Calculating VaR

Parametric and Nonparametric VaR:

Nonparametric VaR: When the forward distribution of changes in the portfolio
value is derived using historical data or simulated data
No need to make adjustments for deviation from normality
Parametric VaR: If the VaR is derived assuming a particular theoretical
distribution of the returns
The parameters are estimated using historical data
Historical Simulation
Historical Simulation
The most popular approach for calculating value at risk (VaR) and expected
shortfall (ES) for market risk.
Involves using the day-to-day changes in the values of market variables that
have been observed in the past in a direct way to estimate the probability
distribution of the change in the value of the current portfolio between today
and tomorrow.
The Methodology
Historical simulation involves using past data as a guide to what will happen in
the future.
Suppose that we want to calculate VaR for a portfolio using a one-day time
horizon, a 99% confidence level, and 501 days of data.
The time horizon and confidence level are those typically used for a market risk
VaR calculation.
501 days of data leads to 500 scenarios being created.
The Methodology
First, identify the market variables affecting the portfolio.
◦ These market variables are sometimes referred to as risk factors.
◦ They typically include exchange rates, interest rates, stock indices, volatilities, etc.
Then, data are collected on movements in these market variables over the most
recent 501 days.
This provides 500 alternative scenarios for what can happen between today and
tomorrow.
The Methodology
Denote the first day for which we have data as Day 0, the second day as Day 1,
and so on.
Scenario 1 is where the percentage changes in the values of all variables are the
same as they were between Day 0 and Day 1, Scenario 2 is where they are the
same as between Day 1 and Day 2, and so on.
For some variables such as interest rates, credit spreads, and volatilities, actual
rather than percentage changes in market variables are considered.
For each scenario, the dollar change in the value of the portfolio between today
and tomorrow is calculated.
The Methodology
The 99 percentile of this distribution can be estimated as the fifth worst
outcome.
The estimate of VaR is the loss when we are at this 99 percentile point.
We are 99% certain that we will not take a loss greater than the VaR estimate if
the percentage changes in market variables in the past 500 days are
representative of what will happen between today and tomorrow.
ES is the average loss conditional that we are in the 1% tail of the loss
distribution.
VaR is estimated as the fifth worst loss.
ES can be estimated by averaging the losses that are worse than VaR—that is,
the four worst losses.
Weighting of Observations
The natural weighting scheme to use is one where weights decline exponentially
The weight assigned to Scenario 1 (which is the one calculated from the most
distant data) is λ times that assigned to Scenario 2.
This in turn is λ times that given to Scenario 3, and so on.
So that the weights add up to 1, the weight given to Scenario i is

where n is the number of scenarios

Weighting of Observations
VaR is calculated by ranking the observations from the worst outcome to the
best.
Starting at the worst outcome, weights are summed until the required percentile
of the distribution is reached.
For example, if we are calculating VaR with a 99% confidence level, we continue
summing weights until the sum just exceeds 0.01.
The parameter λ can be chosen by trying different values and seeing which one
back-tests best.
Model-Building
Approach
Model-Building Approach
An alternative to the historical simulation approach
Sometimes also referred to as the variance–covariance approach
Involves assuming a model for the joint distribution of changes in market
variables and using historical data to estimate the model parameters
Model-Building Approach
When the loss in the portfolio value has a mean of μ and a standard deviation of
σ (under assumption of normal distribution),

X is the confidence level

N−1(.) is the inverse cumulative normal distribution [similar to z-score]
(which can be calculated using NORM.S.INV in Excel)
Model-Building Approach
When the loss is assumed to be normally distributed with mean μ and standard
deviation σ, ES with a confidence level of X is given by

Y is the Xth percentile point of the standard normal distribution

[similar to z-score]
Model-Building Approach
Suppose that the loss from a portfolio over a 10-day time horizon is normal with
a mean of zero and a standard deviation of $20 million.
The 10-day 99% VaR is $46.5 million.

Because 2.326 is the point on a standard normal distribution that has a 1%

chance of being exceeded, the 10-day 99% ES is $53.3 million.
Back-Testing
Back-testing is an important reality check for a risk measure.
It is a test of how well the current procedure for calculating the measure would
have worked in the past.
Suppose that we have developed a procedure for calculating a one-day 99% VaR.
Back-testing involves looking at how often the loss in a day would have
exceeded the one-day 99% VaR when the latter is calculated using the current
procedure.
Days when the actual loss exceeds VaR are referred to as exceptions.
Back-Testing
If exceptions happen on about 1% of the days, we can feel reasonably
comfortable with the current methodology for calculating VaR.
If they happen on, say, 7% of days, the methodology is suspect and it is likely
that VaR is underestimated.
From a regulatory perspective, the capital calculated using the current VaR
estimation procedure is then too low.
On the other hand, if exceptions happen on, say, 0.3% of days, it is likely that the
current procedure is overestimating VaR and the capital calculated is too high.
Stress Testing
Stress testing involves evaluating the impact of extreme, but plausible, scenarios
that are not considered by value at risk (VaR) or expected shortfall (ES) models.
VaR/ES models are useful, but they are inevitably backward looking.
Risk management is concerned with what might happen in the future.
Events that could happen but are quite different from those that occurred
during the period covered by the data, are not taken into account.
Stress testing is an attempt to overcome this weakness of the VaR/ES measure.
Stress Testing
Stress testing involves estimating how the portfolio of a financial institution
would perform under scenarios involving extreme (but plausible) market moves.
Alternative procedures:
◦ Stressing Individual Variables: A large move in one variable and other variables are
unchanged.
◦ Scenarios Involving Several Variables: Several variables change at the same time.
◦ Scenarios Generated by Management: Use their understanding of markets, world
politics, the economic environment, and current global uncertainties to develop
plausible scenarios that would lead to large losses.
◦ Reverse Stress Testing: Use of computational procedures to search for scenarios that
lead to a failure of the financial institution
Problems
Problem: VaR
A portfolio has a daily VaR of ₹15 million at the 95% confidence level.
Meaning?
On 95% of the days, the portfolio loss will not exceed ₹15 million.
A 95% VaR can also be interpreted as the level of capital that is sufficient to
absorb losses 95% of the time.
Problem: VaR
Over the next 30 days, the change in the value of a loan portfolio can vary
between a loss of ₹25 million and a profit of ₹25 million.
The occurrence of each change is equally likely.
The VaR at 90% confidence level is?
Answer:
₹20 million
Problem: VaR
Suppose that the gain from a portfolio during 30 days is normally distributed
with a mean of 0 and a standard deviation of ₹15 million.
Compute the VaR for the portfolio with a time horizon of 30 days and confidence
level of 95%.
Answer:
From the properties of the normal distribution, the one-percentile point of this
distribution is 0 − 1.645 × 15 or –₹24.68 million.
The VaR for the portfolio with a time horizon of 30 days and confidence level of
95% is therefore ₹24.68 million.
Problem: VaR
An endowment fund has a ₹10,000,000 portfolio.
The manager of the fund assumes that there are 250 trading days in a year and
uses a 1% level of significance to estimate the VaR.
The annual return is expected to be 10% with a standard deviation of 15%.
Assuming that the returns are distributed normally, the daily VaR is?
Answer:
0.15
Daily volatility = = 0.009487
250
Daily return = 10,000,000*0.10/250=4,000
Daily VaR = 4,000-(0.009487*10,000,000*2.33)=₹217,047
Problem: ES
A market risk manager calculates the daily VaR at the 99th percentile of ₹14
million.
Loss observations beyond the 99th percentile VaR level in ₹ million are 15, 18, 21,
24, and 32.
The ES is?
Answer: 110mn/5 = ₹22 million
Problem: Aggregation of VaR and ES
The VaR of one business unit is 180 and the VaR of another business unit is 300.
If the correlation between the losses of the two business units is 0.1, the
combined VaR is?
Answer:
365
Problem: Weighted HS
A fund estimates the VaR of its portfolio using weighted HS.
The weights assigned to the daily changes in the portfolio value decline
exponentially as the observations move back in time.
The fund identifies the following high-value losses with their associated weights.
Loss (₹’000) Weight
423.3 0.00228
380.1 0.00526
326.8 0.00169
518.1 0.00255
716.7 0.00517
415.7 0.00246
307.9 0.00093
Problem: Weighted HS
To identify the 99% VaR, the losses are first arranged in descending order and
weights are accumulated
Loss (₹’000) Weight Cumulative Weight
716.7 0.00517 0.00517
518.1 0.00255 0.00772
423.3 0.00228 0.01000
415.7 0.00246 0.01246
380.1 0.00526 0.01772
326.8 0.00169 0.01941
307.9 0.00093 0.02034
The 99% VaR is ₹423,300.
References
Textbook
Hull (RM&FI)
Hull (O,F &OD)