Beruflich Dokumente
Kultur Dokumente
Sohail Zafar
Lecture 6 7
TOTAL RISK
Total risk of an investment, single stock or portfolio of stocks, is uncertainty
about its expected rate of return.
returns expected a year ago when the respective share was bought or the
respective portfolio was constructed.
DPS1 refers to expected annual cash dividends per share during the next year,
P1 refers to expected share price after one year, and P o refers to current
market price of the share.
Total Risk is defined as uncertainty of ROR of a stock; and it is quantified as
variance of the stocks rate of return.
uncertainty about next years ROR but in real life returns of past periods are
used to calculate variance of returns (total risk).
assumption that total risk calculated by using past data of returns is a good
estimate of total risk of the next year. Therefore using historic (past) data of
RORs you can calculate total risk of a share as variance of its rate of
return, and then find its under root to get standard deviation (SD) of returns.
The formula for variance of stock returns when using past returns data is:
Total Risk of Stock i = (VARi) =
In this formulation t refers to time period which can vary from time period 1
to time period n. Generally more observations of historical returns make this
calculation more valid, so instead of using last 5 years returns if you use last
10 years returns data then your VAR is more valid.
In practice, however,
usually monthly returns data for the last 60 months is used to calculate VAR of
returns of a stock. Under root of VAR of returns is standard deviation (SD) of
returns. If monthly returns data of last 60 months were used and the resulting
52
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
PSO
( 0 refers to beginning of
PSO
PSO
VAR
= [(40 - 7)
PSO
+ 324
144
16
289
+ 256) /6
=2118/6
= 353 %2 (note units are percentages squared)
SD
PSO
= VAR
PSO
= 353 %2
= 18.78 % (note units are percentage)
to find under root of 353 do this in FC 100: Hit green button COMPUTE
mode; enter 2; the hit shift key then x key, then enter 353, then
bracket close , then EXE. You get 18.78.
Please note you can do the same calculations variance (of total risk of stock
and SD of stock) returns, using your FC-100 as shown below.
53
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
Hit Green button STAT. You see a menu, choose first option, 1-VAR to
do calculations of one variable, and hit EXE key on bottom right, you see a
data entering screen with one column.
2
Now enter ROR data as: 40 EXE, -11 EXE, -5 EXE, 3 EXE, 24 EXE, -9
EXE.
3
This is SD of PSO returns and it is called total risk of PSO stock. You can find
variance of PSO returns by squaring this number because SD 2 = VARIANCE.
(18.78)2 = 353 . Using FC 100 you can find variance by : Compute mode;
enter 18.78, then black button with inverted v , then enter 2, then close
bracket, then EXE.
each stock, (amount invested in a stock / Your OE) and sum of weights of all
the securities in a portfolio is ALWAYS ONE. Note the mistake occurs when
weight of a stock is worked out as : investment in that stock divided by total
investment in the portfolio; because total investment in the portfolio may be
composed of some of your money (OE) and some of borrowed money. For
example weight of ICI stock in your portfolio is: X
54
ICI
= Rupee investment in
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
ICI shares / your OE. And R 1, R2, till Rn are expected returns on stocks1, stock
2, and till stock n.
Total Risk of Portfolio is variance of its expected ROR and is denoted as
VARp here (or 2p is used as its symbol in most of the text books). Unlike
expected return of portfolio, total risk of portfolio is not weighted average of
total risk of stocks included in that portfolio. Total risk of portfolio is
not
is:
(Equation one)
Since covariance of something with itself is called variance, therefore : COV
is VAR
i, i
(Equation two)
These 4 formulations for total risk of portfolio are just different ways of saying
the same thing, that is, calculate covariance between all pairs of stocks and
multiply (weigh) each covariance with respective weight of those 2 stock in
the portfolio, and then sum all such weighted covariance. In the above
formulae some terms have double summation , such as, XiXj CORRi,j SDi
SDj , it is so because there is correlation between returns of stock i
55
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
and j
terms twice.
Please note that:
SDp = VARp
CORRi,j = COVi , j / (SDi x SDj ).
Correlation of ROR between any two stocks can be between + 1 to -1 , but
Covariance between RORs of 2 stocks can be any number, small or large,
positive or negative.
stocks can be found if you know total risk of each stock (SD) and correlation
between returns of these 2 stocks, as shown below:
COV
i ,j
= CORR
i ,j
* SDi * SD
This is useful because your FC 100 in STAT mode calculates only correlation (
r) and SDs of 2 stocks; and from that output you can calculate covariance
between returns of 2 stocks as shown above.
Stock j). Opening the second term for three stock portfolio gives: + X i Xj
COVi, j =
+ X1X2Cov1, 2 + X1X3 Cov1,3
X1X2COV
2,1
X3X1COV3,1
X1X3COV
1,3
X2X2VAR2
X2X3COV
2,3
X3X2COV3,2
X3X3VAR3
1,2
56
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
and
between j and i is same and therefore this number has to be added twice
because each box above the the diagonal has same data as a box below the
diagonal, e.g. Cov1,3 and Cov
3,1
twice , once above the diagonal and once below the diagonal;therefore double
summation sign is used in the formula.
written as X1X1COV1,1; then the whole matrix is a matrix of covariance and the
restriction i can not be j is no more there. Then total risk of portfolio formula
can be written as :
VARp = Xi Xj COV i,j
And that is just saying in mathematical notations that add all weighted
covariance in a matrix; for a 3-stock portfolio such a matrix has 9 boxes and
thus 9 covariance; for a 100-stock portfolio such a matrix has 10,000 boxes
and 10,000 covariance.
As COV
i,j
weights (Xs) in your portfolio to those stocks which have low or negative
correlation because doing so would also gives lower total risk of portfolio. One
important aspect of the job of security analysts is to identify pairs of stocks
whose returns have lower, or ideally negative, correlation.
Number of estimates needed to calculate Total Risk of Portfolio(VAR p)
From the above example of 3 stock portfolio, you need to estimate 3 VAR i for 3
stocks, and n(n - 1) = 3(3 - 1) = 6 COVs. Total estimates needed were: 3
variances + 6 covariances = 9, i.e. n2 =32 = 9. The total number of boxes in
a 3 x 3 matrix is 9.
are: n(n - 1)/2= 3(3 - 1)/2= 6/2 =3. Total number of unique estimates needed
to estimate total risk of a 3-stock portfolio are: 3 variance + 3 covariance = 6
Similarly for 100- stock portfolio, to estimate its total risk (VAR p ) you need
estimates of 100 VARi of 100 stocks, and n(n - 1) = 100(100 - 1) = 9,900 COVs
57
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
between all pairs of stocks, and n(n - 1)/2 =100(100 - 1)/2= 9,900/2 = 4,950
unique COVs.
As a generalization, to estimate total risk of a portfolio,
total number of
unique estimates needed are n + n(n - 1)/2. That is n VARs for n stocks
in that portfolio; and n(n - 1) /2 unique COVs. For 100-stock portfolio, 100
variance of 100 stocks + 4,950 unique covariance between the pair of stocks,
and it adds up to total 5,050 unique estimates of variance and covariance of
returns that you need as inputs to estimate total risk of a 100-stock portfolio.
You can also estimate 4,950 correlations instead of covariance.
This sheer number of input estimates was daunting in 1950s when the Modern
Portfolio Theory (MPT) was presented by Markowitz, therefore in spite of the
elegance of the theory, practitioners could not apply it in real life.
By 1970s
two developments took place, first, the emergence of computers, and second
the simplifications proposed by Sharpe, Lintner, and Mossin that significantly
reduced the required number of input estimates for the calculation of total
risk of portfolio (VARp).
popularity among the practitioners since 1970s ; and a whole new industry
known by such titles as Money Management, Investment Management, Funds
Management, Mutual Funds , etc , took off; and now in USA individuals have
more of their savings invested through mutual funds than placed in bank
deposit accounts.
1
2
3
COVs
146
187
145
187
854
104
145
104
289
58
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
X2
X3
X
1
0.2325x0.3605x145
= 12.15
X
2
0.407x0.3605x104
= 15.25
X
3
0.3605x0.23235x145=12.15
0.3605x0.407x104=15.25
0.3605x0.3605x289
= 37.55
Total risk of portfolio ,VAR P , is sum of the data of all the 9 boxes in the above
matrix
VAR
37.55
VAR
= 277.10%
SDp =
VARp
277.1%2
SDp =
SDp
16.46%
You must have noticed that certain numbers are appearing twice in the above
matrix; these are the terms for which double summation sign was used.
Therefore a simpler way of writing this variance formula for a 3-stock portfolio
is:
VARp = X12VAR1 + X22VAR2 + X32VAR3 +
X1X3COV
1,3
VARp = 7.9
) + 2(X2X3COV
+ 141.6
2(X1X2COV
1,2
2(
2,3
+ 37.55
2(17.7)
2(12.15)
+ 2(15.25)
VARp = 277.1
Whereas 1, 2 and 3 are different stocks such as ICI, UBL, and PSO. Similarly
for portfolios made up of more than 3 stocks you can extend the above
formula.
59
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
Estimates for total risk (standard deviation of RORs) and correlation of RORs of
3 stocks, A, B, and C, have been provided by your staff of security analysts.
And you, as portfolio manager, have decided to build the portfolio whose
weights (Xs) are:
XA = 0.20 , i.e. you have invested 20% of your OE in stock A.
XB = 0
XC = 0.80
S.D
12%
15%
10%
A
B
C
1
-1
0.2
1
-0.2
Solution
You may choose to use data of Correlations and SDs, or you may first convert
correlations into covariance. Let us find COVs first, Note COV
i,i
= VARi , and
= 1 X 12 X 12
= -1 X 12 X 15
= 44%
= -180%2
= 1 X 15 X 15
= 0.2 X 12 X 10
= 225%
= 24%2
CovB,C = CorrB,C SDB SDC
= -0.2 X 15 X 10
=100%2
= 30%
60
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
Note: Since weight of stock B is zero, so you used only Stocks A & C to build
this portfolio, Stock B is not in your portfolio, it is a 2- stock portfolio, its VAR p
formula in matrix form would have only 4 boxes
Xa Xa COVa,a
Xa Xc COV
Xc Xa COV
c,a
Xc Xc COV
As COV
a,a
Xa Xa VARa
Xc Xa COV
Xa Xc COV
c,a
a,c
c,c
a,c
Xc Xc VAR,c
Please note that on the diagonal are variances and off-diagonal terms are
covariance. You can write it in a equation format:
VARp
a, c
= COV
c,a
, so instead of writing it
SDp
= 81.28
= 9%
Similarly for a 3-stock portfolio you can write formula of total risk as :
VARp
= X 2a VARa
2(XaXcCOVa,c) + 2(XbXcCOVb,c)
61
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
RICI
RPSO
1999
10%
-5%
2000
-5%
7%
2001
20%
2%
2002
15%
4%
Find
1) Total Risk of ICI Stock, i.e. VARICI & SDICI
2) Total Risk of PSO Stock, i.e., VARPSO & SDPSO
3) COV
ICI,PSO
4) CORR
ICI, PSO
Please note that in real life, security analysts use monthly returns of last 60
months as data set to estimate correlation (or covariance) between returns of
any 2 stocks. Here a shorter and annual return data set is used to save time.
These annual realized returns were calculated as:
[DPS / P
begin
] + [(P
end
- P
begin
) / P
begin
ICI
VARICI = [(R99 - RAvg )2 + (R2000 - RAvg )2 + (R2001 - RAvg )2 + (R2002 - RAvg )2 ]/n
= {(10-10)2 + (-5-10)2+ (20-10)2 + (15-5)2 }/ 4
= {0 + 225 + 100 + 25} / 4
= 87.5%2
SDICI = 87.5%2
= 9.35%
We are using mean returns of ICI as proxy for the expected returns i.e. Avg R ICI
is a proxy for expected ROR of ICI stock, so one degree of freedom is lost and
therefore denominator should be n 1 if it is sample data; and n should be
denominator if it is population data. As for these years from 1999 to 2002 the
returns are not sample but actual returns in these years earned by the
shareholders of these two stocks so it is population data, and n is used as
62
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
PSO
= 19.5%
=4.41%
COVICI, PSO = [(R ICI99-R ICIAvg )(R PSO99 - R PSOAvg) + (R ICI2000 R ICIAvg)(R PSO2000
- R PSOAvg) +
(R ICI 2001 R ICIAvg )(R PSO2001 R PSOAvg) + (R ICI2002 R ICIAvg)(R PSO2002
R PSOAvg )] / n
COVICI, PSO = [(10-10)(-5-2) + (-5-10)(7-2) + (20-10)(2-2) + (15-10)(4-2)] / 4
COVICI, PSO = (0-75+0+10)/4
= -65/4
= -16.25%2
Many text books use returns data in decimal format, that is, 12% is written as
0.12.
expected ROR of ICI and PSO were not available and means were used as
proxy for those, therefore 2 degrees of freedom are lost and denominator of
formula is n 2; but we have used n as denominator, that implies it was
assumed that the data was population data, not the sample data.
COV of returns of 2 stocks can be a positive or negative number, it can be a
small or a large number, and its units are percentages squared which is
something
not
easily
conceptualized,
therefore
it
is
standardized
as
ICI, PSO
= COV
ICI, PSO
/ (SDICI x SDPSO )
returns of these 2 stocks are taken for the last 100 years, then in 39 years
63
Investment Analysis & Portfolio Management. MBA II. Spring 2015. Lahore School of Economics. Dr. Sohail Zafar
when returns of ICI stock were going up compared to its last years returns,
the returns of PSO stock were going down compared to its last year returns.
You can find correlation and then covariance between returns of 2
stocks using FC 100 calculator as shown below:
1. Green button STAT; on the menu choose item A + BX to do
calculations of 2 variables, and then hit EXE
2. On the data entering screen you see 2 columns X and Y. Enter returns
of ICI in column X and returns of PSO in column Y.
3. Hit red button AC
4. Hit SHIFT and STAT buttons, you will see a menu, choose item 7
regression
5. On the new menu choose item 3, r for correlation; and EXE. You see
-0.39. This is correlation of returns between ICI and PSO stocks as you
calculated by hand earlier.
6. To find Covariance of returns of ICI and PSO you need the SD of the 2
stocks.
Hit SHIFT and STAT again, from the menu choose item 5
Var . On the next menu choose item 3 and EXE ; you get 9.35, it is
7.
SD of ICI.
Again hit SHIFT and STAT and choose item 5, and on the next
Please note the skill to use calculator for calculating total risk of a stock and
correlation between returns of any pair of stocks is a must learning for you
in this course, and a clear understanding of their meanings is also essential.
Also notice that correlation of returns of any 2 stock is not in percentages or
any other units of measurement, it is just a number; while covariance is
percentages squared.
64