UNIT 10 PORTFOLIO ANALYSIS

Portfolio Analysis

Objectives
The objectives of this Unit are to:

explain and illustrate the concepts and measures of return and risk as they
apply to individual assets as well as portfolio of assets

highlight the concept of diversification of risk

discuss the portfolio selection problem and the process.

Structure
10.1

Introduction

10.2

Inputs to Portfolio Analysis

10.3

10.2.1

Return and Risk Characteristics of Individual Assets

10.2.2

Expected Return and Risk of a Portfolio

10.2.3

Diversification of Risk

Portfolio Analysis and Selection

10.3.1

Portfolio Risk Selection Problem

10.3.2

Selection of Optimal Portfolio

10.4

Summary

10.5

Key Words

10.6

Self-Assessment Questions/Exercises

10.7

Further Readings

10.1

INTRODUCTION

Suppose you believe that investments in stocks offer an expected return of 20% while
the expected return from bonds is around 10%. Would you invest all your money in
stocks because stocks offer two times of return of bonds? Probably not; because you
may be aware of an axiom that `putting all eggs in a basket is not desirable'. You
would prefer a portfolio of securities rather than investing all your savings in a single
security or group of securities belonging to same category. In this book, we will
discuss more about how one should go about in constructing a portfolio that suits
specific objectives of the investment. In this unit, we will address some of the basic
issues measuring return and risk of the portfolio before addressing the main issue of
selecting optimal portfolio.
The term `portfolio' generally means a collection or combination and in the context of
investment management, it means a collection or combination of financial assets (or
securities) such as shares, debentures and government securities. However, in a more
wider context the term `portfolio' may be used synonymously with the expression
`collection of assets', which can even include physical assets (gold, silver, real estate,
etc.). What is to be borne in mind is that, in the portfolio context, assets are held for
`investment' purposes and not for `consumption' purposes.
We will begin with the analysis of return-risk characteristics of individual assets, and
then proceed to examine how individual assets combine into a portfolio to determine
its return and risk attributes. Having done so, our next logical step would be to
consider the question; how can an investor make a choice when facing an infinite
number of possible portfolios? Or, more precisely, how can the investor decide which
assets to hold and how much to invest in each? Quite obviously, the ultimate choice
of a portfolio will hinge on the investor's attitude towards risk and return.

Portfolio Theory

10.2

INPUTS TO PORTFOLIO ANALYSIS

Portfolio analysis builds on the estimates of future return and risk of holding various
combinations of assets. As we know, individual assets have risk return characteristics
of their own. Portfolios, on the other hand, may or may not take on the aggregate
characteristics of their individual parts. In this section, we will reflect on the
assessment of return-risk attributes of individual assets and portfolios.
10.2.1
Return and Risk Characteristics of Individual Assets
Any investment decision requires an estimate of return and risk associated with the
investment. However the most difficult task of investment decision is estimation of
return and risk. We spent the whole of Block III in discussing how should one
estimate the future value of the asset so that return can be measured. If we are able to
estimate a range of expected return, then it is possible to estimate the probabilities
associated with the range of expected return to get the risk measure. In practice,
however, the return and risk of the securities are estimated based on the historical
return and risk of securities. A stock's single period basic return is:

Total Return t =

Dividend + (Market Price t - Market Price

Market Price t-1

t-1

There are different measures of historical return. The most elementary form of return
measure is holding period yield or return. Here, the dividend received during the
holding period is added along with the capital gain and divided by the purchase price.
If the holding period is more or less than one year, normally the holding period return
is stated for one-year period. This measure is not much useful if one wants to
measure the risk associated with the security. There are two other measures of return
by which one can measure risk.
a) Arithmetic Average: The arithmetic average return is equal to sum of returns
of period and divided by `n'. For instance, if the stock has offered a holding
period return of 11% in period 1, 12% in period 2 and 16% in period 3, then the
arithmetic average return is equal to 13%. Though it is better than holding
period return, this measure suffers because of its failure in considering time
value of money. Another problem of this measure is differential treatment of
positive and negative return. For instance if a stock price increases from Rs. 10
to Rs. 20 in period 1 and declines back to Rs. 10 in period 2, the Arithmetic
average return is still positive value of 25% (Period I return is 100% and Period
2 return is -50%; Total return is 50% and hence average return is 25%).
b) Geometric Average: The geometric average return is based on the compound
value and is also called time-weighted average return. It addresses the problem
of differential treatment of positive and negative return described above. The
geometric average return is computed as follows:

GMR = [(1 + R1 ) x (1 + R 2 ) x (1 + R 3 ).....x (1 + R n )]1/n 1

Illustration: Five years back, you have applied and was allotted 100 shares of a
company at the rate of Rs. 50 per share (Face Value Rs. 10). The price at the end of
each year along with annual dividend per share received from the stock are as
follows:
1
2
3
4
5
Year

Dividend per share (Rs.)

1.5

1.5

Market Price (Rs.)

40

55

70

77

91

Find the Holding Period Return (HPR), Arithmetic Average and Geometric Average
return of the stock.
HPR : [Dividend (Rs. 8) + Capital Appreciation (Rs. 41)] / Investment (Rs. 50)
: 49/50 = 98% for five years or 19.60% per year
AA Return : [R1 (-18%) + R2 (41.25%) + R3 (30%) + R4 (12.86%) + R5(20.78%)]/5
17.38%
Note: R1 is equal to [(41-50)/50], R2 is equal to [(56.5 - 40)/40], etc.

: [(1+R1)x(1+R2)x(1+R3)x(1+R4)x(1+R5)]1/5 1
: [(.82)x(1.41)x(1.30)x l.13)x(1.21%)]1/5-1
: 15.47%
As you may observe, for the same set of data, we get different values of return. HPR
is the highest and GAR is the lowest. The Geometric Return is lower than other two
returns because of compounding. In Table 10.1, the different measures of return of
NSE-50 companies are given based on last ten years data. The list contains only for
the companies, which have 10-year listed life.
In addition to the above two types of return, a foreign investor or foreign fund would
compute dollar-weighted return to adjust differences in the foreign exchanges
between the point of investment and sale. For example, if a foreign fund purchased a
stock at Rs. 50 today when the US Dollar - Rupee rate is Rs. 50 per US Dollar and sold
the stock at Rs. 55 at the end of one year, the holding period return in Rupee term is
10% [(55-50)/50]. However, if the Rupee depreciates during this period and quotes
Rs. 56 per US Dollar, the foreign fund incurs a loss because it can get less than one
Dollar with the sale value of the stock. The loss is equal to 1.79% [$1- $(55/56)].
While historical return gives a fair idea about the future return, they are often used to
measure the risk. It is true that it is more relevant to use expected risk by measuring
the probability associated with various returns, often such measure is, not used in
practice. Historical data is generally used to measure the risk. The risk associated
with an investment in. stocks is measured using variance or standard deviation of the
historical return. Table 10.1 also shows the standard deviation of stock return along
with different return measures. A question with variance as a measure of risk is: why
count `happy' surprises (those above the average historical return or expected return)
at all in a measure of risk? Why not just consider the deviations below the average
historical return or expected return (i.e. the downside danger)? Measures to do so
have m u c h to recommend them. But if a distribution is symmetric, such as the
normal distribution, the result will be the same. Because, left side of a symmetric
distribution is a mirror image of the right side. Although distributions of historical or
forecasted returns are often not normal, analysts generally assume normality to
simplify their analysis.
Table 10.1: Return and Risk Measures of select stocks of NSE-50
GA Return

Portfolio Analysis

Portfolio Theory

In Block 1 of this course, we had discussed how to compute mean and variance (or
standard deviation), so we need not reiterate the procedure here. You may, however,
look up appendix at the end of this unit to quickly revise the concepts of portfolio
return and risk. We may now refer to Figure 10.1 that depicts the distribution of
returns that might be expected for two investments, A and B.
Figure 10.1 : Possible Outcomes of two Independent Investments

Expected Rate of Return

The mean or expected return, at the vertical dotted line, is the same for both
investments. But, investment B is more risky. With investment A, the distribution of
returns (or possible outcomes) is more closely grouped about the mean value. In
other words, the variance is smaller than that of investment B. Consequently, it can
be said with greater degree of accuracy that our forecast will be close to the actual
return from investment A.
When we move from evaluating a single asset in isolation to evaluating a portfolio,
our return-risk analysis changes. Return is still the expected return, but for a portfolio
the return will be the average return from all the assets held in the portfolio. Risk is
still the variance (or standard deviation) of the expected returns from the portfolio.
The investor is still concerned with upside potential and downside danger. However,
the risk of a combination of assets is very different from a simple average of the risk
of individual assets. Most dramatically, the variance of a portfolio of two assets may
be less than the variance of either of the assets themselves. We will examine all these
aspects in the discussion that follows.
10.2.2
Expected Return and Risk of a Portfolio
The return on a portfolio of assets is simply a weighted average of the return on the
individual assets. The weight applied to each return is the fraction of the portfolio
invested in that asset. Thus,

R (p)

x R
i

i=1

(10.1)

Where

Portfolio Analysis

R(p)
Xi

= the expected return of the portfolio;

= the proportion of the portfolio's initial fund invested in
asset i
Ri
= the expected return of asset `i'; and
n
= the number of assets in the portfolio.
To illustrate the application of the above formula, let us consider a portfolio of two
equity shares A and B. The expected return on A is, say, 15 per cent and that on B is
20 per cent. Further assume that we have invested 40 per cent of our fund in share A
and the remaining in B. Then, the expected portfolio return will be
0.40 x 15 + 0.60 x 20 = 18 per cent.
It may be noted here that portfolio weight can be either positive or negative. In case
of securities, the weight will be negative when investor enters into `short sales'.
Usually, the investors buy securities first and sell them later. But with a `short sale'
this process is reversed; the investors sell first the securities that they do not possess,
and buy them later to cover the sales. Since institutional investors in our country do
not enter into such short sales, we will ignore the situation of short sales in the
present discussion as well as in our dealing with the subject matter in subsequent
units.
Having discussed the computation of expected portfolio return, we now turn to the
measurement of variance of portfolio's return (i.e., the risk of the portfolio). As
mentioned earlier, assets when combined may have a greater or lesser risk than the
sum of their component risks. This fact arises from the degree to which the returns of
individual assets move together or interact. It is vital, therefore, to consider
covariance of returns in estimating portfolio variance.
Let us intuitively understand why the risk of portfolio of stocks is lower than the sum
of the risk of the individual stocks. Suppose you have invested your wealth in two
securities. Assume the securities prices move in opposite direction such that when
one security gains, the other security incurs loss. For instance, if one security reports a
gain of 10% in a period, the other one reports a loss of 5%. In the next period, the
security, which lost 5% shows a gain of 12% but the other one lost 7%. Though the
two securities individually has a variance, an investment of equal amount in these
two securities will have a constant return of 5% during the period. In other words, the
variance of the portfolio return is zero.
We may not find assets that move in opposite direction in the real life because the same
set of factors affect the performance of several assets. At the same time, assets are
also not perfectly affected by the factors and hence there is a scope for reduction in the
variance of portfolio of assets. We will explain the same with the help of portfolio of
investments in three stocks. Three equity shares with the following return-risk
characteristics are considered for this purpose:
Monthly Average
Standard
Proportion
Return (%)
Deviation (%)
Invested (%)
ACC
8.89
19.55
33
Century Textiles
Hindustan Lever

5.12
3.42

7.99
6.18

33
34

Monthly returns here represent average appreciation of share prices estimated on the
basis of price movements over 26 monthly intervals during the period 1989 to
February 1991. A weighted average of standard deviation of each share returns works
out to (.33 x 19.55 + .33 x 7.99 + .34 x 6.18) 11.18 per cent. However, a direct
estimation of standard deviation of historical portfolio returns yields a figure of 9.61
per cent. Thus, the portfolio risk, as measured by standard deviation, is less than the
sum of component risks. The lower portfolio risk in this case is due to the fact that
the returns of the select scrips have not exhibited greater tendency to move together.
In fact, the correlation co-efficient of returns (we will discuss about covariance and
correlation co-efficient after a while) between ACC and Hindustan Lever and that
between ACC and Century Textiles were found to be low (.3 and 4 respectively)
during the period under consideration.

Portfolio Theory

The computation of the portfolio variance in the above example is based on the
following formula:

2 (p) =

i=1

j=1

x x
i

ij

where ij denotes the covariance of returns between asset i and asset j. An

explanation of the formula is now in order.
We start off with the most important element of this formula, namely, covariance. It
is a statistical measure of how two random variables, such as the returns on asset
and `move together'. A positive value for covariance indicates that the assets
returns tend to go together. For example, a better-than-expected return for one is
likely to occur along with a better-than-expected return for the other. A negative
covariance indicates a tendency for the returns to offset one anther. For example, a
better-than-expected return for one asset is likely to occur along with a worse-than expected return for the other. A relatively small or zero value for the covariance
indicates that there is little or no relationship between the returns for two assets.
Closely related to covariance is the statistical measure known as correlation. The
relationship is given by

ij = ij i j

where ij denotes the correlation coefficient between the return on asset `i' and that
on j. The correlation coefficient simply rescales the covariance to facilitate
comparison with corresponding values for other pairs of random variables. The
coefficient ranges from - 1 (perfect negative correlation) to + 1 (perfect positive
correlation). A co-efficient of 0 indicates that returns are totally unrelated.
Given an understanding of covariance and correlation, next logical step is to know
how the double summation of equation (10.2) is performed. The easiest way to
understand equation (10.2) is form a n x n table. Suppose there are three stocks in the
portfolio, then the equation 10.2 is equal to
Given an understanding of covariance and correlation, next logical step is to know
how the double summation of equation (10.2) is performed. The easiest. way to
understand equation (10.2) is f o r m a n x n table. Suppose there are three stocks in
the portfolio, then the equation 10.2 is equal to

The above table has to be expanded if the number of securities are more than three.
For example, if the number of stocks are 5, then we have to frame 5 x 5 table. The
variance of the portfolio is equal to sum of the values in the above cells. In the
above table, Wx, Wy and Wz are proportions of investments made in security X, Y
and Z. The variance of the security X, Y and Z appears in the diagonal cells as
xx , yy, and zz . The covariance between the securities appears in non-diagonal
cells as xx , yy, and zz . You may also note, the covariance of xy is equal to
covariance of yx . Thus, if the number of assets in the portfolio is three, then the
portfolio variance can be expressed as follows:

2 (p) = w x 2 xx + w y 2 yy + w z 2 zz + 2w x w y xy + 2w x wz xz + 2w y wz yz

10

The variance of the portfolio used for the computation of portfolio return can be
computed as follows. To explain this computation, we have used weekly price
returns of three well-know stocks of Indian market namely, Hindustan Lever
(HLL), Infosys and Tisco from January 1997 to December 2001. We need the
following details:

Weekly Return, Variance and Standard Deviation of HLL, Infosys and Tisco
Mean

HLL
0.38%

Infosys
0.64%

Tisco
-0.25%

Variance

0.0029

0.0133

0.0048

Standard Deviation

5.37%

11.55%

6.93%

Portfolio Analysis

Variance and Co-variance Matrix of HLL, Infosys and Tisco

Covariance between

HLL

Infosys

Tisco

HLL

0.0029

0.0001

0.0011

Infosys

0.0001

0.0133

0.0016

Tisco

0.0011

0.0016

0.0048

Note: The values of diagonal elements are variance.

The variance of the portfolio is equal to:
Stocks

Proportion

HLL

Infosys

Tisco

of

50%

30%

20%

HLL

50%

. 5 x . 5 x.0029 . 3 x . 5 x . 0 0 0 1

.2x.5x.0011

Infosys

30%

.5 x .3 x .0001 .3 x .3 x .0133

.2 x .3 x .0016

Tisco

20%

.5 x .2 x .0011 .3 x .2 x.0016

.2 x .2 x .0048

The sum of the cells is equal to 0.002538, which is equal to the variance or risk of the
portfolio. The risk of the portfolio can also be expressed in terms of standard
deviation. In such case, the portfolio risk is equal to:

2 (p) = .002538

( p ) = .002538 = 0.050375 or 5.03%

The return of the portfolio is 0.33% per week. It may be noted that the risk of the
portfolio that we computed is much lower than the weighted average variance or risk
of the individual securities in the portfolio. The reduction portfolio risk is mainly on
account of diversification and less than perfect correlation between the three stocks.
10.2.3

Diversification of Risk

Efforts to spread and minimize portfolio risk take the form of diversification. Most
investors prefer to hold several assets rather than putting all their eggs into one
basket, with the hope that if one goes bad, the others will provide some protection
from extreme loss. Surely enough, there is merit in this approach; although some
investors hold a contrary view point that recommends putting all eggs into one basket
and then keeping a sharp eye on the basket.
It is not difficult to understand that adding more assets in the portfolio can reduce the
overall portfolio risk. Consider the table drawn earlier to compute the portfolio risk
and look into the diagonal cells. The diagonal cells contain the variance of securities
in the portfolio. In that example, we assumed that an equal investment is made in
three stocks. The sum of the diagonal cells is equal to sum of the variance of three
securities multiplied by (1/3)2. Suppose, we add one more stock in the portfolio and
revise our weights to 0.25 for each stock. The values of diagonal cells is now equal to
sum of the variance of four securities multiplied by (1/4)2. We know (1/4)2 < (1 /3)2.
Suppose, if the number of securities in the portfolio is increased to 20, then the value
of the diagonal cells is equal to sum of the variance of individual securities multiplied
by (1/20)2. The value of (1/20)2 is equal to .0025 and close to zero. Since the
multiplier is now close to zero, the sum of the diagonal cells will

11

Portfolio Theory

reach close to zero. Thus, when a security is added to the portfolio, the value of
diagonal cells is close to zero and thus reduced the variance of the portfolio. However,
there is a limitation in adding securities to reduce the risk because the diagonal cells
value can not be reduced below zero (i.e. negative) to reduce the portfolio risk further.
Thus, beyond a level, diversification fails to yield further benefit by way of reducing
the risk. This is being illustrated in Figure 10.2.
Figure 10.2 : Diversification of Risk Risk

Number of Securities
It may be noted that beyond certain portfolio size, the reduction in risk is marginal and
insignificant.
We will discuss more about diversifiable and non-diversifiable risk in Unit 12.A word
of caution may, however, be urged here. The above discussion would appear to suggest
that the overall portfolio risk can be reduced by only increasing number of assets in the
portfolio. This is not true. Several empirical studies have indicated that a portfolio
comprising a few assets selected carefully for their risk-diversifying characteristics (i.e.
nature. and degree of variance and covariance), would be less risky than a portfolio of
considerably greater size with assets being selected without regard to risk. Thus, -what
matters in diversification is not the number of assets per se, but right kinds.
Activity 1
a)

Define the following terms.

i)

Portfolio Risk.

.....
.................................................................................................................................
ii)

Variance-Covariance Matrix.

b)

12

What is the major point illustrated through Figure 10.2.

10.3 PORTFOLIO ANALYSIS AND SELECTION

Portfolio Analysis

In the previous section, we have discussed how portfolio risk is measured. Let us
summarise important points before discussing how an investor can use the concept in
selection of the portfolio. Risk associated with investments can be reduced through
diversification and such diversification helps the investors to reduce the risk of the
portfolio. Investments in individual securities, risk (variance) associated with
individual securities and the relationships (covariance) between the securities are the
three variables that affect the risk of the portfolio. While diversification reduces the
unsystematic risk of the portfolio, the number of securities required to minimize the
portfolio risk is not very large. Finally more than the number of securities, what
matters in reducing the risk of the portfolio is the kind of securities included in the
portfolio. The last observation is stressed in this section.
10.3.1 Correlation between Securities and its Impact on Portfolio Risk
We have discussed that risk is reduced when the portfolio includes one stock in the
portfolio. The above observation is not universal in a sense that if the new stock is
perfectly correlated with other securities in the portfolio: In other words, the job of
investment analysts or any other persons responsible in constructing the portfolio is
to identify stocks or securities that are less related with each other for portfolio
construction. The risk of the portfolio can be reduced to zero if the correlation
between the assets included in the portfolio is equal to minus 1. However, such
securities are difficult to identify in the market. If two securities are perfectly
correlated, then there is no diversification benefit and such combination will not
reduce the risk of the portfolio. There are only very few securities in the market
whose correlation is equal to minus one. What is more prevalent in the market is
securities whose return are correlated between minus 1 to plus 1. Depending on the
level of correlation, diversification reduces the risk of the portfolios. The relationship
between the assets and its impact on portfolio risk is explained below in Figure 10.3
with the help of two securities.
Figure 10.3: Correlation and Portfolio Risk

(a)Correlation = - 1

(b) Correlation = +1

(c) Correlation = 0.72

Figure 10.3 (c) is more relevant for our discussion since the correlation between the
securities is often less than 1 and greater than zero. In such a situation, when an
investor combine such securities, the risk of the security is initially reduced. We will
show you with a real life example in the following Table:
Figure 10.3 (c) is more relevant for our discussion since the correlation between the
securities is often less than 1 and greater than zero. In such a situation, when an
investor combine such securities, the risk of the security is initially reduced. We will
show you with a real life example in the following Table:
SI.No.
Portfolio Risk
Proportion of Investment in Portfolio
Return
(Variance)
Hindustan Lever
Infosys
1
100%
0%
0.38%
0.00289
2
80%
20%
0.43%
0.00240
3
60%
40%
0.48%
0.00320
4
40%
60%
0.54%
0.00529
5
20%
80%
0.59%
0.00867
6
0%
100%
0.64%
0.01334
Note: Correlation between Hindustan Lever Ltd. and Infosys is .0087

13

Portfolio Theory

The risk and return of the portfolio is plotted below to show how the graph looks
similar to one shown in Figure 10.3 (C)
Figure 10.4: Risk and Return of Portfolios of HLL and Infosys

If there are 10 securities in the market, it is possible to draw the diagrams of the above
for a number of combination of two-securities.
10.3.2

Portfolio Selection

In the above section, we have shown that combination of securities normally reduces
the risk. Often, it also leads to an increase in return, which is good for investors. That
is, you are able to achieve higher return and also lower risk through diversification. The
problem is if there are large number of securities in the market, how to determine the
optimum portfolio, which reduces the risk while keeping the return constant or
increasing the return. We first provide an intuitive understanding of the concept. If
there are large number of securities in the market and if you are able to form a twosecurity portfolio and find the portfolio return and risk for various combinations as
discussed above, then you will have a large number of graphs as in Figure - 10.3 (C).
Figure 10.5: Risk and Return of Two-Stock Portfolios
Return

Risk
In the above Figure 10.5 we have shown six combinations. Now the issue is how to
select a portfolio, which is good in terms of minimizing risk and maximizing return. Now
carefully look into the above Figure particularly on the dashed line. There are five
portfolios offering same risk but different returns. Consider the two extreme Portfolios
- Portfolio X and Portfolio Y. While X offers lowest return, Y offers highest return for
the same level of risk. Now, we can say all four portfolios below Y are inefficient in a
sense that you would not buy such portfolio with the same risk level to earn lower
return. If we eliminate all such inefficient portfolios, we will get a smooth curve, which
connects the left extreme values of the curves. Such an efficient set of the portfolios is
shown in Figure 10.6.

14

The new curve A and B connects all left-extreme values of earlier portfolios and
become efficient set of portfolios. For instance, we don't have any portfolios above this
curve to show better return for a given level of risk. All portfolios below this curve of
A and B are inefficient and hence no one prefer such combination of stocks. All points
in the curve are efficient because it is not possible to evaluate two points in the curve
and conclude one is

better than the other. They are all efficient portfolios because for a higher risk, the
expected return is also high. Depending on the investors risk and return expectation,
they can pick up any combination. If an investor like to have low risk, then she or he
will select a combination of stocks close to point A. On the other hand, if an investor
likes to assumes more risk, she or he will prefer a portfolio close to point B.
Figure 10.6: Efficient Set of Portfolios

Portfolio Analysis

If the above understanding is clear intuitively, we can now proceed to learn how to
find an optimal portfolio. This requires an application of quadratic programming.
Minimize Variance of Portfolio Z :
Subject to :

i=1

j=i

Cov w w
ij

x E(R ) - E* = 0
x - 1 = 0
i

Combining the above three equations, we get an optimization equation to minimize the
risk:

Z = (

Covij w i w j )+(1 X i E(R i )-E*)+( 2 X i -1)

For a three securities portfolio, the optimization equation is as follows.

Z = X12 11 + X 2 22 2 + X 32 33 + 2X1 X 2 12 + 2X1 X 3 13 + 2X 2 X 3 23 + 1 (X1E1 + X 2 E 2 +

X 3 E 3 -E*) + 2 (X1 + X 2 + X 3 - 1)
Setting partial derivatives of Z with respect to all variables equal to zero (dz /dx1, dz
/dx2, dz/dx3, d1 and dz/ d 2 ), we get a set of five equations and solving the five
linear equations for the unknowns X1 X2 and X3, the proportion of investment to be
made in each of the stocks to get the desired return. The above quadratic
programming results will be in the form of three equations in the form of
X1= a+b1E (R)
X2 = a+b2E (R)
X3=a+b3E (R)
Where `a' and `b,' are known and one has to substitute the expected rate of return to
know the investment to be made in the three stocks.
The portfolio selection process as described above is not something new; the model
was presented by Harry Markowitz briefly in 1952 and later in a complete book
entitled Portfolio Selection-Efficient Diversification of Investments (1959). One
important concept that Markowitz emphasized for the first time was that some
measure of risk, and not just the expected rate of return, should be considered when
dealing with investment decision. Markowitz's approach to portfolio analysis and
selection attracted a number of academicians and practitioners, who subsequently
began to adjust the basic framework so that practical application could be more
readily considered. Another interesting thing happened. Following the presentation of
the model, there had been a wide spread realisation of how computers could be
utilized in investment decision making. Markowitz's own solution to portfolio
selection problem necessitates, as we will see in the next unit,

15

Portfolio Theory

application of computers. As a final remark, we may mention that Markowitz's work

marks the beginnings of what is today known as modern portfolio theory.
Activity 2
1)

List out four basic steps of portfolio selection process.

2)

Whose work marked the beginning of modern portfolio theory?

10.4

SUMMARY

The unit describes the basic components of portfolio selection process. Beginning
with the estimation of a portfolio's expected return and risk, which in turn involves
estimation of such input data as expected return, variance and covariance for each of
the assets contained in the portfolio, we have explained why an investor should
consider only the `efficient set' out of the feasible set of portfolios. Once the efficient
portfolios are delineated, the investor will next `select the `optimal' portfolio depending
upon his or her `trade-off' between return and risk. In terms of graphical analysis such
optimal portfolio will be located at the point where indifference curve that
summarises the investors risk-return trade-off, is tangent to the efficient set. In this
kind of approach to portfolio selection, it is assumed that rational investors are risk
averse and prefer more return or loss. Finally, the portfolio selection approach
presented here epitomises the Markowitz's model developed in early 1950s.

10.5

KEY WORDS

Portfolio refers t o collection of assets (financial or physical or both).

Portfolio weight refers to the fraction of available fund that is being invested in a
particular asset in a particular asset in the portfolio.
Expected rate of return is the return on an asset (or portfolio) over a holding period
that an investor anticipates to receive.
Standard deviation is a measure of the dispersion of possible outcomes around the
expected outcome of a random variable.
Variance is the squared value of the standard deviation.
Covariance is a statistical measure of the relationship between two random
variables.
Correlation coefficient is a statistical measure similar to covariance in,that it measures
the degree mutual variation between two random variables. The correlation coefficient
re-scales covariance to facilitate comparison among pairs of random variables. The
correlation coefficient is bounded by the values - 1 and +I.

16

Variance-covariance matrix is a table, which symmetrically arrays the covariance

between a number of random variables. Variances of the random variables lie on,the
diagonal of the matrix, w h i l e covariance between the random variables lie above or
below the diagonal.

Diversification means the spreading of investments over more than one asset with a
view to reduce the portfolio's risk (i.e., the variability of expected portfolio returns).

Portfolio Analysis

Feasible set (or opportunity set) represents the set of all portfolios that can be formed
by an investors, given a population of assets.
Efficient set (Efficient frontier) is the set of portfolios of a given population of assets
which offer the maximum possible expected return for a given level of risk.
Optimal portfolio means the feasible portfolio that offers an investor the maximum
level of satisfaction, given his or her own preference for return and risk. This
portfolio is located at the point of tangency between the efficient set and an
indifference curve of the investor.

10.6
1)

SELF-ASSESSMENT QUESTIONS/EXERCISES
The following forecasts have been made for investments A and B.
Investments A
Investments B
Return (%)
Probability
Return (%)
Probability
10
15
20
25
30

.05
.20
.50
.20
.05

2
12
20
28
38

.05
.25
.40
.25
.05

Calculate the expected rate of return and standard deviation. Which

investment has more upside potential and downside danger?
2)

If a portfolio's expected return is always equal to the weighted average of the

expected return of the component assets, why is not portfolio risk always
equal to the weighted average of the component assets' variances? Explain.

3)

Suppose an analysts has provided you with the following estimates in respect
of equity shares of Century Textiles, Escorts and Hoechst:
Century
Escorts
Hoechst
Expected monthly Return (%)
5
4
9
Standard Deviation (%)
8
7
17
Correlation coefficients
Between
Century and Escorts

of

Returns
0.4

Century and Hoechst

0.6

Escorts and Hoechst

0.3

Assuming that equal amounts of the available funds will be invested in the three
scrips, estimate the portfolio's mean return and standard deviation.
4)

Consider two securities with the following characteristics:

A

Expected Return

12

02

Standard Deviation

08

10

Assuming no correlation between the returns on two securities, calculated expected

return and standard deviation for each of the following portfolios:
Portfolio Weights (XA)
Security A

1.0

.75

.50

25

0.0

17

Portfolio Theory

Plot these portfolios with expected portfolio returns on x-axis and standard deviation
on y-axis. Locate the efficient frontier' and the portfolio with least risk or standard
deviation.
Can you precisely determine XA corresponding to the portfolio with minimum standard
deviation? (Hint: Obtain the equation for p with zero correlation between returns on
two securities. To find XA for which p is minimum, set the first order derivative of

p with respect to XA equal to zero, and then solve for XA).

5)

6)

7)

8)

18

Explain the following in your own words and using graphs:

a)
diversification of risk
b)
indifference curves
c)
selection of optimal portfolio
For a portfolio with the following characteristics, calculate the rate of return and
the standard deviation of the rate of return (r1 and p ):

Consider the data given below:

a) Form all possible portfolios consisting of two securities each, calculate the rate
of return and standard deviation of rate of return for each one of these
portfolios. You may assume that each portfolio has equal proportions of the
two securities.
b) Out of the set of portfolios formed in Q. 7a, identify the efficient portfolio (s)
Refer to the following observations for securities X and Y:

a)

Compute the sample mean returns for X and Y.

b)

Compute sample standard deviations for X & Y

c)

Compute the sample correlation co-efficient between the returns on X&Y.

10.7

Portfolio Analysis

FURTHER READINGS

Elton, Edwin J. and Gruber, Matin J., 1987 Modern Portfolio Theory and Investment
Analysis, John Wiley 84 Sons.
Alexander, Gordon J., Sharpe, William F., and Jeffery V. Baibey Fundamentals of
Investments, (3rd ed.) Prentice-Hall, Inc.
Answers to select Self-Assessment Questions
6)

7a)

rt

13.13%

1.77%

rAB

12.5%

AB

.63%

rAC

27.5%

AC

.06%

rBC

20%

BC

.62%

7b)

Portfolio constituted by securities A and C because it has the highest rate of

return and the lowest standard deviation.

8.

a)

rX

011;

b)

XY

-.0017

c)

XY

-.798

rY

.0625

19

Appendix

Portfolio Theory

Rate of return of a portfolio

The rate of return from a portfolio is computed as:
n

R (P)

X R
i

it

i=1

where
R(p)

rate of return of portfolio during period t

Xi

proportion, in terms of value, of security i in the portfolio

Rit

rate of return of security i during period t

number of securities comprising the portfolio

Rit, rate of return of security i during t is in turn computed as follows

Pit

(Pit-1 - Pit + Iit )/Pit

Pit+1

market price of security `i' at beginning of period d t+1 (or

alternatively at the end of period t)

Pit

market price of security `i' at beginning of period `t'

Iit

income in the form of dividends/interest etc. received from

holding security `i' during period `t'

where

To illustrate the concept, let us take an example of a portfolio comprising three securities A, B
and C. The relevant data is given below:
Price

No. held beg.

Price

Dividend

beg 1990
Rs.25/ share

1990
100

end 1990
Rs.28/ share

recd 1990
Rs. 2/ share

Rs.50/ share

30

Rs.49/ share

Rs. 4.5/ share

Rs.60/ share

100

Rs.65/ share

Rs. 1/ share

Security
A

Now, from the above data, we can compute XA, XB, and Xc as follows:
XA

XB

XC

(25 x 100)/[(25 x 100) + (50 x 30) + (60 x 100)]

2500/(2500 + 1500 + 6000)

2500/10000

0.25

(50 x 30)/[(25 x 100) + (50 x 30) + (60 x 100)]

1500/(2500 + 1500 + 6000)

1500/10000

0.15

(60 x 100)/[(25 x 100) + (50 x 30) + (60 x 100)]

6500/(2500 + 1500 + 6000)

6000/10000

0.6

Similarly, we can compute rA1990 , rB1990 , rC1990 as follows:

rA1990
20

= (28 25 + 2)/25
= 5/25
= 0.2 or (20%)

rB1990

Portfolio Analysis

= (48 - 50 + 4.5)/50
= 2.5/25
= .05 or (5%)

rC1990

= (65 -60 + 1)/60

= 6/60
= 0.1 or (10%)
Now we can compute the rate of return from the portfolio as follows:

rP1990

x r

i it

= ( . 2 5 x . 2 ) + ( . 1 5 x.05) +.6 x .1)

= .05 +.0075 + .06
= 0.1175 (11.75%)
From the above illustration we can see that the rate of return from a portfolio is a
function of the rate of returns from individual securities comprising the portfolio.
Risk of a portfolio
The most commonly used measure of variability of returns from a portfolio is the
standard deviation of the returns of the portfolio and this is computed as follows:

2 (p)

i=1

j=1

x x
i

(10.2)

ij

where

2p

Variance of returns of the portfolio

Xi

Proportion (in terms of value) of security ` i' in the

portfolio.

Xj

Proportion (in terms of value) of security `j' in the

portfolio.

ij

= Co-variance between the security i' and security j

ij

= ij i j

= Standard deviation of returns from security `i'

= Standard deviation of returns from security `j'

ij

= coefficient of correlation between the rates of return of

Where

securities `i' and j

n

= the number of securities in the portfolio

To illustrate the concept, let us take the example considered in the earlier section,
with some additional data provided as follows:
Security

Xi
(as computed above)

i
(as computed above)

A
025
0.1
B
0.15
0.1
C
0.60
02
The correlation matrix (between the rates of return of securities A, B and C) is given
below:
A
B
C
A
1.0
+.1
+.8
B
+.1
1.0
-.6
C
+.8
- .6
1.0

21

Portfolio Theory

Given this data op is computed as:

= (2.5) 2 + (.1) 2 + (.15)2 + (.1) 2 + (.6)2 + (.2) 2

+ (2 x .25 x .15 x .1 x .1 x .1) + (2 x .25 x .6 x .8 x .1 x .2)
+ (2 x .15 x .6 x (-.6) x .1 x .2)

= .00625 + .000225 + .0145 + .0075 + .000075 + .0048

= .025465
= 2.5%
From the above formula it is apparent that the risk of a portfolio which has a high
degree of correlation between the returns from its constituent securities would be
higher than the risk of a portfolio which has a low degree of correlation between the
returns from its constituent securities. Herein lies the crux of portfolio management in
order to reduce the risk of a portfolio, a portfolio manager would have to pick and
choose a diversified basket of securities such that the degree of co-movement between
their returns is very low.
Though the standard deviation of returns is a well-accepted measure of the risk
associated with a security, modem portfolio theories believe that a better index of risk
would be the "beta" value of a security. While the standard deviation measures the
total variability in returns from the security, the beta value is an index of that portion
of the variation, which can be attributed to market level factors, which are not unique
to the firm. Modem portfolio theorists argue that the risk, which arises from factors
unique to the firm are not that important because they can be eliminated through
diversification. The beta value of a portfolio indicates the degree of sensitivity of
returns from the portfolio to changes in the returns from the market as a whole and is
computed as follows
n

i i

i=1

j
erin
h
W
Where

p = beta of the portfolio

Xi = proportion (in terms of value) of security in the portfolio.

i = beta value for security i

n

= number of securities comprising the portfolio.

The beta value of individual securities (which indicates the degree of sensitivity of
returns from the security to changes in the returns from the market as a whole) is in
turn got as follows:

i =

Cov22(ri , rm )
Var(rm )

Where:

= beta value of security `i

Cov (ri, rm) = covariance between the returns from security `i' and returns
from the market
Var (rm ) = variance of returns from the market

22

From the above discussion on the beta value of a portfolio, it becomes clear that if we
use the beta value as an indicator of the risk associated with the returns from a
portfolio and if we wish to minimize this risk, we would have to pick and choose
securities which have very low beta values. In other words, in order to reduce the risk
of a portfolio we have to choose securities whose returns are fairly insensitive to
changes in the returns from the market as a whole.