Risk Management

Value-At-Risk

Lecture 7
Prof Youwei Li

1
Introduction to VAR
l Measures such a delta, gamma and vega represent
different aspects of the risk in a portfolio consisting
of options and other financial assets.
l Financial institutions usually calculate each of
these measures each day for every market variable
to which it is exposed.
l These risk measures provide valuable information
for the financial institution’s traders, but they are of
limited use to senior management.

2
Introduction
l Value at risk (VaR) is an attempt to provide a single
number summarizing the total risk in a portfolio of
financial assets for senior management.
l It has become widely used by corporate treasurers
and fund managers as well as by financial
institutions.
l Bank regulators also use VaR in determining the
capital a bank is required to keep for the risks it is
bearing.

3
Introduction
l VaR was devised at JP Morgan.
l In the aftermath of the 1987 stock market crash,
the CEO demanded that the market risk team
produced a simple report of the potential losses for
the next trading day – on a single page and
delivered no later than 45 minutes after market
close.
l Nowadays, it is a major activity for the bank to help
clients to calculate VaR of their financial asset
portoflios .

4
The VaR Measure
1. Introduction

Definition
VaR is the largest likely loss from market risk (expressed in currency
units) that an asset or portfolio will suffer over a time interval and with
a degree of certainty selected by the user (Choudhry (2006), p.32)

Or
VaR is a measure of the amount that could be lost from a position, portfolio
etc. VaR is generally understood to mean the maximum loss an investment
could incur at a given confidence level over a specified time horizon. (Wilmott
(2006)

5
The VaR Measure
1. Introduction

l When using the VaR measure, an analyst is interested

in making a statement of the following form:
“I am X percent certain that there will not be a loss of more
than V dollars in the next N days”
Where:
l V is the VaR of the portfolio. It is a function of two parameters: the time horizon,
N days, and the confidence level, X percent. It is the loss level over N days that
has a probability of only (100-X) percent of being exceeded.
l Bank regulators require banks to calculate VaR for market risk with N=10 and
X=99.

6
The VaR Measure
1. Introduction

l VaR is an attractive measure because it is easy to

understand. In essence, it asks the simple question:
“How bad can things get?”

l This is the question all senior managers want to answer.

They are very comfortable with the idea of compressing
all the Greek letters for all the market variables
underlying a portfolio into a single number.

7
The VaR Measure
2. Calculation

l The calculation of VaR estimate follows four steps:

1. Determine the time horizon over which the firm wishes to
estimate a potential loss
• This horizon is set by the user,
• In practice time horizons of 1 day to 1 year is the most common.

2. Select the degree of certainty required, which is the confidence

level that applies to the VaR estimate
• For banks a confidence level of 95% is usually enough,
• Regulators may require 99% confidence level
• Senior management and shareholders are often interested in the potential
loss arising from catastrophe situations, such as a stock market crash, so for
them a 99% confidence level is more appropriate.

8
The VaR Measure (cont.)
2. Calculation

3. Create a probability distribution of likely returns for the

instrument or portfolio under consideration
• Several methods may be used. The easiest is a distribution of recent historical
returns for the asset or portfolio which often looks like the curve associated
with the normal distribution.
4. Calculate the VaR estimate
• This is done by observing the loss amount associated with that area beneath
the normal curve at the critical confidence interval value that is statistically
associated with the probability chosen for the VaR estimate in Step 2.

• The Time Horizon: VaR has two parameters: the time horizon in days, and the confidence interval,
X. In practice, analysts almost invariably set N = 1 in the first instance. This is because there is not
enough data to estimate directly the behaviour of market variables over time longer than one day.
The usual assumption is:

N-day VaR = 1-day VaR x N

• Note: this formula is exactly true when the changes in the value of the portfolio on successive
days have independent identical normal distributions with mean zero. In other cases, it is an
approximation. 9
Calculating VaR
l Historical Simulation
l Model-Building
Historical Simulation
Calculating VaR -
Historical Simulation
l Create a database of the daily movements in all
market variables.

l The first simulation trial assumes that the

percentage changes in all market variables are
as on the first day
Historical Simulation continued

l We are at end of day m, we want to simulate

what might happen on the next day m+1
l Suppose we use m days of historical data
l Let v1 be the value of a variable on day 1
l The (m+1) th trial assumes that the changes
of the market variable are as on the 1st day,
so the value of day m+1 is
v1
vm
v0
Calculating VaR -
Historical Simulation

l The second simulation trial assumes that the

percentage changes in all market variables are
as on the second day

l and so on
Historical Simulation continued
l We are at end of day m, we want to simulate
what might happen on the m+2 day

l Suppose we use m days of historical data

l The (m+2)th trial assumes that the changes

in all market variables are as on the 2nd day,
so the value on (m+2) day is, is
v2
vm
v1
Generate 500 Scenarios of what
might happen on day 501
Historical Libor FT All $/£ ………

Date Share
0 0.05 4000 1.51

1 0.05 4010 1.53

2 0.0525 4050 1.51

: : : : :

: : : : :

499 0.06 3500 1.39

500 0.06 3550 1.40

 Example - Scenario 1/ 501st day

Libor 0.06 x 0.05/0.05 = 0.06

(V500 x V1/ V0)
FT ALL Share 3550 x 4010/4000 = 3559
(V500 x V1/ V0)
$/£ 1.40 x 1.53/1.51 = 1.42
(V500 x V1/ V0)
Scenario Libor FT All $/£ ……… Portfolio
Number Share Value £m

1 0.06 3559 1.42 25

2 0.063 3585 1.38 25.7

: : : : : :

: : : : : :

: : : : : :

499 : : : :

500 : : : :
Example - Scenario 500/ 1000th
day

What are the value on Scenario 500/ the

1000th day?
Historical Libor FT All $/£ ………

Date Share
0 0.05 4000 1.51

1 0.05 4010 1.53

2 0.0525 4050 1.51

: : : : :

: : : : :

499 0.06 3500 1.39

500 0.06 3550 1.40

Example - Scenario 500/ 1000th
day
Libor 0.06 x 0.06/0.06 = 0.06
(V500 x V500/ V499)
FT ALL Share 3550 x 3550/3500 = 3601
(V500 x V500/ V499)
$/£ 1.40 x 1.40/1.39 = 1.41
(V500 x V500/ V499)
VaR from Scenarios
l When the portfolio value has been
calculated for all scenarios, we can work
out VaR for 500 scenarios.

l If 500 scenarios and looking for 99% VaR

look at the drop in value occurring with the
5th (500*1%) worst scenario.
2008 Crisis and using Scenarios
l Maybe Financial Institutions weren’t
looking back far enough.
l Prof Clarkson (Cass) – if engineers built
bridges based on ten years of wind data
they would be in big trouble if it blew down
in a ‘20 year’ storm.
Model Building Approach
 Calculating VaR -
The Model-Building Approach

l The main alternative to historical simulation is to

make assumptions about the probability
distributions of the portfolio return

l This is known as the model building approach or

the variance-covariance approach
Microsoft Example
l We have a position worth $10 million in
Microsoft shares
l The volatility (standard deviation) of
Microsoft is 2% per day (about 32% per
year)
l Therefore, one standard deviation
changes of the value of Microsoft shares
per day : 2%*$10million=$20,000
Microsoft Example continued
l We use N=10 and X=99
l (10 day period and 99% confidence)

l The one standard deviation of the change

in the value of portfolio in 10 days is
200,000 10 = $632,456
Microsoft Example continued
l We assume that the expected change in
the value of the portfolio is zero (This is
OK for short time periods)
l We assume that the change in the value of
the portfolio is normally distributed
l Since N(2.33)=0.99, the VaR is

2.33 ´ 632,456 = $1,473,621

AT&T Example
l Consider a position of $5 million in AT&T
l The daily volatility of AT&T is 1% (approx
16% per year)
l The S.D per 10 days is
50,000 10 = $158,144
l The VaR at 99% confidence level is
158,114 ´ 2.33 = $368,405
Portfolio
l Now consider a portfolio consisting of both
Microsoft and AT&T
l Suppose that the correlation between the
returns is 0.3
S.D. of Portfolio
l A standard result in statistics states that

s X +Y = s + s + 2rs X s Y
2
X
2
Y

l In this case sX = 200,000 and sY = 50,000

and r = 0.3. What is the standard
deviation of the change in the portfolio
value in one day?
VaR for Portfolio

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

220,227 ´ 10 ´ 2.33 = $1,622,657
lThe benefits of diversification are
(1,473,621+368,405)–1,622,657=$219,369
The Linear Model
We assume
l The daily change in the value of a portfolio
is linearly related to the daily returns from
market variables
l The returns from the market variables are
normally distributed
The General Linear Model
continued
n
dP = å a dx i i
i =1
n
s = å a i2s i2 + 2å a ia js is j r ij
2
P
i =1 i< j

where s i is the volatilit y of variable i

and s P is the portfolio' s standard deviation
δP – Change in value of whole portfolio in a day.
δxi – Return on asset i in a day.
αi – Amount in asset i.
ρij – correlation between asset i and asset j.
Delta
l Revision - delta measures the number of
stocks needed to mimic the returns of an
option
Linear Model and Options
l Similar when there are many underlying
market variables
dP = å S i D i dxi
i
where Di is the delta of the portfolio with respect to the ith asset
δP – Change in value of whole portfolio in a day.
δSi – Change in stock price i in a day.
Si – Stock price i.
δxi – Return (percentage) on stock price i in a day.
Example
l 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
l As an approximation
dP = 120 ´ 1,000dx1 + 30 ´ 20,000dx2
where dx1 and dx2 are the percentage changes
in the two stock prices
Merits of Historical Simulation
and Model Building
l Historic simulation has the advantage that
historical data determines the joint probability
distribution of the market variables.
l Historical simulation may be computationally
slow
l Historical simulation does not easily allow for
volatility to be updated
Merits of Historical Simulation
and Model Building
l The model building approach can produce
results quickly (if done analytically rather than by
Monte Carlo simulation). It is possible to change
parameters to allow for changes in volatility
l The model building approach has a
disadvantage if it assumes that market variables
have a multivariate normal distribution which is
often not justified.
Potential Problems of VaR

41
Potential Problems of VaR

l Some researchers have argued that VaR may tempt traders to

choose a portfolio with a return distribution similar to that in Figure
18.2.

l In addition, some assets may have heavy tails compared to a

normal distribution.

l The portfolios in Figures 18.1 and 18.2 have the same VaR, but the
portfolio in Figure 18.2 is much riskier because potential losses are
much larger.

42
Potential Problems of VaR

l A measure to that deals with that problem is the “expected shortfall”.

l Whereas VaR asks the question how bad can things get?”, expected
shortfall asks: “If things do get bad, how much can the company
expect to lose?”

l Expected shortfall is the expected loss during an N-day period

conditional that an outcome in the (100-X) percent left tail of the
distribution occurs.

l For example, with X=99 and N=10, expected shortfall is the average
amount the company loses over a 10-day period when the loss is in
the 1% tail of the distribution.

43
 Potential Problems of VaR
l For example, with X=99 and N=10, expected shortfall is the average
amount the company loses is greater than the loss in the 1% tail of
the distribution.

44
Partly Moving to Expected
Shortfall
l For banks’ trading books (not the other
parts of their business), the regulators are
now requiring banks to move to use
expected shortfall by end of 2019.
l Basel Committee of Bank Supervision
consultative paper (2012): Fundamental
Review of Trading Book.
Next Week
l Stress testing
l Banking Regulation