16-12-2020

Value-at-Risk (VaR)

Sankarshan Basu
Professor of Finance
Indian Institute of Management
Bangalore

The Question Being Asked in VaR

“What loss level is such that we are X%

confident it will not be exceeded in N business
days?”

VaR and Regulatory Capital

• Regulators base the capital they require

banks to keep on VaR
• The market-risk capital has traditionally been
calculated from a 10-day VaR estimated
where the confidence level is 99%
• Credit risk and operational risk capital are
based on a one-year 99.9% VaR

1
 16-12-2020

Advantages of VaR
• It captures an important aspect of risk
in a single number
• It is easy to understand
• It asks the simple question: “How bad can
things get?”

VaR vs. Expected Shortfall

• VaR is the loss level that will not be exceeded with a specified
probability
• Expected shortfall (ES) is the expected loss given that the loss
is greater than the VaR level (also called C-VaR and Tail Loss)
• Regulators have indicated that they plan to move from using
VaR to using ES for determining market risk capital
• Two portfolios with the same VaR can have very different
expected shortfalls

Distributions with the Same VaR but

Different Expected Shortfalls

VaR

VaR

2
 16-12-2020

Normal Distribution Assumption

• When losses (gains) are normally distributed with

mean m and standard deviation s
VaR  m  s N 1 ( X )
e Y 2
2

ES  m  s
2 (1  X )
1
where Y is N (X).

Changing the Time Horizon

• If losses in successive days are independent,
normally distributed, and have a mean of zero

T - day VaR  1 - day VaR  T

T - day ES  1 - day ES  T

VaR - Calculation

• Historical simulation
• Model building approach
• Monte Carlo simulation

3
 16-12-2020

Historical Simulation
• Historical simulation involves using past data in a
very direct way as a guide to what might happen in
the future.
• The first step is to identify the market variables
affecting the portfolio.
– These will typically be exchange rates, equity prices,
interest rates, and so on.
• We then collect data on the movements in these
market variables over the most recent 500 days.
– This provides us with 500 alternative scenarios for what
can happen between today and tomorrow.

10

Historical Simulation – contd.

• For each scenario we calculate the dollar change in the value
of the portfolio between today and tomorrow.
– This defines a probability distribution for daily changes in the value of
our portfolio.
• The fifth-worst daily change is the first percentile of the
distribution.
• The estimate of VaR is the loss when we are at this first
percentile point.
• Assuming that the last 500 days are a good guide to what
could happen during the next day, we are 99% certain that we
will not take a loss greater than our VaR estimate.

11

Historical Simulation – contd.

• Suppose we use m days of historical data.
• Let vi be the value of a variable on day i
• There are m-1 simulation trials
• The ith trial assumes that the value of the market
variable tomorrow (i.e., on day m+1) is

vi
vm
vi 1

12

4
 16-12-2020

Historical Simulation – contd.

0.1343*(0.1162/0.1159)

13

The Model-Building Approach

• The main alternative to historical simulation is to
make assumptions about the probability
distributions of the returns on the market variables
• This is known as the model building approach (or
sometimes the variance-covariance approach)

14

Microsoft Example
• We have a position worth $10 million in
Microsoft shares
• The volatility of Microsoft is 2% per day (about
32% per year)
• We use N=10 and X=99

15

5
 16-12-2020

Microsoft Example – contd.

• The standard deviation of the change in the
portfolio in 1 day is $200,000
• The standard deviation of the change in 10
days is

200,000 10  $632,500

16

Microsoft Example – contd.

• We assume that the expected change in the
value of the portfolio is zero (This is OK for
short time periods)
• We assume that the change in the value of the
portfolio is normally distributed
• Since N(–2.326)=0.01, the VaR is

2.326  632,500  $1,471,300

17

AT&T Example
• Consider a position of $5 million in AT&T
• The daily volatility of AT&T is 1% (approx 16%
per year)
• The SD per 10 days is
50,000 10  $158,144
• The VaR is
158,100  2.326  $367,800

18

6
 16-12-2020

Portfolio
• Now consider a portfolio consisting of both
Microsoft and AT&T
• Suppose that the correlation between the
returns is 0.3

19

S.D. of Portfolio
• A standard result in statistics states that

s X Y  s 2X  sY2  2rs X s Y
• In this case sX = 200,000 and sY = 50,000 and
r = 0.3. The standard deviation of the change
in the portfolio value in one day is therefore
220,200

20

VaR for Portfolio

• The 10-day 99% VaR for the portfolio is

220,200  10  2.326  $1,620,100
• The benefits of diversification are
(1,471,300+367,800)–1,620,100=$219,000
• What is the incremental effect of the AT&T
holding on VaR?

21

7
 16-12-2020

ES Results (10-day 99%)

• Microsoft shares: $1,687,000
• AT&T shares: $421,700
• Portfolio: $1,857,600

22

The Linear Model

We assume
• The daily change in the value of a portfolio is
linearly related to the daily returns from
market variables
• The returns from the market variables are
normally distributed

23

Markowitz Result for Variance of

Return on Portfolio
n n
Variance of Portfolio Return   rij wi w j s i s j
i 1 j 1

wi is weight of ith asset in portfolio

s i2 is variance of return on ith asset
in portfolio
rij is correlation between returns of ith
and jth assets

24

8
 16-12-2020

VaR for Bond Portfolios

Task: To calculate VaR of a T-Bond with a principal of $1
mn. and coupon of 10% per annum payable semi-
annually maturing in 0.8 years
Method: Cash flow mapping and Variance-covariance
method for calculating VaR
The bond cash flows have to be converted to equivalent
3-,6- and 12-month cash flows using cash flow
mapping since zero rates, bond price volatilities and
return correlations are available for these maturities
only.

25

Example: VaR for Bond Portfolios

DATA:
3-m 6-m 1-year
Zero rate 5.50 6.00 7.00
(% with annual Compounding)
Bond price volatility 0.06 0.10 0.20
(% per day)

26

Example: VaR for Bond Portfolios

DATA:
Correlations 3-m 6-m 1-year
between daily returnsbond bond bond
3-month bond 1.0 0.9 0.6
6-month bond 0.9 1.0 0.7
1-year bond 0.6 0.7 1.0

27

9
 16-12-2020

Example: VaR for Bond Portfolios

The T-bond cash flows are:

Time Amount

0.3-year $ 50,000
0.8-year $1,050,000

28

Example: VaR for Bond Portfolios

0.8-year T-Bond
= 0.3-year ZCB ($50,000)
+ 0.8-year ZCB ($1,050,000)

0.3-year ZCB is mapped into equivalent

positions in 3-m and 6-m ZCBs and 0.8-year
ZCB is mapped into equivalent positions in
6-m and 1-year ZCBs

29

Example: VaR for Bond Portfolios

The cash flow mapping procedure
The present value and variance of
equivalent positions should remain the same
As the original cash flow.

30

10
 16-12-2020

Example: VaR for Bond Portfolios

First, the 0.3-yr and 0.8-yr zero rates and
volatilities are worked out by interpolation.

0.3-yr 0.8-yr
Zero rate (%) 5.60 6.60
Bond price volatility 0.068 0.16
(% per day)

31

Example: VaR for Bond Portfolios

The present values of T-Bond cash flows are as
follows:

PV(0.3-yr ZCB)
= 50,000/1.0560.3 = 49,189.32
PV(0.8-yr ZCB)
= 1,050,000/1.0660.8 = 997,662

32

Example: VaR for Bond Portfolios

Cash flow mapping of 0.3-yr ZCB
Suppose, we allocate  of the PV to the 3-m bond and
(1- ) to the 6-m bond.
Then, matching variances, we obtain,
0.000682 = 0.00062 2+0.0012(1- )2
+2*0.9*0.0006*0.001* *(1 - )
 = 0.760259
Position in 3-m ZCB = 0.760259*49,189.32 = $37,397
Position in 6-m ZCB = 0.239741*49,189.32 = $11,793

33

11
 16-12-2020

Example: VaR for Bond Portfolios

Cash flow mapping of 0.8-yr ZCB
Suppose, we allocate  of the PV to the 6-m bond and
(1- ) to the 1-yr bond.
Then, matching variances, we obtain,
0.00162 = 0.0012 2+0.0022(1- )2
+2*0.7*0.001*0.002* *(- )
 = 0.320337
Position in 6-m ZCB = 0.320337*997,662 = $319,589
Position in 1-yr ZCB = 0.679663*997,662 = $678,074

34

Example: VaR for Bond Portfolios

Cash flow mapping of 0.8-yr T-Bond

$50,000 $1,050,000 Total

(0.3-yr) (0.8-yr) ($)

3-m bond 37,397 37,397

6-m bond 11,793 319,589 331,382

1-yr bond 678,074 678,074

35

Example: VaR for Bond Portfolios

Calculation of VaR
The variance of the portfolio of 3-m, 6-m and 1-
yr ZCBs can be worked out using the
volatilities, correlations and value weights.

Variance = $2,628,518 and

Std. Dev. = $1621.3
So, 10-day 99% VaR = 1621.3*√10*2.33 =
$11,946

36

12
 16-12-2020

When Linear Model Can be Used

• Portfolio of stocks
• Portfolio of bonds
• Forward contract on foreign currency
• Interest-rate swap

37

Approximate VaR for Options using The Linear

Model

• Consider a portfolio of options on a single

stock price, S.
• Let δ be the delta of the position and Δx, the
percentage change in stock price in 1 day.
Then,
P S
  and x 
S S

38

Approximate VaR for Options using The

Linear Model
• As an approximation
P   S  S x
• Similarly when there are many underlying market
variables
n n
P   S  x    x where   S 
i i i i i i i i
i 1 i 1

and i is the delta of the portfolio with respect to the ith

asset.
• VaR can then be calculated using linear model.

39

13
 16-12-2020

Approximate VaR for Options using The Linear

Model– Example
• Consider an investment in options on Microsoft and
AT&T. Suppose the stock prices are 120 and 30
respectively and the deltas of the portfolio with
respect to the two stock prices are 1,000 and 20,000
respectively.
• As an approximation
P  120 1,000x1  30  20,000x2
• where x1 and x2 are the percentage changes in the
two stock prices and ∆P is the resultant change in the
value of the portfolio.

40

Approximate VaR for Options using The

Linear Model– Example
• Assuming that the daily volatility of Microsoft is 2%
and the daily volatility of AT&T is 1%, and the
correlation between the daily changes is 0.3, the
standard deviation of ∆P is
 (120,000 * 0.02) 2  (600,000 * 0.01) 2  2 * (120,000 * 0.02) * (600,000 * 0.01) * 0.3
 7,099

• Because N(-1.65) = 0.05,

5-day 95% VaR =1.65*√5*7,099 = $26,193

41

Shortcoming of the Linear Model for

Options
• The linear model fails to capture skewness in
the probability distribution of the portfolio
value.
• It does not take into account gamma of the
portfolio.
– Gamma measures the curvature of the
relationship between the portfolio value and an
underlying market variable.
• For a more accurate estimate of VaR for
options quadratic model should be used.

42

14
 16-12-2020

Quadratic Model
For a portfolio including options dependent
on a single stock price it is approximately true
that
1
P  S   (S ) 2
2

This becomes (using Δx),

1 2
P  S x  S  ( x ) 2
2

43

Monte Carlo Simulation

To calculate VaR using MC simulation we
• Value portfolio today
• Sample once from the multivariate
distributions of the xi
• Use the xi to determine market variables at
end of one day
• Revalue the portfolio at the end of day

44

Monte Carlo Simulation continued

• Calculate P
• Repeat many times to build up a probability
distribution for P
• VaR is the appropriate fractile of the
distribution times square root of N
• For example, with 1,000 trial the 1 percentile
is the 10th worst case.

45

15
 16-12-2020

Model Building vs Historical Simulation

Model building approach can be used for
investment portfolios where there are no
derivatives, but it does not usually work when
for portfolios where
• there are derivatives
• positions are close to delta neutral

46

Improving VaR/ES estimates

• VaR/ES calculation is backward looking. But some
modifications are possible.
a) Volatility updating can lead to more extreme
outcomes considered when market is highly
volatile.
b) Extreme value theory provides a way of
extending the tails of the loss distribution
obtained from historical data.
c) Calculating stressed VaR / ES considering the
impact of a particularly difficult period of 250
days that has occurred in the past.

47

Volatility Updating

• Use a volatility updating scheme to calculate values

of each variable under historical simulation as
follows.

• Value of market variable under ith scenario

v  (vi  vi 1)s n 1 / s i
 vn i 1
vi 1

48

16
 16-12-2020

Extreme Value Theory (EVT)

• EVT is the term used to describe the science of

estimating the tails of a distribution.
• EVT can be used to improve VaR estimates and to
help in situations where analysts want to estimate
VaR with a very high confidence level.

49

17