
n 1
• Mean of returns (x-bar) is usually zero, especially if returns are over short-time period (say, daily returns). In
that case, variance estimate for next day is nothing but simple average (equally weighted average) of
previous ‘n’ days’ squared returns.
n

 ix 2
Variance  i 1
n 1
What if the volatility is dependent on the values of volatility observed in the recent past?
What if they also depend on the latest returns?

EWMA & GARCH
This topic has already been covered in Quants Lectures

EWMA Model
 In an exponentially weighted moving average model, the weights assigned to the u2 decline
exponentially as we move back through time.
 This leads to:  2n   2n 1  (1   ) u n21
 Apply the recursive relationship:
 n2    n2 2  (1   )u n2 2   (1   )u n21

 n2  (1   )u n21  u n2 2   2 n2 2
 Hence we have:
 m i 1 2 
 2
n  (1   )    u n  i   m n2 m
 i 1 
• Variance estimate for next day (n) is given by (1-λ) weight to recent squared return and λ weight to the
previous variance estimate.
• Risk-metrics (by JP Morgan) assumes a Lambda of 0.94.

EWMA Model (cont.)
 Since returns are squared, their direction is not considered. Only the magnitude is considered.
 In EWMA, we simply need to store 2 data points: latest return & latest volatility estimate.
 Consider the equation:  t21  (1  0.94) t2  0.94 t2
 In this equation, variance for time ‘t’ was also an estimate. So we can substitute for it as follows:
 t21  (1  0.94) t2  0.94(1  0.94)t21  0.94 t21 

 t21  0.06 * t2  0.94 * 0.06 * t21  (0.94 * 0.94 t21 )
 What are the weights for old returns and variance?

 λ is called ‘Persistence factor’ or even “Decay Factor”. Higher λ gives more weight to older
data (impact of older data is allowed to persist). Lower λ gives higher weight to recent data
(i.e. previous data impacts are not allowed to persist).
 Higher λ means higher persistence or lower decay.
 Since, (1- λ) is weight given to latest square return, it is called ‘Reactive factor’.

EWMA – Question
 Example 1: On Tuesday, return on a stock was 4%. Volatility (Std. deviation) estimate for Tuesday
was 1%. Find volatility estimate for Wednesday using λ of 0.94

EWMA – Solution
 Variance estimate for Wednesday = (1-0.94)*(4%)^2 + (0.94)*(1%)^2 = 1.9%2
Std. Dev. = sqrt (1.9%) = 1.378%
 Tuesday volatility (Std. Dev.) estimate was 1%. Actual return on Tuesday was 4%. Therefore,
volatility estimate for Wednesday is estimated upwards than Tuesday i.e. 1.378% as compared to
1%.
 Notice how the volatility estimate has been revised due to high return

EWMA – Question
 Example 2: Continuing the previous example, volatility estimate for Wednesday was 1.378%.
Assume that actual return on Wednesday was 0%. What is the variance estimate for Thursday?

EWMA – Solution
 Variance estimate for Thursday = (1-0.94)*(0%)^2 + 0.94*(1.378%)^2 = 1.78%2

Std. Dev. = 1.34%
 In very short-term like daily returns, estimated volatility is the expected return
 Since latest return of 0% was lesser than estimated volatility (and estimated return) of 1.378%,
volatility for next day is revised downward from 1.378% to 1.34%
 Notice the downward revision in the estimate due to lower return

EWMA Weights Graph
λ = 0.7 (faster decay)

Weight of
variance terms
λ = 0.9 (slower decay)
Days into the past

How to Select λ (Decay Factor)
 Selection of Decay Factor

• We are free to select the decay factor we use in our calculation of volatility
• We can use common sense in our estimation
• If we expect volatility to be very unstable then we will apply a low decay factor (giving a lot of weight to
recent observations)
• If we expect volatility to be constant we would apply a high decay factor (giving a more equal weight to older
observations)
 Sum of Weights
• One special property of the weights used in the EWMA formula is that their sum will always equal to 1
m
S   (1
i 1
  ). i 1
  m
1
• Obviously we do not have an infinite set of historical data. We just have to make sure that our data set is
large enough so that this sum is close to 1
• Or alternatively we can rescale the weights so that the sum is 1
• What happens to the second term when m tends to infinity?

Question: FRM Exam
 Using a daily RiskMetrics EWMA model with a decay factor λ = 0.95 to develop a forecast of the
conditional variance, which weight will be applied to the return that is 4 days old?
A. 0.000
B. 0.043
C. 0.048
D. 0.950

Solution
 B.
• A. Incorrect. The wrong factor has been squared. The EWMA RiskMetrics model is defined as:
 ht = λ*ht-1 + (1- λ)* r2t-1. For t = 4, and processing r0 through the equation three times produces a factor of (1-
0.95)3*0.95 = 0.00012 for r0 when t = 4.
• B. Correct. The EWMA RiskMetrics model is defined as:
0.95)*0.95^3 = 0.043 for r0 when t = 4.
• C. Incorrect. The 0.95 has not been squared. The EWMA RiskMetrics model is defined as:
0.95)*0.95 = 0.048 for r0 when t = 4.
• D. Incorrect. The weight is not simply λ. The EWMA RiskMetrics model is defined as:
 ht = λ*ht-1 + (1- λ)* r2t-1. For t = 4, and processing r0 through the equation three times produces a factor of 0.95 =
0.950 for r0 when t = 4.

GARCH (1,1)
 GARCH stands for Generalized Autoregressive Conditional Heteroscedasticity
 Heteroscedasticity means variance is changing with time.
 Conditional means variance is changing conditional on latest volatility.
 Autoregressive refers to positive correlation between volatility today and volatility yesterday.
 (1,1) means that only the latest values of the variables.
 GARCH model recognizes that variance tends to show mean – reversion i.e. it gets pulled to a
long-term Volatility rate over time.
 2
t 1  VL     t
2
t
2
Long-term average Volatility

GARCH (1,1) (cont.)
 2
t 1       t
2
t
2
 Generally γ*VL is replaced by ω

 Since the sum of all the weights is equal to 1 we get the following equation as well:

VL 
1   

GARCH – Question
 Suppose a GARCH model is estimated using MLE from daily data as follows:
 t21  .000005 0.12 t2  0.85 t2

 Suppose that on a particular day ‘t’; actual return was -1% and the volatility (std. deviation)
estimate for that was 1.8%. Calculate the volatility estimate for next day (t+1) and long-term
average volatility (to which the model shows reversion over-time).

GARCH – Solution
 Solution: In the GARCH model, 12% is the weight given to latest squared return (reactive factor).
85% is the weight given to latest variance estimate (persistence factor). Therefore,
1-0.12-0.85 = 3% is weight given to long-term average Volatility.
• Therefore, 3%*VL = 0.000005 i.e. VL = 0.017%
• Also, variance estimate for t+1 = .000005 + 0.12*(-1%)^2 + 0.85*(1.88%)^2 = 0.0317%
• Volatility (Std. Dev.) estimate for t+1 = sqrt (0.0317%) = 1.782%
• For a stable GARCH model, alpha + Beta <=1. If alpha + Beta>1, then weight given to long-term volatility is
negative and the model becomes ‘mean-fleeing’

Question: FRM Exam
 Which of the following GARCH models will take the shortest time to revert to its mean?
A. ht = 0.05 + 0.03r2t-1 + 0.96ht-1
B. ht = 0.03 + 0.02r2t-1 + 0.95ht-1
C. ht = 0.02 + 0.01r2t-1 + 0.97ht-1
D. ht = 0.01 + 0.01r2t-1 + 0.98ht-1

Solution
 B.
• A. Incorrect. The model that will take the shortest time to revert to its mean is the model with the lowest
persistence defined by α1 + β. In this case the persistence factor is the second largest:
α1 + β = 0.03 + 0.96 = 0.99.
• B. Correct. The model that will take the shortest time to revert to its mean is the model with the lowest
persistence defined by α1 + β. In this case the persistence factor is the second lowest:
α1 + β = 0.02 + 0.95 = 0.97.
• C. Incorrect. The model that will take the shortest time to revert to its mean is the model with the lowest
persistence defined by α1 + β. In this case the persistence factor is the largest:
α1 + β = 0.01 + 0.97 = 0.98.
• D. Incorrect. The model that will take the shortest time to revert to its mean is the model with the lowest
persistence defined by α1 + β. In this case the persistence factor is the lowest:
α1 + β = 0.01 + 0.98 = 0.99.

Question
 Suppose the long-run variance rate is 0.0002 so that the long-run volatility per day is 1.4%
 n2  0.000002  013
. un21  0.86 n21
 Suppose that the current estimate of the volatility is 1.6% per day and the most recent percentage
change in the market variable is 1%. What is the new variance rate?

Solution
 The new volatility is 1.53% per day
0.000002  0.13  0.0001  0.86  0.000256  0.00023516

VAR Methods for Estimating Risk
 Historical based Approach
• Parametric Approach
 Exponential smoothing
 Risk Metrics (EWMA model with λ = 0.94)
 Typical example of parametric approach is the delta-normal VAR
• Non-Parametric Approach
 Historical Simulation
 Multivariate Density Estimation
• Hybrid Approach
 Implied Volatility based Approach

Hybrid approach Steps
 Estimating returns percentile directly as in HS
+
 Exponential smoothing as in Risk Metrics
 Step 1:
• Denote by r (t – 1 ,t ) the realized returns from (t-1) to t
• For the most recent k returns
• Choose lamda based on weight requirements (L)
• Assign a weight (1 – L) / (1 – L ^K) for t = 1 , (1 – L) / (1 – L ^K) * L for t = 2 …
 Step 2 :
• Order returns in ascending order
 Step 3 :
• Keep accumulating the weights till you reach x% VaR
• Linear interpolation is used to reach exactly x % of the distribution

Question 14 (Hybrid approach)
Hybrid Approach Problem Worksheet
Lamda 0.98 K 100
Return Periods Ago Hybrid Wt Cum Hybrid Wt HS Wt Cum HS Wt
-3.3 3 0.022144839 0.022144839 0.01 0.01
-2.9 2 0.022596774 0.044741613 0.01 0.02
-2.7 65 0.006328331 0.051069944 0.01 0.03
-2.5 45 0.009479113 0.060549057 0.01 0.04
-2.4 5 0.021267903 0.08181696 0.01 0.05
-2.3 30 0.012834429 0.09465139 0.01 0.06
Hybrid Interpolation 0.16618552
Ans -2733814418
HS Ans -2.35

Portfolio Volatility
 Variance – Covariance Approach

• Extend the parametric approach to entire portfolio
• Disadvantage: Correlations tend to increase in periods of stress. This approach might underestimate the
portfolio Volatility
 Extend Historical Simulation Approach

• Apply current portfolio weight to old period returns
 Third Approach
• Aggregate the simulated returns and then apply the parametric approach to aggregated portfolio

Implied Volatilities
 Forward looking, predictive
 Use specific Derivative pricing model
 Black Scholes pricing model for ATM options
 The implied volatility calculated from a European call option should be the same as that calculated
from a European put option when both have the same strike price and maturity
 Implied volatility: The value for σ such that Black & Scholes price is equal to the observed
market price
Truly Predictive in nature and is not a historical simulation but it assumes that Black Scholes
(or any other model) is correct model for calculating market Value of Option

Discussion (VaR Methodology)
 Three value at risk (VaR) methods are reviewed: Delta-normal, historical simulation and Monte
Carlo. Which are true of the following? Among the VaR methods, which:
i. Efficient in terms of data use?
ii. Requires a (parametric) distributional assumption?
iii. Is LEAST appropriate for a portfolio that contains many embedded derivatives (e.g., options)?
iv. Is computationally fast?
v. Handles fat tails?
vi. Suffers from sampling variation?

Question 15 (Historical Simulations)
 Suppose we have an asset with ordered simulated price returns as below for sample of 500 days
and is trading at 70. What is the VaR at 99% confidence if the returns for the last 500 days are:
• -7%, -6.7%,-6.6%, -6.5%,-.6.1%,-5.9% … 4%, 4.75%,5.1%, 5.2%,5.3%

Solution
 99% of 500 = 495th i.e. -5.9% = 0.059 x 70 = 4.13

Question 16 (HS)
 We have an asset with ordered simulated price returns as below for sample of 400 days and is
trading at 100. What is the VaR at 99% confidence if the returns for the last 400 days are:
• -7%, -6.7%,-6.6%, -6.5%,-.6.1%,-5.9% … 4%, 4.75%,5.1%, 5.2%,5.3%
 What is the expected shortfall?

Solution
 VaR = 396th = -6.1% of 100 = -6.1
 Expected shortfall = - (7 + 6.7 + 6.6 + 6.5)/4 = -6.7%
Expect atleast one direct question on the calculation of Expected Shortfall

Putting VaR to Work

Full Valuation Method
 Full Valuation method is the process of measurement of risk of a portfolio by fully re-pricing it
under a set of scenarios over a time period. It can be used to cover a large range of values of the
portfolio returns in order to provide more accurate results. It generally provides more accurate
results compared to delta normal approach but is a complicated process.
 Advantages over delta normal:

• It accounts for non-linearities of derivatives whereas delta normal assumes a linear approximation.
• It accounts for extreme fluctuations.
 Two popular methods under full revaluation approach have been explained in the subsequent
slides.

Historical Simulation
 No assumptions is required about the distribution of returns.
 VaR is estimated directly from the past data which includes all the correlations among
historical data.
 Historical data has fat tails and skewness already accommodated in itself.
 The most important parameter in historical simulation is the Look back window.
• For example, in 250 observations window the 5th percentile is between 12th and 13th observation.
 Historical Simulation method is not exposed to any model risk.
 The biggest drawback with the Historical Simulation method is that the changes in volatility and
correlation from structural changes are not recognized.
 We have an asset with ordered simulated price returns as below for sample of 100 days and is
trading at 100. What is the VaR at 99% confidence if the returns for the last 100 days are:
• -7.5%, -6.7%,-6.6%, -6.51%,-.6.12%,-5.92% … 4.34%, 4.5%,5.1%, 5.2%,5.3%

Monte Carlo Simulation
 Straddles: Long position in call and put with same exercise price.
 Straddle is a non Linear derivative whose payoff increases with the increase in the volatility.
 Also delta normal VAR increases with the increase in the volatility.
 Going by the delta normal VAR for straddles, as the volatility increases, VAR should increase but in
reality the payoff is becoming positive.
Payoff
Gains !
Price
Risk !

 The Monte Carlo approach assumes that there is a known probability distribution for the
risk factors.
 The usual implementation of Monte Carlo assumes a stable, Joint-Normal distribution for the risk
factors.
 This is the same assumption used for Parametric VaR.
 The analysis calculates the covariance matrix for the risk factors in the same way as
Parametric VaR.
 Unlike Parametric VaR Monte Carlo Simulation:
• Decomposes the covariance matrix and ensures that the risk factors are correlated in each scenario.
• The scenarios start from today's market condition and go one day forward to give possible values at the end
of the day.
• Full, nonlinear pricing models are then used to value the portfolio under each of the end-of-day scenarios.
• For bonds, nonlinear pricing means using the bond-pricing formula rather than duration.
• For options, it means using a pricing formula such as Black-Scholes rather than Greeks.

 Advantages:
• Unlike Parametric VaR, it uses full pricing models and can therefore capture the effects of nonlinearities.
• Unlike Historical VaR, it can generate an infinite number of scenarios and therefore test many possible future
outcomes.
 Disadvantages:
• The calculation of Monte Carlo VaR can take 1,000 times longer than Parametric VaR because the potential
price of the portfolio has to be calculated thousands of times.
• Unlike Historical VaR, it typically requires the assumption that the risk factors have a Normal distribution.

Comparison between Methods
 The best methods depend on speed required and whether the portfolio has non-linear elements.
 For large portfolios where non linearity is not a major factor, the delta normal method is fast
and efficient.
 For portfolios with substantial non-linear elements full valuation may be necessary.

Visualizing WCS
 The worst case scenario (WCS) answers:
• What is the worst loss that can happen over a period of time?
• The probability of a “worst loss” is certain (100%); the timing is uncertain.
 Focuses on distribution of loss during the worse trading period:
 A worst case scenario can be simulated only using Monet Carlo Simulations

Visualizing WCS
 The area under the normal curve for confidence value is:
Confidence (x%) ZX%

90% 1.28
95% 1.65
97.5% 1.96
0.016 99% 2.32

Distribution of WCS
0.011
0.005 Mean = 0
+- stdev
Approx. Normal curve representing Value at Risk

Question (WCS)
 H = 100
 Z < -2.33 ( 99% VaR )  1 out of 100
 Distribution of WCS mean = -2.51 and it’s 1st and 5th percentile are -3.1 and -3.9 resp.
 What does this mean to a financial risk manager?

Solution
 Mean of distribution of WCS = -2.51

• It means that if the loss exceeds the VAR number(2.33), the average value of the loss would be 2.51 which is
nothing but the expected shortfall (ES)
 The 5th and 1st percentile of WCS is -3.1 and -3.9
• It means that if the loss exceed the VAR, then probability of loss exceeding 3.9 is 1%
• Similarly the probability of loss exceeding 3.1 would be 5%

Stress Testing

Stress Testing
 Limitations of VaR
• Tells that n number of times in 100 days, the loss is not going to exceed N$
• But it cannot predict the loss when it exceeds!
• Does not focus on large losses (Tails of distribution)
 Stress Testing:
• Supplement to VaR
• "VaR should always be supplemented with stress testing" has been one of the recommendations
of the supervisor
• Testing how well a portfolio performs under some of the most extreme market moves seen in the last 10 to
20 years

Stress Testing
 Advantages:
• Can take a large number of risk factors into consideration
• Can specifically focus on the tails (extreme losses)
 Disadvantages:
• Highly subjective and can become overcautious
• Requires complete top management support
 Two independent sections to the risk report:

• VAR-based
 Top-down identification of the relevant risk generators for the trading portfolio
• Stress testing-based risk report proceeds in one of two ways
 It examines a series of historical stress events and
 It analyzes a list of predetermined stress scenarios

Stress Testing
 Capital Allocation
 Exposure
 Ride out Turmoil
 The one significant shortcoming of VaR that stress testing does address is sudden changes in
historical correlations
 If two currencies have been pegged to one another, they will exhibit a high historical correlation. A
VaR analysis based on that historical correlation will not address the risk that one of the currencies
may be devalued relative to the other. If this is a scenario that concerns management, a simple
stress test will offer more insights than would, say, a VaR analysis performed with a modified
correlation assumption

Scenario Analysis
 Oct19, 1987 : The Black Monday, a -20 sigma event

 Sep 1992, breakup of fixed exchange system
• The 99% VAR would have totally missed the magnitude of actual loss
 Goals
• Identify scenarios that would not occur under standard VAR models
• Simulating shocks that have never occurred
• Simulating shocks that reflect permanent structural breaks or temporally changed statistical patterns
 Principles of Scenario Analysis

• Identify risk factors
• Forecast and Change Risk factors
• Revalue Portfolio using VaR system in place
• Portfolio Vs Event Driven

Scenario Analysis
 Parallel shift in yield curve

 Changes in steepness of yield curve
 Parallel shift in yield curve along with the changes in the steepness of yield curve
 Changes in yield volatilities
 Changes in the values of equity indices
 Changes in equity index volatilities
 Changes in the values of key currencies with respect to US Dollar
 Changes in foreign exchange rate volatilities
 Changes in Swap spreads in G7 countries plus Switzerland

Standard Portfolio Analysis of Risk (SPAN)
 Example of scenario based method for measuring portfolio risk

 SPAN uses full valuation methods that makes it a good tool for analyzing portfolio
including options
 Margin is set to the worst portfolio loss after considering all scenarios
 Useful when considering only two risk factors. Becomes very complex when risk factors increase.
 Drawback:
• Ignores correlation between risk factors
• With more number of risk factors, alternative scenarios could become unmanageable

Unidimensional and Multidimensional Scenarios
 Uni-dimensional scenario analysis:

• Identifies important risk factors
• Shocks the factor by a large amount
• Measures the impact on portfolio value
 Multi-dimensional Scenario Analysis

• Incorporates correlation between risk factors.
• This increases the complexity of the analysis
• This analysis can be backward or forward looking

Various Approaches to Scenario Analysis
 Prospective Scenarios – Factor Push Method and Conditional Scenario Method.
 Factor push method shifts each risk factor in the direction that would have an adverse impact on
the portfolio.
 Conditional Scenario Method incorporates the correlation between various key risk factors .

Improving Stress Tests
Stress tests can be improved by adopting one or more of the following methods:
 Buy protection through insurance or other derivative products
 Change portfolio composition to decrease the exposure or diversify
 Change business strategy to suit the changing business environment
 Develop back up plans for unforeseen events
 Secure alternative funding in stressed environment

Five Minute Recap
Key Terms to Remember: Volatility calculation: EWMA: Confidence (x%) ZX%

 Delta Normal VAR 1 n
 Implied Volatility 2 
n
 (R
i 1
i  E ( R )) 2  2n   2n 1  (1   ) u n21 90% 1.28
 Regime Switching 95% 1.65
 The worst case scenario
GARCH (1,1): 97.5% 1.96
 Stress Testing
VAR  Z X % * * AssetValue
 t21  VL   t2   t2 99% 2.32
VaR (n days) (in %)  VaR (daily VaR) (in %) * n VARLinear Derivative   *VARUnderlying Risk Factor
VaR portfolio (in %)  wa2 (%VAR a ) 2  w 2b (%VAR b ) 2  2wa w b * (%VAR a ) * (%VAR b ) * ab
Value at Risk: Value-at-Risk Measurement Methods: Monte Carlo Simulation:

0.5 Probability
Payoff
0.4 VaR Gains !
0.3
0.2 95% daily-VAR
Z values Linear Valuation Full Revaluation
0.1 Method Method
0 Price
-4 -2 0 2 4
Mean = 0 Delta Normal Historical Monte Carlo
Method Simulation Simulation Risk !
VAR X % (in %)  Z X % *

Thank You !
support@edupristine.com
www.edupristine.com
© EduPristine – www.edupristine.com
© EduPristine For VaR-I (2016)

VaR I PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

VaR I PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Valuations and Risk Models – I

© EduPristine For VaR-I (2016) 1

I own risk-less assets as My assets are very risky

© EduPristine For VaR-I (2016) 2

There may be a day out of 100 when your loss is $5000,

© EduPristine For VaR-I (2016) 3

Confidence (x%) ZX%

© EduPristine For VaR-I (2016) 4

© EduPristine For VaR-I (2016) 5

VaR (n days) (in %)  ZX% * * Asset Value * n

 portfolio  wa2  a2  w 2b  b2  2wa w b *  a *  b * ab

© EduPristine For VaR-I (2016) 6

© EduPristine For VaR-I (2016) 7

© EduPristine For VaR-I (2016) 8

© EduPristine For VaR-I (2016) 9

 10 day VaR = 0.3728 x (10)^0.5 = 1.1789

© EduPristine For VaR-I (2016) 10

 What is the daily portfolio VaR at 97.5% confidence level?

© EduPristine For VaR-I (2016) 11

 Portfolio VaR = [(40 x 0.1078)2 + (60 x 0.0833)2 + 2x0.1078x0.833x40x60x0.20]0.5 = 7.22 mn

© EduPristine For VaR-I (2016) 12

© EduPristine For VaR-I (2016) 13

 Market Value of asset $10 mn

© EduPristine For VaR-I (2016) 14

© EduPristine For VaR-I (2016) 15

 For an uncorrelated portfolio what is the VaR if:

© EduPristine For VaR-I (2016) 16

 The VaR comes out to be 22.36 mn

© EduPristine For VaR-I (2016) 17

Linear Valuation Method Full Revaluation Method

Delta Normal Historical Monte Carlo

© EduPristine For VaR-I (2016) 18

© EduPristine For VaR-I (2016) 19

 Linear Derivatives: Payoff diagrams that are linear or almost linear:

VARLinear Derivative   *VARUnderlying Risk Factor

© EduPristine For VaR-I (2016) 20

• For Many Underlying variables: P  S

© EduPristine For VaR-I (2016) 21

© EduPristine For VaR-I (2016) 22

© EduPristine For VaR-I (2016) 23

© EduPristine For VaR-I (2016) 24

© EduPristine For VaR-I (2016) 25

© EduPristine For VaR-I (2016) 26

Risk Management is all about understanding tails of distribution

© EduPristine For VaR-I (2016) 27

High Vol Estimate

© EduPristine For VaR-I (2016) 28

 In this case we observe distinct periods of high and low volatility.

© EduPristine For VaR-I (2016) 29

© EduPristine For VaR-I (2016) 30

© EduPristine For VaR-I (2016) 31

This topic has already been covered in Quants Lectures

© EduPristine For VaR-I (2016) 32

 n2    n2 2  (1   )u n2 2   (1   )u n21

© EduPristine For VaR-I (2016) 33

 t21  (1  0.94) t2  0.94(1  0.94)t21  0.94 t21 

 What are the weights for old returns and variance?

© EduPristine For VaR-I (2016) 34

© EduPristine For VaR-I (2016) 35

© EduPristine For VaR-I (2016) 36

© EduPristine For VaR-I (2016) 37

 Variance estimate for Thursday = (1-0.94)*(0%)^2 + 0.94*(1.378%)^2 = 1.78%2

© EduPristine For VaR-I (2016) 38

VaR (n days) (in %)  ZX%  Asset Value * n

 Variance estimate for Thursday = (1-0.94)(0%)^2 + 0.94(1.378%)^2 = 1.78%2