Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 44

Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.

Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Lecture: 17 to18
Simplification by Sharpe
As you saw in earlier classes that the Modern Portfolio Theory by Markowitz requires “n” estimates of
variances and “ n(n - 1)/2” estimates of unique covariances as inputs to calculate total risk of portfolio
(VARp). Thus the theory required calculating a large number of variances and covariances. These
computational requirements were daunting in 1950s when the theory was presented by Markowitz.

Sharpe simplified the formula of total risk of portfolio ; and his formula
requires calculation of fewer estimated inputs for calculating total risk of portfolio. Let us understand in
some detail the simplification proposed by Sharpe in calculations of total risk of portfolio.

Sharpe proposed that ROR of any security (Ri) is sensitive to ROR of overall stock market (R M) in a linear
fashion. Up till now we have looked at expected ROR of a stock as composed of 2 components, namely:
expected capital gains yield plus expected dividend yield , i.e.:
Expected ROR of a stock, Ri = {(P1 – P0) / P0 } + DPS1/P0 .

But Sharpe argued that historical data has shown that as stock market’s ROR, symbolized as RM ,
increases or decreases, RORs of most of the stocks also increase and decrease; he further argued that this
relationship is linear. If his assumption is accepted then it follows that relationship of return of any stock
with the stock market returns can be written as a linear model:

Expected Ri = αi + βiRM + ei
Y = A + B* X
Depend Variable = intercept + Slope* Indep variable
Whereas: i refers to any stock , such as MCB
Ri ……..is the dependent variable on y-axis, i.e. expected ROR of any stock.
αi ........intercept of a straight line with y-axis, ROR of stock i when R M (ROR of Stock Market) was zero
βi ……. Slope of a straight line. It tells 1 %age point change in R M causes what %age point change in
ROR of Stock i
RM…….independent variable on x-axis, i.e. expected ROR of the overall stock market. Usually %age
change in an index such as KSE-100 index is used as proxy for RM
ei……..random error term. Expected value (or mean) of error term is assumed to be
zero, but it does have a variance which is denoted as VARei (variance of error term of stock i)

137
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

For practical purposes RM is estimated as %age change in a stock market index such as KSE-100 index.
Usually such change is estimated for a one year time period.
Expected return of stock market for the next year , expected R M , can be estimated as:
Expected RM = (expected KSE-100 Index end of the year – KSE-100 index now ) /KSE-100 index now
Expected RM = (50,000 - 46,000)/ 46,000 = 0.087= 8.7%

Sharpe also showed that, with some assumptions, total risk of a stock ( for example VAR MCB) is:

VARMCB = βMCB2 VAR M + VAR eMCB,

Please note that VARM is total risk of stock market, This is the expression by Sharpe for total risk
of a single stock.

Note:

Estimated Ri = αi + βiRM + ei
represents a linear relationship between ROR of stock i and ROR of stock market
Such relationship when drawn on paper appears as a straight line. Graph of historical returns of 6 years
of ‘MCB ’ and stock market is shown below as circles

e 2005
e2003
e1998
RMCB e2001 Characteristic Line for Stock of MCB
αMCB e1999 Beta of MCB = COV MCB, M / VAR M e2000
e2004

RM ( %age change in KSE-100 index)

The circles show actual RORs of MCB stock and stock market for that year; whereas the stright line

captioned as charecteristic line shows estimated RMCB from the linear equation model given
above. Vertical distances shown by arrows are called error terms for those years; such as e 2001 , e 2004.

138
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

These vertical distances between actaual ROR of MCB stock and estimated ROR of MCB stock for each
year from the linear model (straight line) are estimation errors present in this linear model; and each
year’s error of estimation is represented as “e” of that year.

Regression theory says that for a large data set the errors above the line (calculated as vertical distance
between actual circle and the straight line of each year) are + ive. And errors below the line such
distances are negative , and sum of positive deviations or distances would cancel -ive distances or
deviations, and therefore average error or expected value of error , or expected distanc or deviation from
the line is zero . But error term does have a variance called VARe i.

Simplification of Total Risk of Portfolio by Sharpe


With these assumptions, Sharpe simplified the Rp and VARp formulae as follows.
We know that according to Markowitz, expected
expected Rp = ∑XiRi = X1R1 + X2R2 + …..+XnRn
But according to Sharpe for each stock expected rate of return is Ri = α i + βiRM + ei
So inserting this expression of Ri in the Rp formula, we get

Rp = ∑Xi [α i + βiRM + ei] ; which can be written as:


Rp = ∑Xi αi + {∑Xi βi} RM + ∑Xi ei
Rp = αp + ( βp * RM ) + ep.
Note this is also a linear relationship. Where
αp = ∑Xi α i = X1 α 1 + X2 α 2 + ………………. + Xn α n . It is intercept of straight line with y-axis; and
intercept of portfolio, αp ,is weighted average of intercepts of the stocks included in that portfolio.
βp = ∑Xi βi = X1β1 + X2β2 + ………………….. + Xnβn. It is slope of straight line; and it is weighted
average of the betas (slopes) of the stocks included in that portfolio. This expression of beta of portfolio
is saying that beta of portfolio is weighted average of betas of stocks included in that portfolio
ep = ∑Xi ei = X1e1 + X2e2 + ……………….. + Xn en . It is error term of portfolio whose expected value
(mean ep ) is zero by definition because error terms of stock 1 to stock n have expected value of zero
therefore their weighted average error terms of all the stocks is also zero; but error term of portfolio , ep

does have a variance called variance of the error term of portfolio and denoted as:
VAR ep.
139
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

The ep , the error term of portfolio, is weighted average of the error terms of stocks included in that
portfolio. As expected value of error term of each stock was zero, therefore their weighted average is also

zero, so ep is expected to be zero for a portfolio, and therefore return of a portfolio ,


Rp, is estimated as shown below without mentioning error term of portfolio:
Estimated Rp = αp + ( β p * RM )

Sharpe also proved that covariance of returns of any 2 stocks (COV i,j ) = βiβjVARM ; so if you
know betas of 2 stocks and VAR M of stock market returns (total risk of market) then you can calculate
COV between returns of those 2 stocks.
As you know that Markowitz expression for total risk of portfolio is :
VARp = ∑Xi2 VARi + ∑∑ Xi Xj COV i,j (i≠j)
And Sharpe has given total risk of a stock as : VAR = β i i
2
VAR M + VAR ei

Putting this value of total risk of a stock in total risk of portfolio formula of Markowitz, we get:

VARp = ∑Xi2 [βi2 VARM + VARei] + ∑∑ Xi Xj COV i, j (i≠j).


Since Sharpe also proved that COVi,j = βiβjVARM , therefore putting this expression of COV i,j in VARp
formula we get :

VARp = ∑Xi2 [βi2 VARM + VARei] + ∑∑ Xi Xj βiβj VARM . (i≠j).


Further opening it, on RHS of equation the first term becomes

[ (∑Xi2 βi2 )VARM + ∑Xi2 VARei ] ,


and it can also be written as: ∑XiXi βiβi VARM + ∑Xi2VARei
if the condition stock’ i’ is not stock ’ j’ is removed , then the last term on the RHS , that is,
∑∑ Xi Xj βiβj VARM can also be written as : ∑XiXi βiβi VARM
which is exactly the same expressions as the first term on the RHS. This leads to interesting result:

VARp = ∑XiXi βiβi VARM + ∑Xi2 VARei + ∑XiXi βiβi VARM ; and this can be
written with double summation sign because the same term is appearing twice, as

VARp = ∑ ∑ XiXi βiβi VARM + ∑Xi2 VARei. Note the first term on RHS with double
summation is a matrics and in each box you have XiXi βiβi VARM and this can be written as
VARp = [(∑Xi βi ) (∑Xi βi )]VARM + ∑Xi2 VARei

140
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

since ∑ Xi βi is βp therefore; and ∑Xi2 VARei is called variance of error term of poirtfolio and is denoted
as VARep, therefore you can write as

VARp = ( βp * βp) VARM + VARep


VARp = βp2 VARM + VARep
This is the simplified Formula of total risk of portfolio by Sharpe
Where: 1 to n are stocks such as UBL, MCB, Engro, etc
βp = ∑ Xi βi = X1β1 + X2β2 + ………………….. + Xnβn
VAR (ep) = ∑ Xi2 VAR (ei) = X12 VAR e1 + X22 VAR e2 + .....+ Xn2 VAR en. (1 to n are stocks)
So in conclussion:
According to Markowitz Total Risk of portfolio is:

VARp = ∑ Xi2 VARi + ∑∑ Xi Xj COVi.j (i is not j)


According to Sharpe Total Risk of portfolio is:

VARp = βp2 VARM + VARep

Concept of Beta of a Stock, βi


The expression beta has appeared frequently in previous discussion about simplification proposed by
Sharpe in calculation of total risk of portfolio, VAR P , Beta appears both in return and risk formulas of
Sharpe as shown below:

Beta is part of the following formulas proposed by Sharp


1. Estimated Ri = ai + Bi *RM. estimated ROR of single stock
2. Estimated Rp = ap + B p*RM. estimated ROR of portfolio of stocks
3. VAR i = Bi2 VARM + VAR ei total risk of single stock
4. VAR p = Bp2 VARM + VAR e p. total risk of portfolio of stocks

therefore it is important to have clear understanding about its various meanings.

141
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Mathematically beta of a stock is a ratio: Beta of stock MCB is calculated as:

βMCB = COV MCB,M / VARM = 200 /100 = 2 ; and beta of stock j is βj = COV j,M / VARM . Please note that
any stock i with Market, COV i,M , can be negative therefore beta of a stock i can be negative, though

such stocks are rare in actual life. Countercyclical stocks may have negative correlation
( or covariance) of their reteurns with the returns of stock market and therefore can
have negative beta.
If beta is a ratio of covariance of an asset’s returns with market returns divided by variance of market
returns then it follows that if you take the whole stock market as a portfolio then beta of stock market, βM
= COV M,M / VARM
But we know COV M,M = VARM . Because COV of something with itself is VAR of that thing.

βM = COV M,M / VARM =VARM / VARM = 1


This is a proof that beta of stock market is always ONE for any stock market. Beta of
Pakistani market is 1, beta of Indian market is 1, beta of US market is 1 according to the Sharpe’s
formulation.

Beta of 1 is considered average beta because beta of market is weighted average of betas of all the
stocks in that market; βM = X1B1 + X2B2 + …. +XnBn ; whereas 1 to n are different stocks in the market.

Geometrically beta of any stock (βi ) is the slope of straight line shown above where R M is on the
horizontal axis and RMCB is on the vertical axis, this line is termed characteristic line for MCB stock ;
and the straight line is drawn by first calculating alpha and beta , the 2 constants of straight line
equation) for MCB stock using actually realized RORs of the past years for MCB stock and RORs for the
stock market. Then for each past year ROR of MCB is estimated by inserting actual RM of that year and
alpha and beta in the equation given below

Ri = αi + BiRM
Suppose you had ROR data from 2015 to 2020, and you have estimated αMCB + BMCB using
regression. Then you estimate from this model ROR of MCB for each of those years. Then you use
actual RM of 2015 year and estimated R of MCB for 2015 year to place a point on graph paper , then
another point on graph where actual RM actual of 2016 and estimated R MCB is used, and so on until 2020

142
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

you place a dot on graph paper for each year , then you join those dots and you get the straight line shown
above as characteristic line of MCB. Beta MCB is slope of this line.
You know that slope of any straight line is: rise / run; in this case run is change in variable on the
horizontal axis, that is Rm; and rise is change in variable on vertical axis, that is R MCB. So slope of the
straight line is change in RMCB for a unit change in RM , and therefore beta of a stock tells us that 1 % age
point change in RM causes what % age point change in ROR of that stock.
For example if RM changes from 14% to 16%,from 2015 to 2016, and commensurate change in estimated
RMCB is from 20% to 24% then on the graph shown above, then
The rise is 24 -20 = 4%;
and run is 16 -14 = 2%;
and slope of straight line is : rise / run = 4% / 2% = 2, it is beta of MCB, beta MCB has no units of
measurement, beta is just a number, it is not % age.
and you would conclude that beta of that stock is 2.

Please note that % signs cancel each other and the answer is not 2% but just 2. Alpha of linear model
(αMCB ) is in percentages but beta is just a number, it is not a percentage.

Conceptually beta of any stock (βi ) is the sensitivity of returns of a stock to returns of stock market.
For example if beta of ICI stock is 2, it means if R m goes up by 1 %age point this year compared to the
last year then, expected ROR of ICI would go up 2 % age points this year compared to its ROR last year.

For example

suppose alpha of ICI , αICI , is 1% and beta of ICI is 2; estimated through regression, and in 2015
actual ROR of stock market was 10%, then you would estimate that expected ROR of ICI stock for 2015
from Sharpe linear model as:
Estimated RICI = αICI + ( βICI * RM ) 2015

= 1 % + 2* 10%
= 21%
Now suppose next year , 2016, actual RM was 11% (that is 1 % age point increase in R M compared to
last year’s RM), so you would estimate ICI stock return for 2016 as:
Estimated RICI = αICI + βICIRM 2016

= 1% + 2* 11%

143
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

= 23%
So we saw that 1% age point increase in R M from 10% to 11% has caused 2% age point increase in
estimated RICI from 21% to 23%. It Is so because ICI beta is 2. So beta shows sensitivity (response) of
ICI estimated ROR to one unit change in actual RM
As another example:
If αUBL was 3% and beta of UBL stock was 0.8, and actual R M was 10% in 2015, then 2015 estimated R UBL
from linear model proposed by Sharpe would have been:
Estimated RUBL = αUBL + βUBLRM for 2015
= 3 + (0.8*10)
= 11%
and next year , 2016, actual RM was 11% , then estimated RUBL for 2016 according to Sharpe linear
model would be
estimated RUBL = αUBL + βUBLRM
= 3 + (0.8*11)
= 11.8%
Again you saw that 1% age point rise in R M from 10% to 11% would have caused 0.8% age point rise in
returns of UBL stock from 11% to 11.8%, and this was so because beta of UBL stock is 0.8.
The bigger the beta of a company’s stock, more is sensitivity of its estimated stock returns to changes in
stock market returns. That means if beta of a stock is more than one then a slight change in stock market
ROR would cause a big change in ROR of such a stock. But if beta of a stock is less than one, then one
percentage point change in RM would cause a less than 1%age point change in ROR of that stock. In other
words, such a stock’s ROR would be less sensitive to changes in market ROR.
Please note that in this example actual ROR of UBL for past years would be found as capital gains yield
plus dividends yield. And then error term of UBL , say for 2015, would be calculated as:

2015 e UBL = Actual R UBL – estimated R UBL from linear model.


And this error term may come out negative or positive for that year. For example actual ROR of UBL
stock in 2015 was 13% and you estimated above from linear model 11% then e UBL for 2015 was : 13 % –
11% = 2%. And it is shown above on the graph as vertical distance between the straight line and a circle
for 2015

In the context of portfolio theory, Sharpe designates beta of a stock as relevant risk of that stock. It
is a justified name for the beta of a stock because according to Sharpe total risk of any stock is :
VARi = βi2 VAR M + VAR ei

144
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

In this expression VARM is total risk of stock market, and it is a given macro-economic fact and
therefore it is same for all stocks, for all portfolios, and for all investors, in that country. if VAR M is
40% then it is 40 for all the investors, all the stocks, all the portfolios of stocks as long as these are made
2

in Pakistan. Variance of error term of any stock (VAR e i ) is termed as company specific risk of a stock;
and it is assumed as diversifiable when a stock is mixed with other stocks in the form of a portfolio. It is
so because the impact of company specific good event in one company may be cancelled out by company
specific bad event in another company when both stocks are in a portfolio; thus overall impact of
company specific events on the total risk of portfolio may actually be very small; that is, VARe p is likely
to be small in a well diversified portfolio. Later on you shall see that it is about 40 stocks that make your
portfolio almost well diversified by making VAR e p very close to zero in such a portfolio, though it would
not be zero in most portfolios, except some portfolios that are made with certain conditions that would be
explained in the coming lectures..

Therefore only thing relevant for making decision about total risk of a portfolio is
betas of the stocks included in the portfolio. If high beta stocks are chosen then
resulting portfolio beta would be high and consequently total risk of portfolio
would be high; and if low beta stocks are chosen, the resulting portfolio beta would
be low and consequently total risk of portfolio would be low. But in any case
VARM (total risk of the stock market, also called Market Risk) would be there as
part of total risk of all portfolios no matter who is building the portfolio as long as
it is being built in Pak market. The VAR e p would be so small that it can be
ignored by the time a portfolio manager has included about 40 stocks in her
portfolio. This fact would be made clear with an exercise using hypothetical data.
So VAR of portfolio would depend on VAR of M and beta of portfolio. VAR of M
is not manageable and every investor has to swallow it as part of her portfolio’s
total risk; but beta of portfolio can be managed by portfolio manager/investor by
deliberately including low beta or high beta stocks in her portfolio; and resultantly
building low total risk or high total risk portfolio.

145
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Therefore Sharpe argued that in this decision making context, beta of a stock is the relevant risk of that
stock as it has impact on the total risk of portfolio, while other expressions appearing in below stated total
risk formula of a portfolio given by Sharpe
VAR p = Bp2 VARM + VAR e p
are not relevant for decision making as they cannot be managed by portfolio manager/investor: they are
given facts such as VAR of M. Because with VAR M you have to live as long as you are operating in
Pakistani market and you cannot get rid of market risk of Pakistani stock market, while impact of
variances of error terms of stocks included in your portfolio that results in VAR e p can be made to
approach zero by building a portfolio of about 40 stocks.

Example of simplification of calculation of VARp due to


Sharpe’s contribution:
According to Markowitz, in a 4- stock portfolio, number of covariance is n(n - 1)= 4(4 -1) = 12 and
number of variances are 4. Total number of inputs n +n (n - 1) = 4 + 4(4 - 1)= 16 or square of n , ( n 2

= 42 = 16) . Since COV i,j is same as COV j,I therefore number of unique COVs =n(n - 1)/2 = 4(4 -1) /2
= (16 – 4) /2 = 6. Thus according to Markowitz classical portfolio theory, total number of unique
estimated inputs needed for estimating total risk of portfolio ( VAR p ) = n + n (n - 1) /2. In case of 4-stock
portfolios total number of inputs needed to estimate VAR p = 4 + (4(4 – 1)/2 = 4 + 6 = 10

According to Sharpe’s formula for calculating total risk of portfolio (variance of portfolio)
VAR p = Bp2 VARM + VAR e p (sharpe formula for total risk of portfolio)

, you need fewer estimated inputs. In a 4-stock portfolio you need 4 estimates of betas of 4 stocks (B i ) ,
4 estimates of either VAR ei , or 4 estimates of VAR i , and one estimate of VAR M , therefore total
number of input estimates needed for VAR p calculation = 4 + 4 + 1 = 9. You can generalize this for a
portfolio of n stocks as 2n +1 estimated inputs needed to estimate total risk of a portfolio.
For 100 Stock portfolio, number of input estimates needed for VAR p ( Total risk ) calculations
Markowitz
VARp = Sum Xi2VAR i + double Summation X i X j COV I,j ( I can not be j)
‘n’ is 100, so 100 VAR of 100 stock.
n(n - 1) / 2 = 100 (100 -1) / 2 = 9,900/2 = 4,950 unique COVs between RORS of pairs of stocks
Total number of estimated inputs needed = n + n(n - 1) / 2
= 100 + 4,950 = 5,050

146
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Sharpe Formula: VARp = βp2 VAR M + VAR ep

100 VARei for 100 stocks to calculate VAR e p


100 betas ( Bi ) for 100 stocks to calculate Bp
1 VAR M . It is VAR of Stock Market Returns and quantifies total risk of a stock market
Total estimates needed = n +n + 1 or 2n +1
= 100 + 100 + 1 or 200 +1
=201
Please note in lines above it is stated that you need either 100 VAR ei of 100 stocks, but instead of that if
you have estimated 100 VAR i of 100 stock then still you can work out total risk of portfolio. It is
shown below for MCB stock that you can calculate VAR of error term of MCB if you have calculated
VAR of MCB stock and beta of MCB and VAR of Pak market. For example you estimated VAR MCB of
a stock as 64, and you have estimated beta (B MCB ) as 0.9, and have also estimated total risk of Pakistani
market ,VAR M as 20, then you can estimate VAR e MCB of that stock as shown below

VAR MCB = B2 MCB VAR M + VAR e MCB


64 = 0.92 * 20 + VAR e MCB
64 = 16.2 + VAR e MCB
64 - 16.2 = VAR e MCB
47.8 = VAR eMCB
Similarly you can estimate VAR of error terms of 100 stocks. Then use those to calculate the term VAR e
P . And then VAR p
Computational simplification introduced by Sharpe is significant in terms of requirement of estimated
inputs to calculate total risk of portfolio ( VAR p). You just saw that in case of 100 stock portfolios,
5,050 inputs in the form of variance and covariance of stock returns for Markowitz formula are needed,
but you need to estimate only 201 inputs for Sharpe’s formula to calculate total risk of the same 100 stock
portfolio.

Analytical Look at Sharpe’s Total Risk of Portfolio


Let us understand the analytical insight provided by Sharp’s formulation of total risk of portfolio.
According to Sharpe, total risk of a portfolio is:
VARp = B2p VAR M + VAR ep
VARp is total risk of a portfolio. B2p VAR M is non diversifiable risk of a portfolio; and it is
composed of 2 components: relevant risk of portfolio (B p) and market risk of that country’s stock
market (VAR M). In certain text books B2p VAR M is termed as systematic risk of a portfolio because

147
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

risk of stock market (VAR M) is part of this expression and stock market represents the system, or
country, or economy, as a whole. Please note some text books wrongly call this term B2p VAR M as
market risk; market risk is only VARM part of this term; it is better to call the complete term B p VAR 2

M as systematic risk of portfolio or non diversifiable risk of portfolio.

The term VAR ep is called portfolio’s company related risk, or idiosyncratic risk, or non systematic
risk, or a better wording is diversifiable risk. This component of total risk of portfolio is due to
company specific events; as these events are not necessarily same in all the companies so their impact my
cancel out each other, and in a large portfolio this portion of total risk is likely to be so small that it is
negligible; and such portfolios may be called well diversified portfolios; though, strictly speaking those
portfolios may not be called fully diversified portfolios because a fully diversified portfolio is that
portfolio whose VAR ep is zero. AND IN FUTURE LECTURES WE SHALL LEARN TO BUILD
FULLY DIVERSIFIED PORTFOLIOS
It is helpfull for your understanding to visualise total risk, as per Sharpe, in the manner given below:

VARp = B2p VAR M + VAR ep

Total Risk = systematic risk + company specific risk;


and this risk is called by different names such as unique risk, un systematic, idiosyncratic risk.
Total Risk = non- diversifiable risk + diversifiable Risk
Total Risk = (relevant risk squared * market risk) + diversifiable risk

Please notice you have 5 different measures of risks related to a portfolio in the
above formula, namely:
VAR P = B2 P VAR M + VAR e P

1. total risk of portfolio, VAR p

2. diversifiable risk of portfolio, VAR ep = sum (Xi2 VAR e i)

3. non-diversifiable risk, Bp2 * VAR M


4. relevant risk CALLED BETA of portfolio, Bp

148
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

5. market risk, VAR M.


You should be able to calculate those 5 risks for any portfolio. YOU ARE REQUIRED TO
CALCULATE THESE 5 RISKS FOR EACH OF YOUR 4 PORTFOLIOS IN YOUR PROJECT.

Please note all these 5 risks are also applicable not only to portfolios but also to a single stock according
to Sharpe. You can speak about
1. MCB stock’s non-diversifiable risk (BMCB 2 * VAR M)
2. and diversifiable risk of MCB stock (VAR e MCB) ,
3. total risk of MCB ( VAR MCB) and
4. relevant risk of MCB ( B MCB).
5. Market Risk, VAR M
Also note that for all stocks and all portfolios made in Pakistan , the market risk (VAR M) is same, and no
one can get rid off it. If Pak market’s total risk , VAR M , it is 25, then in total risk (VAR) calculation of
all stock it would be 25; and also for calculation of total risk of all portfolios (VAR P) it would be 25.

How does diversification work in Sharpe’s Model:


The folowing expression of total risk of portfolio was derived by Sharpe
VARp = B2p VAR m + VAR ep

The term VAR ep (variance of error term of portfolio) shrinks and can approach to zero as number of
stocks (N) in a portfolio are increased; and therefore this component of total risk is termed by Sharpe as
diversifiable risk of portfolio. Next you shall see with the help of data that by increasing the number of
stocks in a portfolio the diversification effect comes into play and reduces total risk of portfolio
according to the Sharpe’s formulation of total risk of portfolio.

In an equally weighted portfolio of N stocks, same amount is invested in each stock and therefore weight
of each stock in the portfolio is same:
X1 = X2 = ……………. Xn = 1/N ; here N is number of stocks used to build a portfolio.
For example in an equally weighted portfolio of 4 stocks, %age of your OE invested in each stock is
1/N = 1 / 4; or 0.25. so X1 = X2 =X3 = X4 = 1 / 4
We know that Sharpe proved that:

VARp = B2p VAR M + VAR ep

149
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

but VAR e P is ∑ X2i VAR ei, so


VARp = B2p VAR M + ∑ X2i VAR ei.
Now insert 1/N in place of Xi as weight of each stock in an equally weighted portfolio, so the X i is
replaced by 1/N, and you have:
VARp =B2p VAR M + ∑ (1/N)2 VAR ei

VARp =B2p VAR M + ∑ (1/N) (1/N) VAR ei

VARp = B2p VAR M + 1/N [ ∑ VARei / N]


As ∑VARei /N = sum of variances of error terms of all stocks in the portfolio divided by number of error
terms (or number of stocks), so it is average variance of error terms denoted as Avg VARe i

VARp = B2p VAR M + (1/ N * Avg VAR ei)


Please remember this equation is mathematically valid ONLY for the total risk of an equally
weighted portfolio.
If number of stocks in a portfolio is very large, that is, if N approaches infinity, then 1/N approaches zero
and the term {1/N * Avg VAR ei} also approaches 0, so this portion of total risk of portfolio can be made
to approach zero just by adding more stocks in the portfolio. Therefore if a portfolio has many stocks, its
total risk is:
VAR p = B2p VAR M + almost zero, [when N is approaching ∞]

Question
Why B2 p VAR M cannot be made to approach zero by adding more stocks in
the portfolio ?
VARM is always present as positive number; and it is same for all portfolios regardless of which portfolio
manager made the portfolio and included fewer or more stocks; as long as portfolios are being made in
that particular market such as Pakistan. But the portion Bp2 can be managed by a portfolio manager as
she has a choice to include low beta stocks to build a low beta portfolio; or include high beta stocks to
build a high beta portfolio. Therefore only risk that is relevant for portfolio managers’ decision making
about building a low or high risk portfolio is beta of stocks included in the portfolio, hence beta of stock
is called relevant risk of that stock, and beta of portfolio is called relevant risk of portfolio.
Relevant in this context refers to relevant for making decision about total risk of portfolio.
As Bi is part of non- diversifiable risk of a stock, then if you included stocks with high betas then
resulting portfolio’s beta (Bp) would also be high , and that ultimately would result in high non
diversifiable risk of portfolio (B2p VAR M ), which finally results in higher total risk of portfolio (VAR p).

150
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

In real life, portfolio managers are given a target or bench mark beta by their bosses. Let us see how
diversification works when we build portfolios with different number of stocks but keep the beta at the
given target level given by our boss.

Exercise : Please build equally weighted portfolios. Suppose Avg VAR e i is 100%2 for all stocks in
Pakistan, that is the diversifiable risk on average is 100 in Pakistani stocks. Your target is to keep B p =
1.2 for your portfolios as that is the target for the relevant risk given by your boss, and suppose that total
risk of Pakistani market, VAR M = 25 %2, Please remember that it is estimated as variance of returns of
KSE-100 Index using data of previous years
Please build equally weighted portfolios if number of stocks (N) = 1, 2, 4, 10, 40, 100, 200, 400 and
find total risk of these portfolios, i.e. VAR p.

Non-Diversifiable Risk + Diversifiable Risk = Total Risk


N of portfolio B pVAR M
2
+ 1/N *Avg VARei = VARp
1 stock ( 1.2)2 *25 + 1/1 * 100 = 36 + 100 =136
2 stocks ( 1.2)2 * 25 +1/2 * 100 = 36 + 50 = 86
4 stocks (1.2) * 25
2
+ 1/4 * 100 = 36 + 25 = 61
10 (1.2) * 25
2
+ 1/10 * 100 = 36 + 10 = 46
40 (1.2)2 * 25 + 1/40 * 100 = 36 + 2.5 = 38.5
100 (1.2)2 * 25 + 1/100 * 100 = 36 + 1 = 37
200 (1.2) * 25
2
+ 1/200 *100 = 36 + 0.5 = 36.5
400 (1.2)2 * 25 + 1/400 * 100 = 36 + 0.25 = 36.25
In the table given above, as number of stocks increased in a portfolio, total risk of portfolio decreased;
diversifiable risk (1/N Avg VARei ) decreased from 100 to 0.25, a significant decrease approaching to
zero, it is diversifiable risk of portfolio . Please note that total risk of portfolio (VAR p ) kept on
decreasing as you kept on increasing number of stocks, but most of the decrease was achieved by the time
40 stocks were in the portfolio as total risk declined from 136 to 38.5. Please note that the term (Bp2
VARM ) remained constant at 36, it is not reduced by increasing the number of stocks and therefore it is
non-diversifiable risk of portfolio. Please also note that you had diversified 71% of total risk by the
time you had 40 stocks in the portfolio [(38.5 – 136) /136 = -71%]; and diversifiable risk was down to 2.5
from 100; a 97.5% decline in diversifiable risk

151
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

( 2.5 - 100) / 100 = - 0.975.


Therefore in practice, a portfolio composed of 40 assets is considered well diversified; even 10 assets
are good enough because in our example by the time there were 10 stocks in the portfolio the diversifiable
risk was down to 10 from 100; that is a 90% decline: (10 – 100) /100 = - 0.9; and total risk was down to
46 from 136, that is 66% decline in total risk :(46 -136)/136= -0.66

The Concept Of Average Beta, Building Low Risk Portfolios


In real life portfolio managers are given by their bosses a target beta for their respective portfolios; and
according to this bench mark they decide about including stocks in their portfolio. Let us pay attention to
it:
Bp2 = (∑ Xi Bi)( ∑ Xi Bi), since for equally weighted portfolio Xi = 1/N so we can write
Bp2 = ∑ (1/N) Bi * ∑(1 / N) Bi. And it can be written as [∑Bi / N] * [∑Bi / N]. Note: ∑Bi /N is sum of
betas divided by number of betas (or number of stocks) , and that is called average beta. So:
Bp2 = average beta * average beta.
Note also that N of a portfolio is largest when maximum number of stocks are in your portfolio; it
happens when all the stocks present in the market are included in your portfolio, for example all stocks
listed at PSX, Pakistan Stock Exchange. But that means beta of such a portfolio is same as beta of stock
market, and you have already proven that beta of any stock market is always one, so when N is very large
then average beta should be same as market beta, that means beta of such a portfolio would be one.
Therefore:
∑Bi / N * ∑Bi / N = Avg Bi * Avg Bi . Since Avg Beta is 1 always.
Bp2 = Avg B * Avg B = 1 x 1 = 1
So, If N is very large , and 1/N approaches zero, then B 2p approaches 1
, but still the term Bp2 VARM cannot be made to approach to zero because VAR M is still there as a
positive number; and it is the total risk of a stock market which is a macro-economic fact and cannot be
reduced by any portfolio manager as long as she has built her portfolio in a particular market such as
Pakistan.
But the term B2p VARM can be reduced to (1)2 VARM , as you saw above that beta can be 1 if all stocks
are included
So in a large portfolio with many stocks in it:

VARp = Bp2 VARM + VARep

152
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

VARp = 12* VARM + Almost zero


and that means
VARp = VARM

that means, total risk of such a portfolio is equal to market risk (VAR M)
Note: Since VARM is same for all the portfolio managers in Pakistan, say 25, they all have to accept it as
part of the total risk of their portfolio. Therefore it is B p2 portion of total risk of a portfolio that a
portfolio manager can attempt to manage by selecting low or high beta stocks for her portfolio, therefore
Bi of a stock is the relevant risk of a stock from portfolio management view point because it influences
Bp.

An interesting question arises :

Is it possible for a portfolio manager to

build a portfolio whose total risk is less than the


market risk? In other words, can a portfolio manager build a portfolio
whose VARp is less than VARM?
The answer is yes.
A portfolio manager can build a portfolio whose total risk is much lower than the total risk of stock
market by selecting low beta stocks and thereby having portfolio beta less than the market beta of 1.

For Example: a portfolio manager can build a portfolio whose beta is 0.5 using 40 stocks. suppose
Market risk of Pakistani stock market (VAR M) is 64%2. The resulting total risk of such a portfolio
would be:
Total risk of portfolio = VARp = Bp2 VARM + VARep
Total risk of portfolio = VARp = 0.52 VARM + almost zero VAR ep
(because of 40 stocks)
Total risk of portfolio = 0.52 64 + almost zero VAR ep
Total risk of portfolio = 16 + almost zero VAR ep
so such a portfolio’s total risk would be one-fourth of the total risk of the stock market that was 64.

153
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

You just saw with the example of an equally weighted portfolio that VARe p can be made to approach
zero by having a large number of stocks in your portfolio, that is, large N. Therefore only manageable
or controllable item whose contribution to the total risk of portfolio you can attempt to manage is the B p,
and that depends on betas of individual stocks included in the portfolio by the portfolio manager.

Since Bp = ∑ Xi Bi = X1 B1 + X2 B2 + …………. + Xn Bn (1
to n are different stocks)
Therefore, Beta of each security is the relevant contributor to the total risk of portfolio. In finance
literature relevant risk of a stock is beta of that stock. So each stock’s contribution to total risk of
portfolio is beta of that stock. To build low risk portfolio, you as portfolio manager, should include those
stocks in your portfolio that have a low beta.
B m = COV m,m/ VAR m = VAR m /VAR m =1
Assuming stock market as a whole is a portfolio then applying Sharpe formula of total risk on stock
market we get the following analytical results
VAR m = B2m VAR m + VAR e m
VAR m = 12 VAR m + almost zero (as market has more than 4 stocks)
VAR m = VAR m

Aggressive (high risk) and Defensive (low risk) Stocks and Portfolios
In the context of Sharpe’s formulation of total risk of portfolio, when a stock is called as low risk or high

risk , we are referring to their beta. As stock market beta is considered average beta and it is
ALWAYS ONE, therefore with reference to market beta a stock is categorized as high risk or aggressive
if its beta is more than one; and low risk or defensive if its beta is less than one; and the same is true for a
portfolio.
Low risk stock or low risk portfolio: Beta < 1 (defensive stock or portfolio)
High risk stock or high risk portfolio: Beta > 1 (aggressive stock or portfolio)
Average risk stock or portfolio: Beta = 1, and it is also the beta of stock market

Following is the summary of what you have learnt up till now about risk and
return of stand-alone stocks as well as portfolio of stocks:

154
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Stand Alone Stock


Markowitz
Expected ROR of single stock (Ri) = Expected dividend yield + Expected capital gains yield
Total Risk of single stock (VARMCB ) = ∑(RMCBt - Avg RMCB )2 /n . ‘ t’ varies from year 1 to year n.
Sharpe
Expected ROR of single stock, Ri = αi + Bi RM . i refers to any stock such as MCB
Total Risk of single stock (VARi ) = B i VARM 2
+ VARei
= Non-Diversifale + Diversifiable risk

Portfolio of Stocks
Markowitz
Expected Rp = ∑Xi Ri = X1R1 + X2R2 +…….+XnRn . 1 to n are various stocks
VARp = ∑Xi VARi + ∑∑Xi Xj COVi,j ( i is not j)
2

Sharpe
Expected Rp = αp + (Bp RM)
VARp =Bp2 VARM + VARep
while αp = ∑Xiαi , and Bp = ∑XiBi
while VARep = ∑Xi2 VAR ei
Coefficient of determination of a portfolio
R2 = Systematic Risk or non-diversifiable risk / Total Risk
= Bi2 VARM / VAR P .
If total risk is taken to be 100% or 1 then R 2 tells what %age of total risk is non diversifiable; the
remaining %age of total risk is diversifiable risk (1 – R 2). R2 is called Co-efficient of determination, and
it shows %age of changes or variations in returns of a stock or a portfolio due to variations in market
return. If R2 of a stock or a portfolio is 0.3, it means 30% of changes or variations in its returns are due
variations in RM.
Please note you can find correlation (R) between portfolio returns and market returns, that is (CORR P,

M ) by taking under root of R , that is √ R , and the same applies to correlation of a stocks return with the
2 2

market returns,
CORR UBL , M = √R2

155
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Example
If for a portfolio you know its Beta is 1.2; risk of stock market (VAR M) is 25; and portfolio’s diversifiable
risk is 10, then total risk of this portfolio is:
Total risk = non diversifiable + diversifiable
VAR p = B2 p VAR M + VARep
= (1.2)2 25 + 10
= 1.44 x 25 + 10
= 36 + 10
=46

Dividing the whole equation by Total Risk, i.e. VAR p


VAR p / VARp = B2p VARM / VARp + VARep / VARp
46/46 = 36/46 + 10/46
1 = 0.78 + 0.22
1 = R2 + (1 - R2)

78% of total risk of this portfolio is systematic, i.e. non diversifiable


22% of total risk is diversifiable. In this case R 2 is 78% , it means 78% variations in dependent variable
(that is Rp) are explained by variations in the independent variable which is return of market, R M,
according to Sharpe’s linear model
Expected Rp = αp + Bp RM

Note : correlation of returns of this portfolio with the stock market return is :
CORRP,M = √ R2 = √0.78 = 0.88
Correlation of 0.88 means a strong correlation between returns of this portfolio and returns of stock
market.

Please note if past historical data is available then that beta is a ratio:

156
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Beta of portfolio = COV P,M /VAR M; and VAR M is = SDM* SDM; You also know that covariance is:
COV P,M = CORR P,M * SDP * SDm.
Therefore Beta P = (CORR P,M * SDP * SD M) / (SDM * SDM). And that simplifies to:

Beta P = (CORR P,M *SDP ) / SD M .


As you know that SD cannot be negative, but correlation can be negative between returns of a stock (or a
portfolio) and return of stock market, therefore beta of a stock (or a portfolio) can be negative, though in real life
building such a portfolio is desirable but challenging.

Learning to Calculate total risk, covariance,


correlation, alpha, and beta, diversifiable risk (VAR e
PSO ),
non-diversifiable risk (B2MCB VARM), co-efficient of
determination R2 = Non diversifiable risk/total risk,
We have learned classical Markowitz formulae for portfolio risk and return and then we learned
simplification introduced by Sharpe. Sharpe formulations had terms such as α i (alpha of stocks), βi (beta
of stocks), and ei (error term of stocks), non-diversifiable risk of a stock ( β i2 VARM ), diversifiable risk of
a stock (VAR ei ), co-efficient of determination of a stock ( R 2 ), and correlation of a stock returns with
stock market returns (Corr i,M). Similar terms for portfolio were also derived by Sharpe including
expressions for αp (alpha of portfolio), βp (beta of portfolio), VAR ep (variance of error term of portfolio)
which he termed as diversifiable risk of portfolio, non-diversifiable risk of a portfolio ( β p2 VARM ), co-
efficient of determination of a portfolio, R 2, tells what percentage of total risk is non diversifiable
because of the presence in the portfolio certain systematic factors such as VAR M, and correlation of a
portfolio returns with the returns of the stock market (Corr P,M) .

We want to learn to calculate these items for a stock , first by hand calculations, and then by using your
calculator. Please note that in the previous class you learned to calculate the α p, βp, VAR ep.

157
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Exercise: Suppose your staff of security analysts have the following 5-years historical data of actual
rate of returns of PSO stock and stock market. Please note stock market returns are %age change in KSE-
100 index each year; and actual returns of PSO is realized capital gains yield plus realized dividend yield
in each of the past year.
Year R PSO R M ------------------------- (KSE 100 end - KSE 100 beg) / KSE 100 Beg
2011 40% 24%
2012 -11 -7
2013 -5 10
2014 3 18
2015 24 32
2016 -9 -5
Avg 7% 12%
Please note that for VAR of returns calculations we are using in the denominator ‘n’ (number of
observation, that is 6 for 6 years data of RORs) instead of n - 1 because data set represents the actual
returns and is not sample of returns; in each of these 6 years these were the actually realized RORs.
Similarly for COV calculations instead of ‘n – 2’ in the denominator we are going to use ‘n’ in the
denominator.
The following questions you can answer with this data set:
Find:
1. Total risk of PSO stock
2. Total risk of stock market VAR M
3. Covariance of returns of PSO with returns of stock market
4. Relevant risk (Beta) of PSO stock
5. Correlation of returns of PSO with returns of stock market
6. Coefficient of determination (R2) of PSO returns with stock market returns
7. Alpha of PSO stock
8. Estimated returns of PSO stock for each of 6 years from Sharpe’s linear model: estimated
R of PSO = alpha PSO + ( Beta PSO * RM)

158
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

9. Error term of PSO returns for each of 6 years


10. Diversifiable risk of PSO stock ( VAR e PSO , Variance of error term of PSO stock)
11. Non-diversifiable risk of PSO stock :Beta of PSO squared* VAR of market
12. Show total risk of PSO stock = non-diversifiable risk + diversifiable risk

Total risk of PSO stock


VAR PSO= [(40 - 7) 2 + (-11 - 7)2 + ( -5 - 7)2 + (3 - 7)2 + (24 - 7)2 + (-9 - 7)2 ] /6.
=(1089 + 324 + 144 + 16 + 289 + 256) / 6 = 2118 / 6 = 353%2
SD PSO = √VAR PSO = √353 = 18.788 %

Total risk of stock market


VAR M= { (24 - 12) 2 + (-7 - 12)2 + (10 - 12)2 + (18 - 12)2 + (32 - 12)2 + (-5 - 12)2 }/6
=[(144 + 361 + 4 + 36 + 400 + 289)]/6
= 1234/6
=205%2
SD M = √VAR M = √205 = 14.317%
Co-Variance of returns of PSO and stock market

COVPSO,M =[(40-7)(24-12) + (-11-7)(-7-12) + (-5-7)(10-12) + (3-7)(18-12) + (24-7)(32-12) +

(-9-7)(-5-12)]/6
=[(396 + 342 + 24 + (-24) + 340 + 272)]/6
=1350/6 = 225%2
Beta of PSO stock (relevant risk of PSO)
BPSO = COVPSO , M / VAR M =225/205 =1.09
Please note that PSO beta is slightly higher than one, so this stock is slightly more risky than the overall
stock market, and would be included in aggressive or high risk stocks category.
Correlation of returns of PSO stock and the stock market
COVPSO, M = CORR PSO, M *SDPSO*SDM
225 = CORR PSO , M 18.78*14.31
225/(18.78 * 14.31) = CORR PSO , M

0.83 = CORR PSO , M =R

159
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Please note that returns of PSO stock are strongly correlated with the returns of stock market.

Co-efficient of determination (R2) for PSO


Please note R2 is (Correlation PSO,M)2 . You already found correlation above as 0.83; So R 2
is therefore
(0.83)2 = 0.69. It means 69% of variations in return of PSO are due to variations in market returns so
these are systematic variations in PSO returns, or in other more careful wording, 69% of variations in
PSO returns are explained by variations in market returns. But 1 - R 2 = 1- 0.69= 0.31 or 31% of the
variations in PSO returns are not explained by variations in stock market returns, rather company specific
factors are responsible for this portion of variations in PSO returns in last 6 years. So 31% of variations

in PSO returns are unsystematic, or company specific, or idiosyncratic, or simply


diversifiable. You can also say that 69% of total risk of PSO is non- diversifiable while 31% of
total risk is diversifiable. Because R 2 is a ratio of non- diversifiable risk to total risk. Let us double check
it for PSO stock:
R2 = non – diversifiable risk / total risk
R2 = β2PSoVARM / VARPSO = non diversifiable risk / total risk = R squared
= [(1.09)2 * 205] /350
=243.5/350 = 0.69.
This is same R2 we already found as square of correlation above

alpha is intercept of straight line with y-axis, it is the return of PSO on y-axis when return of market is
zero in x-axis
How to estimate Alpha ( αPSO ) of PSO
Avg RPSO = αPSO + (βPSO * Avg RM )
7 = αPSO + (1.09*12)
7 = αPSO + 13.08
7 - 13.08 = αPSO
-6.08% = αPSO.
Please note the data for average returns of PSO stock and the stock market is coming from the table
above. Also note that alpha of PSO refers to intercept of straight line with Y-axis; in previous lecture we
called this characteristic line of a stock when actual historic R M was on x-axis and the line gave estimated
ROR of PSO based on regression.
Note Alpha of PSO means that in case RM is zero, then Beta PSO * RM is also zero, then

160
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

R PSO = aPSO + BPSO * RM


= - 6.08% + 1.09 * 0%
R PSO = - 6.08 + 0
Return of PSO stock would have -6.08% return in a year when stock market returns are 0%
Estimated Sharpe’s Linear Model For PSO
Sharpe has argued that expected returns of any stock are linear function of stock market returns, and it can
be written as:
estimated Ri = α i + βi RM.
Therefore expected returns (or returns estimated from straight line regression of PSO with stock market
returns) can be written as:
estimated RPSO = α PSo + βPSO RM. Please enter the alpha and beta (constants, or parameters) of PSO that
you have estimated above in this model, you get:
estimated RPSO = - 6.08 + 1.09*RM of that year
When written in this form, it is said that model has been estimated because its parameters, that is alpha
and beta, have been estimated. You can insert any values of independent variable (R m) and can estimate
the corresponding values for the dependent variable (R PSO). Please note that one of the methods of
estimating parameters of this linear regression model is called ordinary least square method (OLS) and
that method was used here, and these values of alpha and beta of PSO are called OLS estimates of the
parameters of this model.
Estimated returns of PSO for the last 6 years from this model using actual returns of market as R M
First we estimate RPSO for each of the 6 years using Sharpe’s Market Model, and actual R m of those years
would be inserted as independent variable to get an estimate of dependent variable , that is estimated R PSO

for these 6 years. Once we have model-estimated returns of PSO, then we can compare estimated returns
with actual returns of PSO and calculate error term for each year; such error terms tell us that our model
was not accurate and therefore estimated returns of PSO given by the model are different from the actual
returns. Each year’s estimated return of PSO stock using Sharpe’s linear model are calculated below:
Please note that data for R M is actual historical return of stock market calculated as percentage change in
KSE-100 index in each of those 6 years, and it is shown above along with the actual returns of PSO in the
same 6 years

α PSO % + βPSO * RM % = estimated RPSO %


For 2011: -6.08 + 1.09*24 = 20.08

161
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

For 12: -6.08 + 1.09*-7 = -13.71


For 13: -6.08 + 1.09*10 = 4.82
For 14: -6.08 + 1.09*18 = 13.54
For 15: -6.08 + 1.09*32 = 28.8
For 16: -6.08 + 1.09*-5 = -11.53

Error term of PSO in each of the 6 year


Since linear model: RPSO = alpha PSO + (beta PSO*RM )
is only a model that means it is simplified representation of reality, therefore it is bound to make mistakes
in doing its job. That is, the returns estimated using the model are not going to be correct returns for that
year. For past years, you know the correct actual returns of PSO stock, and now you have also calculated
using the linear model estimated returns of those years. Therefore you can find each year’s error made by
the model for PSO stock returns, please subtract estimated return of PSO from actual return of PSO for
each year to get error term for each year.
et of PSO = Actual R PSO – Estimated R PSO
e2011 =40 % – 20.08% = 19.92%
e12 =-11 – -13.91 = 2.91%
e13 = -5 – 4.82 = -9.82%
e14 =3 – 13.54 = -10.54%
e15 = 24 – 28.8 = -4.8%
e16 = -9 – -11.53 = 2.53%
Once you have error term of each year, the next step is to find expected value or average of the error
terms (Avg e PSo ). Average of error terms is close to zero as shown below:
Avg e PSo =(19.92 + 2.91 -9.82 -10.54 - 4.8 + 2.53) /6 = 0 .2 /6 = 0.03
Sharpe has by definition taken this as Zero, and 0.03 is very close to zero. Therefore when calculating
the variance of error term of PSO stock, we would use zero as its average return

Variance of error term of PSO ( VAR e PSO , diversifiable risk of PSO stock)

162
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Though expected value of error term of PSO is supposed to be zero yet it has a variance, Sharpe calls it
company specific or idiosyncratic or non- systematic, or diversifiable risk of that stock. VARe PSO = n∑
t=1 ( et – Avg e)2/n. Please note we are using again ‘n’ as denominator instead of n -1 because data set is
not a sample, also note that for average error term we use zero instead of 0.03 because that is the
assumption used by Sharpe in building this model.
VAR e PSO =[(19.92 - 0)2 + (-2.91 - 0)2 + (-9.82- 0)2 + (-10.54 - 0)2 + (-4.8 - 0)2 + (2.53 - 0)2] /6
=642.23/6
=107.03%2
This is diversifiable risk of PSO stock

Double check
Total risk of PSO = non –diversifiable risk + diversifiable risk
VAR PSO = β2 PSOVAR m + VAR e PSO
351 = (1.09)2205 + 107.03
= 243.56 + 107.03
= 350.59.
Please compare it with your answer above where you found already VAR PSO as 353. This difference is
due to rounding.
Please also note that out of total risk of PSO (350.59), non-diversifiable risk ( systematic risk) is 243.56;
and diversifiable ( company specific) risk is 107.03. Sometime researchers want to know what
proportion of total risk is non- diversifiable and what proportion is diversifiable, to do that divide the
whole equation by total risk:
350.59/350.59 = 243.56 / 350.59 + 107.03/350.59
1 = 0.69 + 0.31 . and multiplying by 100 you get
100 % = 69% + 31%
Then you can say that out of total risk of PSO, 69% is non diversifiable risk and 31 % is diversifiable
risk. Please note you got the same 69% as R2 already in above calculations.

Some portfolio managers look for stocks with high proportion of diversifiable risk and low beta, such
stocks they want to include in portfolio because once such stocks are included in a portfolio, diversifiable
risks of different stocks cancel out each other and over all diversifiable risk of portfolio (VAR e P) is

163
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

reduced. It is so because diversifiable risk is due to company specific events, and impact of good event in
one company is cancelled out by bad event in another co. Since in a portfolio, beta of portfolio is the only
manageable portion of total risk, therefore a portfolio manager who wants to build low risk portfolio shall
look for low beta stocks so that resulting beta of portfolio is low and thus ultimately total risk of portfolio
(VAR p) is low. Total risk of single stock (VAR i), according to Sharpe formula, has 4 components:
1.Non-Diversifiable risk of PSO = B2PSO VARM = 243.56%2
2.Market Risk of Pakistan’s stock market = VARM = 205%2
3. Relevant risk of PSO = BPSO = 1.09 (it is not a percentage)
4.Diversifiable risk of PSO = VAR e PSO = 107.03%2
5. Total Risk of PSO stock = non diversifiable risk + diversifiable Risk
353 = 243.56 + 107.03
350 = 350.59
The difference is due to rounding
And now you have learned how to calculate each of the 5 risks in a stock as worked out by Sharpe. Also
note these 5 types of risks are applicable to a portfolio as well and you can calculate non-diversifiable
risk, diversifiable risk, relevant risk and total risk of a portfolio; whereas market risk (VAR m) is a macro
economic fact applicable to all stocks and all portfolios: if it is 40% 2 in Pakistani market then for all
stocks and portfolios in Pakistan it is 40 for the calculation of non-diversifiable risk and beta of all those
stocks and portfolios

Using BA II Plus Financial calculator for estimating Sharpe


Linear regression model for PSO,that is estimating Alpha
and beta, of PSO stock

Rm is always entered as X variable, as it is independent variable in


RPSO = Alpha PSO + (Beta PSO *RM) sharp linear model
X Y
Year RM R PSO
2011 24% 40%
2012 -7 -11
2013 10 -5
2014 18 3
2015 32 24
2016 -5 -9
Avg 12% 7%

164
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Hit 2ND and DATA keys, you see on screen


X01 write 24, hit ENTER, hit ↓, see Y01, write 40 , hit ENTER, ↓ ,
see X02, write 7 hit +/- key , hit ENTER, hit ↓, see Y02, write 11, hit +/- key ENTER, ↓,
see X03, write 10, hit ENTER, hit ↓, see Y03, write 5, hit+/- hit ENTER, ↓,
see X04, write 18 , hit ENTER, ↓, see Y04, write 3 hit ENTER, ↓,
see X05, write 32, hit Enter, ↓, see Y05, write 24 hit Enter, ↓,
see X06, write 5, hit+/- key, Enter, see Y06, write 9 hit +/- hit Enter

data has been entered for 6 years for x and y variables , that is data about RM (X) and PSO (Y)
Hit 2ND and SET keys again and again until you see LIN on the screen. It refers to
Y = A + BX, that is linear regression model of 2 variables. You want that, so you
hit ↓, you see: n=6; it is number of observation in data set
hit ↓, you see X bar = 12, this is mean of X variable, return of Market
hit ↓, you see Sx = 15.71. This is sample SD of X , Market, calculated with n-1 in the denominator.

Hit ↓, you see δX = 14.34, this sigma of X variable, that is population SD of RM, with n in the
denominator, you need that
You hit ↓, you see Y bar = 7, it is mean of Y , returns of PSO stock
You hit ↓, you see SY = 20.58; this is sample SD of Y variable , that is PSO

You hit ↓, you see δY = 18.78, this is sigma of Y variable, that is population SD of PSO returns, with
n in the denominator, you need that
You hit ↓, you see, a = -6.128, this is intercept of straight line with Y-axis, straight line starts from this
value on y-axis, that is PSO returns. it is estimated return of PSO (that is Y variable value), when return
of market ( X variable value ) is zero. Due to rounding it is slightly different than found above by hand -
6.08
You hit ↓ , you see, b = 1.09; this is same as found above by hand, this is slope of straight line , it is also
beta of PSO with respect to RM; which tells you that 1 unit increase in RM, which is X variable or
independent variable, causes 1.09 unit increase in PSO returns that is Y variable or dependent variable. In
simple words when return of market increases by 1%age point the return of PSO increase by 1.09%age
points and vice e versa. Please note this b = 1.09 is the famous beta of PSO stock that can be used in
CAPM model. here as X variable we had used R M, return of stock market, then the resulting ‘b’ is called
beta of PSO.
You hit ↓, you see r = 0.835; this is correlation of Y with X, that is, correlation of PSO returns with
Market returns.

165
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Now you have all the data for calculating covariance between ICI and PSO stock returns as :
covariance, COV PSO, M = correlation PSO, M * SD PSO * SD M
1. covariance = 0.835 * 14.34 * 18.78
2. covariance = 224.86 , by hand it was 225 the difference is due to rounding.

Please note : alpha of PSO, beta of PSO, the correlation of PSO with M, , and SD of RM and SD of
PSO, given by BA II plus are same as found above by hand

Doing the same Calculation of alpha & beta of PSO using Casio FC-100 Calculator
1) Go to STAT, green button
2) Choose A + B X, then E X E, you see two columns X and Y in which you can enter data. Y =
A + BX
Y = R PSO, X = RM , A = Alpha of PSO, B = Beta of PSO
3) Enter in X column RM data for 6 years and enter in Y column RPSO data for 6 years.
NOTE: YOU ALWAYS ENTER RM DATA IN X COL, AS IT IS THE INDEPENDENT
VARIABLE
4) Then hit red AC button
5) SHIFT STAT buttons pushed, you see 7 item menu
6) Choose 7 for regression
7) See 5 item menu, choose 1 for α PSO. Hit EXE you get -6.12% . ( You calculated above -6.08 by
hand)
8) Again hit SHIFT STAT choose 7 for regression , and then choose 2 for B PSO, hit EXE, you get
1.09 (you have also calculated above 1.09 by hand)
9) Again hit SHIFT STAT choose 5 for variances you see 5 item menu
Choose 3 for SD of X ( that is SD M ), EXE, you get 14.34 ( your computed answer above is
14.31)
10) Again hit SHIFT STAT choose 5 for variances. Choose 6 for SD of Y (that is SD PSO ), EXE,
you get 18.78 ( you already computed by hand 18.78)
11) Again hit SHIFT STAT choose 7 for regression. You see 5 item menu, choose 3 for CORR,
EXE, You get 0.835. ( you have computed already 0.83 as CORR PSO,M)
12) COV PSO ,M = Corr PSO , M * SD M * SD PSO
= 0.835 * 14.34 * 18.78
= 224.86
You calculate above by hand 225, the difference is due to the rounding.

166
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Please Note: To get your calculator back in a mode to do financial calculation’s data entry such as cash
flows for NPV and IRR, you need to bring data entry setting to one column. To do that, go to STAT and
choose option 1 (that is 1-VAR), that is one variable ; the data entering screen would show now one
column.

Another Exercise:
We have last five years return data for MCB and stock market. Please note return of stock market is
percentage change in KSE-100 index each year.

Years RMCB Y RM X
2005 10 4
2006 3 2
2007 15 8
2008 9 6
2009 3 0
Avg 40/5=8 20/5=4
VAR MCB= [(10-8)2+(3-8)2+(15-8)2+(9-8)2+(3-8)2 ]/5
=[4+25+49+1+25]/5
=104/5
=20.8%2
SD MCB =√(20.8) =4.56
VARM= [(4-4)2+(2-4)2+(8-4)2+(6-4)2+(0-4)2 ] /5
= [0+4+16+4+16 ]/5
=40/5
=8%2
SD M = √ (8) = 2.82843
COVMCB,M= [(10-8)(4-4)+(3-8)(2-4)+(15-8)(8-4)+(9-8)(6-4)+(3-8)(0-4) ] /5
= [(0) + (10) + (28) + (2) + (20)] /5 = 60 /5
=12% 2

167
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Please note instead of n-1 or n-2 in denomination of VAR and COV, we used ‘n’ in the denominator.
BMCB= COV MCB, M/ VAR M
=12/8
=1.5
BM=COVM,M/ VARM
=VARM/VARM
=8/8
=1 Always

COV MCB, M= CORR MCB, M *SD MCB* SDM


12 = CORR MCB, M *4.56*2.82843
12/(4.56*2.82843)= CORR MCB, M
12/12.89764 = CORR MCB, M
0.93040 = CORR MCB, M
R2MCB, M =(CORR MCB, M)2
=(0.93040)2
=0.86564 coefficient of determination
Or
R2MCB, M = non diversifiable risk / total risk
α MCB ( alpha of MCB)
Avg RMCB = α MCB + βMCB * Avg RM
8 = α MCB + 1.5*4
8 = α MCB + 6
8 – 6 = α MCB
2 = α MCB

Sharpe linear Model for MCB


Estimated RMCB = α MCB + β MCB RM , inserting the estimated parameters alpha and beta you get:
Estimated RMCB = 2 +1.5 RM

Estimating RMCB for years 2005 to 2009 using actual RM for these years and estimated α and β of MCB
α MCB + β MCB* RM = Estimated RMCB
2+ 1.5* RM = Estimated RMCB

168
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

2005 2+ 1.5*4 =8%


2006 2+ 1.5*2 =5%
2007 2+ 1.5*8 =14%
2008 2+ 1.5*6 =11%
2009 2+ 1.5*0 =2%

error term of MCB (e MCB) for each year from 2005 to 2009
et = Actual R MCB – estimated R MCB from linear Model
e2005 = 10 - 8 =2%
e2006 =3 - 5 = -2%
e2007 = 15 - 14 =1%
e2008 =9 - 11 = -2%
e2009 =3 - 2 =1%
Avg e = (2 -2 +1 -2 +1 ) / 5 = 0/5 = 0, same as theorized by Sharpe
Theory says expected value (Avg) of error term should be zero and we saw for MCB it is zero.

VAR e MCB (Diversifiable Risk of MCB stock)


VAR e MCB =∑(ei - Avg)2/n (i =2005 to 2009)
={(e2005- 0) + (e2006 - 0) + (e2007 - 0)2 + (e2008 - 0)2 + (e2009 - 0)2}/5
2 2

= [(2 - 0)2 + (-2 - 0)2 + (1 - 0)2 + (-2 - 0)2 + (1 - 0)2 ]/5


= [(4 + 4 + 1 + 4 + 1)] /5
=14/5 =2.8%2 diversifiable risk of MCB
Double check: Total risk = non –diversifiable risk + diversifiable risk
VAR MCB = β2MCB VAR M + VAR e MCB
= (1.5)2 * 8 + 2.8
= 18 + 2.8
VAR MCB = 20.8%2
And it is same answer as we already found above when VAR MCB was calculated by traditional statistical
method.
Please not that R squared can also be calculated as

169
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

R2= Non diversifiable risk/Total risk


=β2MCB VAR M/VARMCB
=[(1.5)2 *8]/20.8
=18/20.8
=0.86538
Corr MCB, M = under root of R2 = under root 0f 0.86538= 0.93

You noticed that total risk can be bifurcated into non diversifiable risk and diversifiable risk. In our
example of MCB stock
Total Risk = non- diversifiable risk+ diversifiable risk
VAR MCB = β2MCB VAR M + VAR e MCB
20.8 =(1.5) 8 2
+ 2.8
20.8 =18 + 2.8
If we divide the whole equation by total risk we get what proportion of total risk is non diversifiable and
what proportion is diversifiable risk
20.8/20.8 = 18/20.8 + 2.8/20.8
1 =0.86538 + 0.13462
100 % = 86.53 % + 13.47 %
1 = R 2
+ (1 - R2)
Later on you will learn that high proportion of diversifiable risk in total risk of a stock does not make that
stock necessarily less attractive for inclusion in the portfolio; in fact the important issues is beta of stock.
As long as it has low beta its inclusion would result in low beta portfolio; and impact of its high
diversifiable risk would be cancelled out just by having a large number of stocks so that the diversifiable
risk of portfolio approaches to zero just by adding more stocks. Therefore from portfolio management
view point the relevant risk is beta of stocks because that has impact on beta of portfolios, and in turn that
would impact total risk of portfolio; later-on you will learn about a theory which takes into account only
the beta as the risk measure for investment decision making and relates expected returns with that risk
alone while ignoring other component of total risk

Using BA II Plus financial calculator for similar calculations

170
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

You are given 6 year’s return data for MCB share and Stock market (that is KSE -100 index % age
change). Note, Market returns are entered as X, and stock’s returns as Y; meaning stock returns are
dependent variable and market returns are independent variable.
Same Data Set as above
Years X (Return of Market, RM) Y (returns of MCB share)
2005 4% 10
2006 2 3
2007 8 15
2008 6 9
2009 0 3
Avg 20/5=4 40/5=8
Estimating the linear regression model given below means finding Alpha and beta of the model, these
are called intercept and slope of straight line respectively. Use above data set of RORs, do the following.
Y = a+ B*X
RMCB = aMCB + B MCB * RM

Hit 2nd and Hit DATA to get into data entry mode, on the screen you see
X01, write 4hit Enter, Hit ↓ see Y01 , write 10 Hit Enter
Hit ↓, See X02, write 2 hit Enter, Hit ↓ see Y02 , write 3 Hit Enter
Hit ↓, See X03, write 8 hit Enter, Hit ↓ see Y03 , write 15 Hit Enter
Hit ↓, See X04, write 6 hit Enter, Hit ↓ see Y04 , write 9 Hit Enter
Hit ↓, See X05, write 0 then hit Enter, Hit ↓ see Y05 , write 3 Hit Enter
Now data has been entered
To do calculations related to the model:
Hit 2nd and STAT
You see Lin on the screen, it refers to Linear regression, you want that. If you do not see Lin on the
screen you hit 2nd and SET keys, you will see Lin on the screen, if not and you see some other thing, hit
2nd and SET again, until on the screen you see Lin, which is saying linear regression, you want the results
of that. so
Hit↓, you see on the screen results of calculations
n =5 (this is number of data observations)
Hit↓, see X bar =4, it is mean of X value, mean return of Market
Hit↓, see S X = 3.16 , it is sample SD of X values ,
Hit↓, see δ X = 2.828, it is population SD of X, SDM, total risk of Market
Hit↓, see Y bar = 8, it is Sample SD of Y values, mean returns of MCB
Hit↓, see SY = 5.09, it is sample SD of Y, of MCB

171
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Hit↓, see δY = 4.561, it is population SD of Y, total risk of MCB


Hit ↓, see α = 2, this is alpha, intercept of linear regression line of MCB with Y- axis
Hit↓, see β= 1.5, this is slope of linear regression line , it is beta of MCB
Hit ↓, see r = 0.93 , this is correlation between X and Y, correlation between RM and R MCB

find COV MCB, M = SD MCB * SD M * Correlation MCB, M


=4.561 * 2.828 * 0.93 = 11.99, almost 12
Note: use population SDs because for these years these are the returns, these returns are not a sample of
returns that were available for these 2 securities for the above stated years.

Find VAR MCB = SD MCB * SD MCB = 4.561 * 4.561 = 20.8 same as found above by hand
Find VAR M = SDM * SDM = 2.828 * 2.828 = 7.99, almost 8, same as found above by hand
Find beta of MCB = COV MCB , M / VAR M = 12 / 8 = 1.5 same as found above by hand
To clear stat data : hit 2nd then DATA, then hit 2nd and hit CLR WORK

using FC -100 calculator.


Finding α, β, CORR ,R2 of MCB stock Example and data of MCB and market
returns already given for 2005-2009 will be used.
1. In FC-100 Go to STAT mode (green button)
2. Choose second option that is A + BX. You see two columns
3. Enter RM data from 2005 to 2009 in X column
4. Enter RMCB data in Y column
5. AC (Red button)
6. SHIFT and STAT buttons, you see 7 items menu
7. Choose 7 for regression, you see 5 items menu
8. Choose 1 for α of MCB , ExE, You get 2 as α of MCB
9. SHIFT and STAT buttons again, you see 7 items menu
10. Choose 3 for SD of market, ExE, you get 2.82 as SDm, you square it to get VARM=(2.8)2=8
11. SHIFT and STAT buttons again, choose 5 for variance
12. Choose 6 for SD of MCB, ExE , You get 4.56. You square it to get VAR MCB=( 4.56)2= 20.8
13. SHIFT and STAT buttons again choose 7 for regression, then choose 3 for correlation of MCB
with market, ExE, You get 0.93026. You can square it to get R 2=( 0.93026)2 =0.86538. You can
double check this R2 as given below.

172
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

A Lengthy Exercise:
Purpose of this exercise is to show you that Markowitz & Sharpe methods give similar expected R p,
VARp, SDP. Please make sure you understand the following exercise fully
Suppose you are portfolio manager, and your staff of security analysts have made the following estimates
for stocks A , B, C, and D using Sharpe’s linear model:
Intercept slope Variance of ROR
Alpha beta total Risk
αi βi VARi
Stocks
A 2% 1 200%2
B 3 1.2 130
C 1 1.5 180
D 2 0.75 50
Please note alpha is percentage ROR but beta is just number, it is not a percentage so it must be
written exactly as given.
expected ROR of stock market (RM ) for the next year is estimated to be 10% and total risk of stock
market (VAR M ) is estimated 70%2.
Required:
Q1: Please build an equally weighted portfolio of 4 stocks a, b, c, and ; show that ∑Xi = 1
Q2: Find expected return of each stock using Sharpe’s linear Market Model.
Q3: Find expected return of portfolio, Rp as :
i) ∑Xi Ri (Markowitz)
ii) ap + Bp Rm (Sharpe)
Q4: Find total risk of portfolio, VARp as :
i) Xi2 VARi + ∑∑ Xi Xj COVi,j (Markowitz)
ii) B2p VARM + VAR ep (Sharpe)
Q5: What is: 1) non diversifiable risk; 2) diversifiable risk; 3) relvant risk; 4) market risk in this
portfolio?
Q6: What percentage of its total risk is diversifiable and what percentage is non diversifiable R 2
Q7: In your view, is it a well diversified portfolio?

173
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Q8: In your view, is it a high risk portfolio?


Q9: What is correlation of portfolio return with the market return?
Solution :
Q1: In an equally weighted portfolio weight of each stock is equal, X a = Xb = Xc = Xd = 1/N . As in this
case N = 4 so each X is 1/N = ¼ = 0.25
∑Xi = Xa + Xb + Xc + Xd
= 0.25 + 0.25 + 0.25 + 0.25
∑Xi = 1

Q2: Expected Return of each stock, according to Sharpe’s Market model:

Ri = α i + B i RM
(note: ei is not used because its expected value is zero)
Estimated ROR of each stock using Sharpe linear relationship between Ri and Rm

Stock αi + βi RM = estimated Ri
A 2 + 1*10 = 12
B 3 + 1.2*10 = 15
C 1 + 1.5*10 = 16
D 2 + 0.75*10= 9.5
Q3 Expected return of portfolio
Markowitz

Rp = ∑Xi Ri
= Xa Ra + Xb Rb + Xc Rc + Xd Rd
= 0.25(12%) + 0.25(15%) + 0.25(16%) + 0.25(9.5%)
= 3% + 3.75% + 4% + 2.375%
= 13.125%
Sharpe:

Rp = αp +( βp * RM )
Where αp = ∑Xiαi
αp = Xa αa + Xb αb + Xc αc + Xd αd
= 0.25 (2%) +0.25 (3%) +0.25 (1%) +0.25 (2%)

174
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

= 0.5 + 0.75 + 0.25 + 0.5


= 2%
βp = ∑Xi βi
βp = Xa Ba + Xb Bb + Xc Bc + Xd Bd
= 0.25(1) + 0.25(1.2) + 0.25(1.5) +0.25(0.75)
= 0.25 + 0.3 + 0.375 + 0.1875
= 1.1125
Inserting values of alpha of portfolio and beta of portfolio and expected return of market in Sharpe model:

Rp= αp + βpRM
Rp = 2% + 1.1125(10%)
Rp = 2% + 11.125%
Rp = 13.125%
Note it is same as given by Markowitz formula above.

Q4 : Total risk of portfolio( VARp )

Markowitz

VARp = ∑Xi 2VARi + ∑∑Xi Xj COVi,j (i is not equal to j)

Since COVi,j = Bi Bj VARM as proved by Sharpe, therefore you can find COVs
COVa,b = Ba Bb VARM
= 1 X 1.2 X 70 = 84
COVa,c = Ba Bc VARM
= 1 X 1.5 X 70 = 105
COVa,d = Ba Bd VARM
=1 X 0.75 X 70 = 52.5
COVb,c = Bb Bc VARM
=1.2 X 1.5 X 70 = 126
COVbd = Bb Bd VARM
=1.2 X 0.75 X 70 = 63
COVc,d = Bc Bd VARM
=1.5 X 0.75 X 70 = 78.75
∑Xi 2 VARi = Xa2 VARa + Xb2 VARb + Xc2 VARc + Xd 2 VARd

175
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

= (0.25)2 200 + (0.25) 2130 + (0.25)2 180 + (0.25)2 50


= 12.5 + 8.125 + 11.25 + 3.125
= 35%2
The second term is:
∑∑Xi Xj COVi,j = Xa Xb COV a,b + Xa Xc COV a,c + Xa Xd COV a,d
5.25 + 6.56 + 3.28
+ Xb Xa COV b,a + Xb Xc COV b,c + Xb Xd COV b,d
+ 5.25 + 7.875 + 3.94
+ Xc Xa COV c,a + Xc Xb COV c,b + Xc Xd COV c,d
+ 6.56 + 7.875 + 4.92
+ Xd Xa COV d,a + Xd Xb COV d,b + Xd Xc COV d,c
+ 3.28 + 3.94 + 4.92

For example : Xa Xb COV a,b = 0.25 * 0.25 * 84 = 5.25. Similarly other values were calculated for the
terms in covariance matrix given above, inserting these values we get the following :
∑∑Xi Xj COVi,j = 5.25 + 6.56 + 3.28
+ 5.25 + 7.875 + 3.94
+ 6.56 + 7.875 + 4.92
+ 3.28 + 3.94 + 4.92
= 63.65%2
VARp = ∑Xi2 VARi + ∑∑Xi Xj COVi,j ( i≠j )
= 35% 2
+ 63.65%2
= 98.65%2
SDp = √VARp = √98.65% 2 = 9.93%

Sharpe:

VARp = Bp2 VARM + VARep


We need to calculate diversifiable risk ( VAR e p ) of portfolio . We know that
VARep = ∑Xi2 VAR ei. For individual stocks we are given VAR i, and betas, and we
also have total risk of market (VAR M ). First we can find for each stock its diversifiable risk (VARe i ) so
that VAR e p (diversifiable risk of portfolio) can be calculated .

176
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Sharpe proved that total risk for a stock is: VARi = Bi2 VARM + VARei .
so: VARei = VARi - Bi2 VARM.

Using this formulation we can estimate diversifiable risk (VARe i ) of each stock as follows:
Stock VARi – Bi2 VARM = VARei
A 200%2 –(1)2 * 70 =130 %2
B 130 –(1.2)2 * 70 =29.2
C 180 –(1.5)2 * 70 =22.50
D 50 –(0.75)2 * 70 =10.62
VARep =∑Xi2 VARei
=Xa2 VAR(ea) + Xb2 VAR(eb)+ Xc2 VAR(ec)+ Xd2 VAR(ed)
=(.25)2*130 + (.25)2*29.2 + (.25)2*22.5 +(.25)2*10.62
=12.02%2 diversifiable risk of portfolio

VARp = B2p VARM + VARep


VARp =(1.1125)2 *70 +12.02
=86.63%2 + 12.02%2
=98.65%2
So: Both formulations, that of Markowitz and that of Sharpe, give exactly same answers for VAR p.

Q5: What is 1) non diversifiable risk; 2) diversifiable risk; 3) relvant risk; 4) market risk in this portfolio
Non diversifiable risk is 86.63%2
Diversifiable risk is 12.02%2
Relevant risk is beta of portfolio that is 1.1125
Market risk is 70%2
Total Risk of portfolio is 98.65
Q6: What percentage of its total risk is diversifiable and what percentage is non diversifiable? R 2
86.63 / 98.65 = 0.878 or 87.8% of total risk is non diversifiable, it is also R squared or co-efficient of
determination. R2
12.02 / 98.65 = 0.122 or 12.2% of total risk is diversifiable , (1 - R2)

177
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

Q7: In your view, is it a well diversified portfolio?


As only 12% of its total risk is diversifiable therefore it seems quite diversified poprtfolio as most of its
total risk is non-diversifiable. If a portfolio had 80% or 90% of its total risk as diversifiable risk then you
would say it is not very well diversified. In any case since this portfolio has some of its risk as
diversifiable risk (about 12% of its total risk ) therefore strictly speaking it is not a fully diversified
portfolio; in a fully diversified portfolio diversfiable risk would be zero % of its total risk. Later in this
course you would learn how to build fully diversified portfolios whose total risk is composed of only
the non-diversifiable risk , and such portfolios have zero diversfiable risk.
Q8: In your view, is it a high risk portfolio?
As beta of this portfolio is more than one, so it is higher risk portfolio. Please remember that only
relevant risk is used in judging a portfolio as high or low risk portfolio, that is the beta of portfolio , and
you would judge riskiness of this portfolio based on its beta 1.1125 that is more than one , and therefore it
is high risk portfolio.
Q9: Correlation of portfolio return with the market return?
Correlation P, M = √R2 = √0.8663 = 0.93. So returns of this portfolio are highly correlated with returns
of stock market.

An Important Point of General Knowledge


Please note percentage decrease in anything is never more than 100% because when something has
declined 100% then it is no more in existence; in case of stock value or currency value it means stock or
currency is worthless if decline is 100%; and there is no possibility of further decline in stock value or
currency value. Therefore when you read in print media or watch electronic media reporting 134%
decline in value of rupee against US dollar during last 4 years, YOU SHOULD KNOW THEY DON’T
REALLY KNOW WHAT THEY ARE TALKING ABOUT. This error occurs due to using ending value
in denominator; whereas percentage decline or rise should always be calculated based on beginning value.

178
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

For example in 2008 US dollar was 60 rupees (0.01667$ = 1 Rs), in 2013 it was 90 rupees (0.01111$
= 1 Rs). You would conclude that rupee depreciated against dollar, therefore value of rupee in dollar
terms should be used:
= (Ending value of Rs – Beginning value of Rs ) / Beginning value of Rs
=(0.01111- 0.01667) /0.01667
=- 0.00556 / 0.1667
= -0.3335
=-33.35%
And you would conclude that rupee has depreciated against US dollar by 33.35% from 2008 to 2013.
On the other hand if you want to talk from US perspective and want to say about appreciation of US
dollar against Pak rupee, you would say dollar appreciated between 2008 and 2013, then you need to use
value of dollar in rupee terms.
= (ending value of dollar – beginning value of dollar) / beginning value of dollar
= (90 - 60)/60
= 30/60
=0.5
=50%
And you would conclude that US dollar has appreciated 50% against Pak rupee between 2008 and 2013.

Please note: percentage loss in value of rupee (devaluation or more correctly depreciation in rupee against
dollar) was 33.35% but percentage gain in value of dollar (appreciation of dollar against rupee) was 50%.
It is always that way; percentage gain is always higher than percentage loss; and percentage loss in value
of shares or currencies can never be more than -100%, but percentage gain can be more than 100%.
Let us look at the actual data from 1972 to 2016, that is 45 years.
For example: US dollar in 1972 was 4.5 rupees (0.2222 $ = 1 Rs) and by the end of 2016 it was 105
rupees (0.0095 $ = 1 Rs)
In 45 years, Depreciation in value of rupee
= (ending – beginning) / beginning value
=(0.0095 – 0.2222)/0.2222 = - 0.957 or – 95.7%

For the same 45 year period, appreciation in value of $


= (ending - beginning) / beginning value

179
Investment Analysis & Portfolio Management. MBA II. Lahore School of Economics. Spring 2022.
Instructor: Dr. Sohail Zafar. TA: Ms Maheen Chaudhry and Ms Roha Asim

= (105 - 4.5) / 4.5 = 22.33, that is 22.33 * 100 = 2,233.33% appreciation in value of dollar against Pak
rupee.
Again notice that depreciation is less than 100% but appreciation is more than 100%, in fact 2 thousand 2
hundred and 33 % was appreciation in value of dollar against rupee; but depreciation in value of rupee
against dollar was 95.7%

As business graduates, this is the type of mistake you are expected not to make; never say a currency lost
value by more than 100% or a share lost value by more than 100%; but a currency or a share can gain
value by more than 100%.

180

You might also like