Value at Risk
relevant certification Financial Risk Management
(FRM)
m 1 ui2
2v
exp
2v
i 1
VaR
VaR
Risk Management and Financial Institutions 3e, Chapter 9,
Copyright © John C. Hull 2012 11
Desirable properties of a risk
measure (Artzner 1999; Jorion)
Monotonicity: if portfolio 1 has systematically lower
returns than portfolio 2 over all states (portfolio 2
strictly dominates 1), portfolio 1 has higher (downside)
risk (note mean-var & stochastic dominance)
Translation invariance: adding W amount of cash to a
portfolio should reduce its risk by W
Homogeneity: scale up a portfolio by a factor K should
increase its risk by a factor K (assume that the larger
portfolio does not increase liquidity risk)
Sub-additivity (Merged portfolio 1 & 2 has less risk
than risk of 1 plus risk of 2 (diversification benefit)
VaR(1) + VaR(2) > VaR (1+ 2)
Options, Futures, and Other Derivatives, 8th Edition,
Copyright © John C. Hull 2012 12
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 fails to address the extreme value (worst
case scenario); C-VaR (expected loss conditional on
loss exceeding VaR) or expected shortfall is an
improvement
VaR tells you 99% chance the loss is less than X
C-VaR tells you if the VaR is breached, what is the
expected value
Both measures do not indicate the
Options, Futures, and Other Derivatives, 8th Edition,
worst case
scenario; Extreme value
Copyright © John theory
C. Hull 2012 13
Computation steps VaR (Jorion)
Mark to market the current portfolio value (e.g., $100
million); expected return = 0
Measure the standard deviation of the portfolio return
(e.g., 15% p.a.)
Set Time horizon (e.g., 10 trading days)
Set confidence level (e.g., 99%; that translate into 2.33
std deviation for normal dist.)
VaR = $100 m (15%) (10/252)^1/2 (2.33) = $7 m; red term
is std dev of return for the 10-day period
+ve statement: 99% confident that the loss is below $7 m
within a 10 day period; -ve statement: there is a 1%
chance that the loss will exceed $7m
Options, Futures and Other Derivatives, 8th Edition,
Copyright © John C. Hull 2012 14
Relationship between time and
variance of returns
Variance of return per year = variance of
return per day (number of days)
Variance of return is linear in Time
Nobel prize laureate (Eugene Fama): efficient
market hypothesis EMH
EMH implies that securities price changes as
random walk; iid is stronger than random
walk
The best prediction of the next price is the
Options, Futures, and Other Derivatives, 8th Edition,
current price; the© John
Copyright returns are uncorrelated 15
C. Hull 2012
What if the returns are
correlated? (ch 23?)
If the stock market exhibit some sort of
herding behavior, algorithm trading too)
Market goes up, people follow and expect the
market to go further up
Up market is fuel by further buying
Asset prices blow up!
Bubble
11,173.59
Example of Calculation: 11,022.06 10,977.08
11,219.38
s X +Y s 2X + sY2 + 2rs X s Y
i 1 j 1
n n
Variance of Portfolio Return P P 2 2
r
i 1 j 1
ij wi w js is j
n n
Standard deviation Portfolio Return P P r
i 1 j 1
ij wi w js is j
n n
s 2P cov ij a i a j
i 1 j 1
s 2P α T Cα
n n
s cov ij a i a j
2
P
i 1 j 1
s 2P α T Cα
P
d
S
Define
S
x
S
Option price change is driven by stock price
movement
Options, Futures, and Other Derivatives, 8th Edition,
Copyright © John C. Hull 2012 50
What drives what?
Linear thinking (cause and effect): A Derivative is written against
an asset S
Price change of the derivatives (futures, call, and put options) is
driven by movement of the underlying asset S
However, what about the statement that the market movement
is trigger by futures price change?
S and F are perfectly correlated; S and F move in perfect
unison; in a perfect market, cannot tell which one is the driver;
no trading opportunity
S moves ahead of F; trade F after seeing S move
F moves ahead of S (norm); informed traders trade derivatives
ahead of S. trade S after seeing F moves; the question is how
long and how much ahead? 1/10 of a second, 1 second, 3-
minutes, how much; 1 bp, 10 bps that determine whether the
trade is viable after deducting
Options, Futures, costs and
and Other Derivatives, fees
8th Edition,
Copyright © John C. Hull 2012 51
Linear Model and Options
continued (equations 20.3 and 20.4)
Then
P d S Sd x
P S d
i
i i x i
54
How to obtain VaR with delta P
Given the quadratic form in estimating
changes in the value of the option portfolio
Two approaches
1. Simulation with historical return data
2. Monte Carlo simulation by using random
draw from a normal distribution (refer to the
ppt for the Black-Scholes-Merton basic
assumption on the return generating
process: geometric Brownian motion
Options, Futures, and Other Derivatives, 8th Edition,
Copyright © John C. Hull 2012 55
But the distribution of the daily return
on an option is not normal
Asset Price
Asset Price
Short
Call
Options, Futures, and Other Derivatives, 8th Edition,
Copyright © John C. Hull 2012 59
Options, Futures, and Other Derivatives, 8th Edition,
Copyright © John C. Hull 2012 60
Quadratic Model
2
this becomes
1 2
P Sd x + S (x) 2
2