0% found this document useful (0 votes)
74 views32 pages

Chapter 2

This document discusses different risk measures used to assess investment risks, including semi-variance, value at risk (VaR), and expected shortfall. It provides definitions and examples of each measure. Semi-variance does not satisfy important properties of a coherent risk measure like monotonicity and subadditivity. VaR satisfies monotonicity by definition but not other properties. Expected shortfall is introduced as another risk measure.

Uploaded by

Ping Leung
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
74 views32 pages

Chapter 2

This document discusses different risk measures used to assess investment risks, including semi-variance, value at risk (VaR), and expected shortfall. It provides definitions and examples of each measure. Semi-variance does not satisfy important properties of a coherent risk measure like monotonicity and subadditivity. VaR satisfies monotonicity by definition but not other properties. Expected shortfall is introduced as another risk measure.

Uploaded by

Ping Leung
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 32

AIN3220

Investment and Risk Analysis

Chapter 2
Chapter 2: Risk Measures
Question:
How do you assess investment risks?
Topics in this Chapter:
2.1 What is Risk?
2.2 Coherent Risk Measure
2.3 Risk Measure: Semi-Variance
2.4 Risk Measure: Value at Risk (V@R)
2.5 Risk Measure: Expected Shortfall (Conditional Value at Risk)
References: Hull (2018) Chapter 12

AIN3220 Investment and Risk Analysis 2-1


2.1 What is Risk?
In everyday speech, the word risk is associated with the possibility of nega-
tive outcomes.
In this course, we are only concerned with investment or market risks
when investing in the stock markets.
For the assets to be considered risky, the final outcome must be uncertain
and there must be some possibility that the final outcome will
have negative values.
Example 2.1:
Consider Portfolio A whose return, measured in $ millions, has the following
distribution:

$10
 with pA($10) = 0.92,
rA = −$10 with pA(−$10) = 0.06, (2.1)

−$20 with pA(−$20) = 0.02.

AIN3220 Investment and Risk Analysis 2-2


In this chapter, we focus on the loss of the asset, instead of the return of the
asset, for the ease of discussion.
Example 2.2: Recall from Example 2.1 that

$10
 with pA($10) = 0.92,
rA = −$10 with pA(−$10) = 0.06, (2.2)

−$20 with pA(−$20) = 0.02.

Denote LA to be the loss of Portfolio A, then we have LA = −rA, i.e.



−$10 with pA(−$10) = 0.92,

LA = $10 with pA($10) = 0.06,

$20 with pA($20) = 0.02.

AIN3220 Investment and Risk Analysis 2-3


Risk measure is used to determine the amount of liquid capital (i.e.
cash) needed to be set aside as reserve due to risky exposures of an
investment.
Hence, riskier investment would have higher risk measure, i.e. requires
higher reserve.
In this chapter, we study the following popular risk measures:
• Semi-Variance
• Value at Risk
• Expected Shortfall
These risk measures only focus on measuring negative outcomes.

AIN3220 Investment and Risk Analysis 2-4


2.2 Coherent Risk Measures
Before studying these risk measures, let us first state the set of criteria under
which a risk measure is a good measure:
A risk measure ρ is said to be a Coherent Risk Measure if it satisfies the
following four properties,
• Monotonicity
If X ≤ Y , then ρ(Y ) ≤ ρ(X).
In words: If a portfolio performs worse than another portfolio in every
market scenario, then its risk measure should be greater.
• Positive Homogeneity
For c > 0, ρ(cX) = cρ(X)
In words: Changing the size of a portfolio by a factor c > 0 should
result in its risk measure being multiplied by c.
• Cash Invariance
For K > 0, ρ(X + K) = ρ(X) − K
In words: Adding cash $K to a portfolio would result in reduction
of its risk measure by $K.
AIN3220 Investment and Risk Analysis 2-5
• Subadditivity
ρ(X + Y ) ≤ ρ(X) + ρ(Y )
In words: The risk measure of two portfolios after they have been
merged should be no greater than the sum of their risk measures
before they were merged. This supports the concept of diversifica-
tion!
Let us now check which of the following popular risk measures:
• Semi-Variance (ρ = SVar)
• Value at Risk (ρ = V @R)
• Expected Shortfall (ρ = ES)
satisfies the properties of a coherent risk measure.

AIN3220 Investment and Risk Analysis 2-6


2.3 Risk Measure: Semi-Variance
Denote L to be a discrete random variable representing the loss of a portfolio,
i.e. L = li with probability pL(li) ≥ 0, for i = 1, ..., M . Then, the risk
measure defined by the Semi-Variance of the portfolio loss is given by
X
SVar(L) = pL(li) (li − µL)2 ,
li >µL

where µL is the expected loss of the portfolio.


P
The symbol li>µL means that only adding up the term li > µL, for i =
1, ..., M .

AIN3220 Investment and Risk Analysis 2-7


2.3.1 Semi-Variance: Properties and Examples

Semi-Variance does NOT satisfies Monotonicity.


Example 2.3:
Consider Portfolio A and B, measured in $ millions, with the following loss
distribution:

−$10 with pA(−$10) = 0.92,

LA = $10 with pA($10) = 0.06, (2.3)

$20 with pA($20) = 0.02.


−$15 with pB (−$15) = 0.92,

LB = $5 with pB ($5) = 0.06, (2.4)

$15 with pB ($15) = 0.02.

Show that Semi-Variance does not satisfy the property of Monotonicity, i.e.
A < B, but SVar(LB ) ≮ SVar(LA).

AIN3220 Investment and Risk Analysis 2-8


Answer:
Note first that Portfolio A performs worse than Portfolio B in every market
scenario, i.e. A < B. (Why?)
Since the expected loss of Portfolios A and B are:

µA = −$10 × 0.92 + $10 × 0.06 + $20 × 0.02 = −$8.2 million,


µB = −$15 × 0.92 + $5 × 0.06 + $15 × 0.02 = −$13.2 million,

Since µA = −$8.2 million and µB = −$13.2 million, the semi-variance of loss


of Portfolios A and B are:

SVar(LA) = ($10 − (−$8.2))2 × 0.06 + ($20 − (−$8.2))2 × 0.02 = 35.78,


SVar(LB ) = ($5 − (−$13.2))2 × 0.06 + ($15 − (−$13.2))2 × 0.02 = 35.78.

In other words, SVar(LA) = SVar(LB ) = 35.78, which shows that Semi-


Variance does NOT satisfy the property of Monotonicity.

AIN3220 Investment and Risk Analysis 2-9


Semi-Variance does NOT satisfies Subadditivity.
Example 2.4:
Consider two perfectly positively correlated portfolios, Portfolios C and
D, measured in $ millions, with an identical loss distribution as follows: For
i = C, D
(
−$1 with pi(−$1) = 0.98,
Li = (2.5)
$5 with pi($5) = 0.02.

Show that Semi-Variance does not satisfy the property of Subadditivity, i.e.
SVar(LC+D )  SVar(LC ) + SVar(LD ).
Answer:
Since the loss distributions of Portfolios C and D are identical, the semi-
variances of the loss of Portfolios C and D are given as follows:

SVar(LC ) = SVar(LD ) = 0.6915.

On the other hand, recall that the loss of the combined portfolios C + D is as

AIN3220 Investment and Risk Analysis 2-10


follows:
(
−$2 with pC+D (−$2) = 0.98,
LC+D = (2.6)
$10 with pC+D ($10) = 0.02,

which yields

SVar(LC+D ) = 2.7660 > 0.6915 + 0.6915 = 1.3830 = SVar(LC ) + SVar(LD ),

i.e. Semi-variance does NOT satisfy the property of sub-additivity.


Exercise: Verify why Semi-Variance does NOT have the properties of Pos-
itive Homogeneity and Cash Invariance.
In summary, Semi-Variance is NOT a Coherent Risk Measure!

AIN3220 Investment and Risk Analysis 2-11


2.4 Risk Measure: Value at Risk (V@R)
Definition Denote L to be the loss of a portfolio. For a given confidence
level, p%, the Value-at-Risk(V@R), denoted as p%V@R, is defined as

P[L ≥ p%V@R] = (100 − p)%.

In words, we are p% confident that the loss of the portfolio would not exceed
$p%V @R.
Example 2.5: When a risk manager says the one-day 99%V @R of Portfolio
A is $10 million, it means that there is a 99% chance that the loss of Portfolio
A does NOT exceed $10 million.

AIN3220 Investment and Risk Analysis 2-12


(100 − 𝑝)%
𝐿𝑜𝑠𝑠
𝐿𝑜𝑠𝑠

𝐺𝑎𝑖𝑛 0 𝑝% 𝑉𝑎𝑅 𝐿𝑜𝑠𝑠


𝐿𝑜𝑠𝑠 𝐿𝑜𝑠𝑠 𝐿𝑜𝑠𝑠 𝐿𝑜𝑠𝑠
𝐿𝑜𝑠𝑠Figure 2.1: Calculation of V@R
𝐿𝑜𝑠𝑠from the Probability 𝐿𝑜𝑠𝑠
Distribution of𝐿𝑜𝑠𝑠
the Loss
in the Portfolio Value Gains are negative losses; confidence level is p%; V@R
level is V .
AIN3220 Investment and Risk Analysis 2-13
2.4.1 V@R: Properties and Examples
V@R satisfies Monotonicity.
Example 2.6:
Consider two portfolios, Portfolio A and B. Denote LA and LB as the losses
of Portfolios A and B, measured in $ millions, respectively.

−$10 with pA(−$10) = 0.92,

LA = $10 with pA($10) = 0.06, (2.7)

$20 with pA($20) = 0.02.


−$15 with pB (−$15) = 0.92,

LB = $5 with pB ($5) = 0.06, (2.8)

$15 with pB ($15) = 0.02.

Show that 95%V @R satisfies the property of monotonicity in this example,


i.e. A < B, but 95%V @RB < 95%V @RA.

AIN3220 Investment and Risk Analysis 2-14


Answer:
Note that

P[LA ≥ $20] = 0.02,


P[LA ≥ $10] = 0.02 + 0.06 = 0.08.

Therefore 95% V @RA is $10 million.


Simiarly,

P[LB ≥ $15] = 0.02,


P[LB ≥ $5] = 0.02 + 0.06 = 0.08.

Therefore, the 95%V @RB is $5 million.


Hence, it follows that 95%V @RB < 95%V @RA, i.e. V@R satisfies the prop-
erty of Montonicity in this example.

AIN3220 Investment and Risk Analysis 2-15


V@R satisfies Positive Homogeneity.
Example 2.7:
Recall Portfolio A in Example 2.6. Consider a multiplier c = 2.
Show that 95%V @R satisfies the property of Positive Homogeneity in this
example, i.e. 95%V @R(cA) = c × 95%V @R(A).
Answer:
Denote 2× Portfolio A as Portfolio 2A. Portfolio 2A, measured in $ millions,
has the following loss distribution:

−$20 with p2A(−$20) = 0.92,

L2A = $20 with p2A($20) = 0.06, (2.9)

$40 with p2A($40) = 0.02.

Then, the 95%V @R2A is $20 million = 2 × 95%V @RA = 2 × $10 million.
The 95%V @R satisfies the property of Positive Homogeneity in this example.

AIN3220 Investment and Risk Analysis 2-16


V@R satisfies Cash Invariance.
Example 2.8:
Recall from Example 2.6 that a new portfolio, denoted as Portfolio Ã, is
created by adding cash K = $5 million to Portfolio A.
Show that 95% V@R satisfies the property of Cash Invariance in this example,
i.e. 95%V @RÃ = 95%V @RA − K.
Answer:
Portfolio à has the following loss distribution:

−$15 with pÃ(−$15) = 0.92,

LÃ = $5 with pÃ($5) = 0.06, (2.10)

$15 with pÃ($15) = 0.02.

Then, the 95%V @RÃ is $5 million = 95%V @RA − $5 million = $10 million
−$5 million.
95%V @R satisfies the property of Cash Invariance in this example.

AIN3220 Investment and Risk Analysis 2-17


V@R does NOT satisfy Subadditivity.
Example 2.9:
Consider two independent portfolios, Portfolio E and F , measured in $
millions, with an identical loss distribution as follows: For i = E, F ,
(
$1 with pi($1) = 0.98,
Li = (2.11)
$10 with pi($10) = 0.02.

Calculate the 97.5%V @R for Portfolio E, F , a combined Portfolio E + F .

AIN3220 Investment and Risk Analysis 2-18


Answer:

Since

P[LE ≥ $10] = 0.02,


P[LE ≥ $1] = 0.98 + 0.02 = 1,

the 97.5%V @RA is $1 million. Since Portfolios E and F are identical, we have

97.5%V @RE = 97.5%V @RF = $1 million.

Consider now that we combine these two portfolios together and denote LE+F
as the loss associated with this combined portfolio. Then the loss of this
combined portfolio has the loss distribution:

$20 with pE+F ($20) = 0.02 × 0.02 = 0.0004,

LE+F = $11 with pE+F ($11) = 2 × 0.02 × 0.98 = 0.0392,

$2 with pE+F ($2) = 0.98 × 0.98 = 0.9604.

AIN3220 Investment and Risk Analysis 2-19


Since,

P[LE+F ≥ $20] = 0.0004,


P[LE+F ≥ $11] = 0.0392 + 0.0004 = 0.0396,

the 97.5%V @RE+F for this combined portfolio is $11 million.


Therefore, it follows that

$11 million = 97.5%V @RE+F > 97.5%V @RE + 97.5%V @RF = $2 million,

i.e., V@R is NOT subadditive.


In summary, V@R is NOT a Coherent Risk Measure!

AIN3220 Investment and Risk Analysis 2-20


2.5 Risk Measure: Expected Shortfall
Beyond V@R, we may also want to know how big the loss will be when we have
an exceedance event. This naturally brings us to the concept of Expected
Shortfall:
Definition
Expected shortfall is an expected value of a loss, given an exceedance of
$p%V @R, i.e.

ESp% = E[L|L ≥ p%V @R],

where E[.|L ≥ p%V @R] denotes the conditional expectation.


Example 2.10
When the 99% expected shortfall of an investment is $10 million, i.e. ES99% =
$10 million, it means that, for the loss exceeding its 99% V@R, the average
loss is $10 million.

AIN3220 Investment and Risk Analysis 2-21


(100 − 𝑝)%
𝐿𝑜𝑠𝑠
𝐿𝑜𝑠𝑠

𝐺𝑎𝑖𝑛 0 𝑝% 𝑉𝑎𝑅 𝐿𝑜𝑠𝑠


𝐿𝑜𝑠𝑠 𝐿𝑜𝑠𝑠 𝐿𝑜𝑠𝑠 𝐸𝑆𝑝% 𝐿𝑜𝑠𝑠
Figure
𝐿𝑜𝑠𝑠 2.2: Probability Distribution 𝐿𝑜𝑠𝑠for Gain in Portfolio 𝐿𝑜𝑠𝑠Value.𝐿𝑜𝑠𝑠
Confidence𝐿𝑜𝑠𝑠
Level is p%. Portfolio has the same V@R level, V , as in Figure 2.1, but a large
loss is more likely. 𝐿𝑜𝑠𝑠
AIN3220 Investment and Risk Analysis 2-22
2.5.1 Expected Shortfall: Properties and Examples

Expected Shortfall satisfies Monotonicity.


Example 2.11
Consider two portfolios, Portfolio A and B, measured in $ millions, with the
following loss distributions:

−$10 with pA(−$10) = 0.92,

LA = $10 with pA($10) = 0.06, (2.12)

$20 with pA($20) = 0.02.


−$15 with pB (−$15) = 0.92,

LB = $5 with pB ($5) = 0.06, (2.13)

$15 with pB ($15) = 0.02.

Show that 95% Expected Shortfall satisfies Monotonicity in this example.

AIN3220 Investment and Risk Analysis 2-23


Answer:
From Example 2.6, we know that 95%V @RA = $10 million and 95%V @RB =
$5 million.
Hence, the 95% expected shortfall for Portfolio A and B are as follows:
A 0.02 0.03
ES95% = × $20 + × $10 = $14 million,
0.05 0.05
B 0.02 0.03
ES95% = × $15 + × $5 = $9 million.
0.05 0.05
B A
Therefore, ES95% < ES95% , i.e. expected shortfall satisfies monotonicity in
this example.

AIN3220 Investment and Risk Analysis 2-24


Expected Shortfall satisfies Positive Homogeneity.
Example 2.12:
Recall from Example 2.7 that the Portfolio 2A with the following loss dis-
tribution:

-$20 with
 p2A(−$20) = 0.92,
L2A = $ 20 with p2A($20) = 0.06, (2.14)

$40 with p2A($40) = 0.02.

Show that 95% Expected Shortfall satisfies Positive Homogeneity in this ex-
ample.

AIN3220 Investment and Risk Analysis 2-25


Answer:
From Example 2.7, we know that 95%V @R2A = $20 million.
Then, the 95% expected shortfall for Portfolio 2A is
2A 0.02 0.03
ES95% = × $20 + × $40 = $28 million.
0.05 0.05
A
On the other hand, from Example 2.11, we know that ES95% = $14 million.
2A A
This shows that ES95% = 2 × ES95% , i.e. expected shortfall satisfies the
property of positive homogeneity in this example.

AIN3220 Investment and Risk Analysis 2-26


Expected Shortfall satisfies Cash Invariance.
Example 2.13:
Recall from Example 2.8 that Portfolio à is created by adding cash K = $5
million to Portfolio A. Then, Portfolio à has the following loss distribution:

-$15 with pÃ(−$15) = 0.92,

LÃ = $5 with pÃ($5) = 0.06, (2.15)

$15 with pÃ($15) = 0.02.

Show that 95% Expected Shortfall satisfies Cash Invariance in this example.

AIN3220 Investment and Risk Analysis 2-27


Answer:
Recall from Example 2.8 that 95%V @RÃ = $5 million.
Hence, the 95% expected shortfall for Portfolio à is
à 0.02 0.03
ES95% = × $5 + × $15 = $9 million.
0.05 0.05
A
On the other hand, recall from Example 2.11 that ES95% = $14 million.
à A
This shows that ES95% = ES95% − $5 million, i.e. the expected shortfall
satisfies the property of cash invariance in this example.

AIN3220 Investment and Risk Analysis 2-28


Expected Shortfall satisfies Subadditivity.
Example 2.14
Recall Example 2.9. Consider two independent portfolios, Portfolio E
and F with an identical loss distribution as follows: for i = E, F ,
(
$1 with pi($1) = 0.98,
Li = (2.16)
$10 with pi($10) = 0.02.

Calculate the 97.5% expected shortfall for Portfolio E, F , a combined Portfolio


E + F.

AIN3220 Investment and Risk Analysis 2-29


Answer:
Recall from Example 2.9 that 97.5%V @RE = 97.5%V @RF = 1 million.
Hence, the 97.5% expected shortfalls for Portfolios E and F are
E F 0.02 0.005
ES97.5% = ES97.5% = × $10 + × $1 = $8.2 million.
0.25 0.25
On the other hand, the combined portfolio E + F has the following loss dis-
tribution:

20 with pE+F ($20) = 0.02 × 0.02 = 0.0004,

LE+F = 11 with pE+F ($11) = 2 × 0.02 × 0.098 = 0.0392,

2 with pE+F ($2) = 0.98 × 0.98 = 0.9604.

Next, from Example 2.9, we also know that 97.5%V @RE+F = $11 million.
Hence, the 97.5% expected shortfall for the combined portfolios E + F is
E+F 0.0004 0.0246
ES97.5% = × $20 million + × $11 million = $11.14 million.
0.025 0.025

AIN3220 Investment and Risk Analysis 2-30


It follows that
E+F E E
$11.14 million = ES97.5% < ES97.5% + ES97.5%
= $8.2 million + $8.2 million = $16.4 million.

Hence Expected Shortfall satisfies the property of subadditivity.


In summary, Expected Shortfall is a Coherent Risk Measure!

AIN3220 Investment and Risk Analysis 2-31

You might also like