Content uploaded by Abubakar Yahaya

Author content

All content in this area was uploaded by Abubakar Yahaya on Jul 15, 2015

Content may be subject to copyright.

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), pp. 98-106

International Journal of Pure and Applied Sciences and Technology

ISSN 2229 - 6107

Available online at

www.ijopaasat.in

Research Paper,

Statistical Analysis on the Advantages of Portfolio

Diversification

Abubakar Yahaya

1,*

, Amina Hassan Abubakar

2

and Jamilu Garba

3

1, 2, 3

Department of Mathematics, Ahmadu Bello University, Zaria, Nigeria.

* Corresponding author, e-mail: (ensiliyu2@yahoo.co.uk)

(Received: 2-8-11; Accepted: 14-10-11)

Abstract:

The classical mean-variance portfolio selection problem (PSP) pioneered

by Markowitz is, undoubtedly, one of the most frequently studied areas in finance, and

several financial analysts regard it as the foundation of modern portfolio theory

(MPT). The model in its basic form deals with making a choice from a universe of

assets to form a master asset known as portfolio of assets. The main aim of such a

strategy is to achieve a reasonable trade-off given the conflicting objectives related to

making a maximum possible return/profit at the most minimum risk possible, provided

that the right choice of constituent assets is made and proper weights (fraction of

investment funds) are correspondingly allotted. In this paper, we looked at the effects

and advantages of constructing a reasonably diversified portfolio from a pool of assets

while giving emphasis on the interrelationship existing among the portfolio’s

constituent assets.

Keywords:

Portfolio Selection, Diversification, Covariance, Correlation.

1. Introduction:

The main goal behind the concept of portfolio management is to combine various securities

and other assets into portfolios that address investor needs and then to manage those

portfolios so as to achieve the desired investments objectives. The investors’ needs are

mostly defined in terms of return and risk, and the portfolio manager makes a sound decision

aimed at maximizing return for investment risk undertaken [2].

The goal of investment decisions which is to maximize shareholders’ wealth and making

sound investment decisions that enhance shareholders’ wealth lies at the very heart of the

financial manager’s job. Wealth-enhancing investment decisions (corporate or personal)

cannot be made without understanding of the interplay between investment returns and

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 99

investment risk. The risk-return relationship is central to investment decision making,

whether evaluating a single investment or choosing between alternative investments.

Potential investors, for instance, will assess the risk-return relationship or trade-off in

deciding whether to invest in company securities such as shares or bonds. Investors will

evaluate whether, in their view, the securities provide a return commensurate with their level

of risk.

The conventional mean-variance portfolio selection model introduced by Harry Markowitz is,

undoubtedly, the corner stone of MPT [7, 8]. The basic assumption underpinning this model

is that: a rational investor has a multivariate normally distributed asset returns characterized

by the first two moments of the distribution in which the first moment (portfolio expected

return) stands for the return/profit of the investment, while the second moment (Variance)

stands for the corresponding risk involved. Given this, among other assumptions; Markowitz

showed that an efficient portfolio lies on a parabola-like frontier of points, often regarded as

Efficient Frontier (EF). A portfolio of assets is considered efficient, if for any expected

return, there exists no other portfolio with lower risk (variance); similarly, for any given

value of risk, there exists no other portfolio with higher expected return. Hence, the EF is

made up of all efficient portfolios. Depending on how many efficient portfolios an

investor/fund manager wants to generate on the EF, there are several computationally

efficient algorithms that can compute a single point/portfolio or even all points/portfolios [1,

6].

The remaining parts of this paper are organized as follows: in the next section we show how

to derive the formula/expression for computing portfolio expected return and risk (variance)

from the returns of portfolio’s constituent assets. Furthermore, we introduce the classical

mean-variance portfolio selection model together with the mathematical formulation of the

problem after which we review some related literature on the model’s possible extensions. In

section 3 we present the main idea behind portfolio diversification and the effect of forming a

diversified portfolio from a set of correlated as well as uncorrelated assets. Finally, we

conclude and offer some recommendations in section 4.

2. Materials and Methods/Definitions and Preliminaries:

2.1 Portfolio Expected Return and Risk (Variance)

This section is further subdivided into two in which the first subsection deals with deriving

the formula for computing the portfolio expected return, while the other section focuses on

deriving the formula for computing the portfolio risk (variance).

2.1.1 Expected return of a portfolio

Suppose there are n assets with rates of return

12

, ,..., 1,2,...,

tt nt

rr r fort T

. The values for the

expected returns of these rates of returns are given by:

11 2 2

, , ...,

nn

rr r r r r

.

Suppose further that, we construct a portfolio from these n assets each having a given fraction

of the investment fund, often regarded as weights

1,2,...,

i

wfori n . The corresponding

rate of return of the [constructed] portfolio in terms of the returns of individual assets is given

by:

11 2 2

1

...

Pnn

n

ii

i

rwrwrwr

wr

1

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 100

By taking expectation on both sides of equation 1 above, we obtain:

1

1

1

n

Pii

i

n

ii

i

n

Pii

i

rwr

wr

rwr

2

From equation 2 above, it can easily be understood that, the portfolio’s expected return,

P

r

can be obtained by taking the weighted sum of the assets expected return.

2.1.2 Variance (Risk) of a portfolio

To determine the variance of portfolio return; let us now denote the variance of asset

i by

2

i

;

the covariance of assets

i and j by

ij

and let the portfolio’s variance be denoted by

2

p

. By

performing a straightforward computation, we now have:

2

2

2

11

11

11

11

,

PPP

nn

ii ii

ii

nn

ii i j j j

ij

nn

ij i i j j

ij

nn

ij i i j j

ij

ij i i j j

rr

wr wr

wr r wr r

ww r r r r

ww r r r r

recall that r r r r the

11

nn

ijij

ij

n

ww

3

It is important to note at this point that, the variance (risk) of a portfolio’s return can be easily

computed from the covariances of the pairs of asset returns and the weights of the asset that

make up the portfolio.

2.2 The Classical Mean-Variance Model

The classical mean-variance model originally developed by Markowitz is aimed at finding a

portfolio of assets that seeks to minimize the (portfolio) risk subject to achieving a given

level of (portfolio) return. In this conventional formulation, the portfolio risk (objective

function) being minimized is quantified by the portfolio’s variance, which is the most

commonly used measure [2]. The model assumes a market composed of n different assets

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 101

with corresponding expected returns

i

r

, and asset covariances

ij

. The aim is to find a set of

fractions (decision variables) w

i

of an investor’s investment fund to be assigned to each asset

i so as to minimize the risk (variance,

2

p

) of the entire portfolio’s expected return, while at

the same time ensuring that the portfolio’s expected return attains a specified target, say ξ.

The only (practical) restrictions that accompany this model are the portfolio’s return and

budget constraints.

The return constraint ensures allotting weights to assets that guarantee achieving a specified

target return, while the budget constraint ensures that these fractions or asset weights must be

nonnegative and their sum must also be unity. The classical Markowitz model can be

represented mathematically by:

2

11

1

1

1

01

nn

pijij

ij

n

ii

i

n

i

i

i

Minimize w w

Subject to

wr

w

w

4

5

6

7

Equation 4 represents the objective function (Portfolio Risk), while equations 5 and 6

respectively represent the return and budget constraints. Constraint 7 ensures that no asset’s

weight falls outside the interval [0, 1], which literally means no short sales are allowed. The

above is a convex quadratic programming problem, since the objective function is quadratic;

the covariance matrix

ij

is positive definite and the constraints composed of both linear

equalities and inequalities.

A quite number of researches have been done and many are still going on purposely to

produce a near-perfect model that tries as much as possible to address the needs of an

investor, which serves as an improvement or rather an extension to the Markowitz’s (basic)

model. Even a superficial glance at the model above reveals that, it comprises of three main

features namely, the objective, constraints and variables/bounds. It is also worthwhile to note

at this point that, most of the studies conducted on PSP focused on making some

modifications to one or more of these attributes. Researchers such as Estrada [9], Ballestero

[4], Konno and Yamazaki [5], as well as Feiring et al [3] substituted variance with other

alternative measures and went extra mile to argue that variance should not serve as a measure

of portfolio risk; because it penalizes both upward and downward deviations from the mean

return.

Other researchers’ interest on improving the basic model tilted towards resolving other issues

to do with the real world investment constraints. For instance, Chang et al [12] while

maintaining the classical objective (variance) introduced additional constraints involving

cardinality as well as floor & ceiling constraints, after which they applied Genetic Algorithms

(GA), Tabu Search (TS) and Simulated Annealing (SA) to solve the constrained PSP. Crama

and Schyns [13] used SA to solve the enriched Chang et al [12] model by further

incorporating turnover (purchase and sale) and trading constraints.

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 102

3. Results and Discussion

3.1 Portfolio Diversification

It is worthwhile at this point to make it clear that, the total portfolio risk comprises of two

types of sub-risks namely: the market/systematic/nondiversifiable risk and the other known as

unique/firm-specific/non-systematic/diversifiable risk (i.e. Total Portfolio risk = Market risk

+ Non-market risk). Thus, it should be made known that, it is (theoretically) possible for

investors to construct portfolios and (possibly) eliminate (non-market/non-

systematic/diversifiable) part of the total portfolio risk through an intelligent investment

practice known as diversification.

Portfolio diversification is, essentially, a defensive technique to counter the problem of

investment risk. Imagine one wish to transport a basketful of eggs from one place to another;

there is, of course, different options open to him to accomplish that task. Some of these

options might include:

Carrying the whole basketful of eggs on foot at one go.

Rationing and taking the eggs on foot to the desired place on several journeys.

Carrying the whole basketful of eggs in a car to the desired place at one go.

Rationing and taking the eggs in a car to the desired place on several journeys.

Rationing the eggs among many people and each will be responsible for some.

Apparently, these are just some of the few possible options aimed at ensuring that the eggs

reach their destination safely. However, it is obvious that different risks are attached to the

different egg-carrying options. Therefore, if it makes sense to divide the eggs in some way to

reduce/spread the risk of disaster, it may also make sense to build an investment portfolio in

order to minimize investment risk; in the sense that, what is a good strategy for the carriage

of eggs may also be a good strategy for investment in marketable securities (stocks, shares,

e.t.c.).

An important way to reduce the risk of investment is to diversify one’s investments. For

instance, if a given investor’s portfolio consisted of [only] stocks from the banking sector,

there is every likelihood that such an investor would face a substantial loss in the value of

his/her portfolio if any major event adversely affected the banking industry – as was

witnessed in 2009 when there was a global economic meltdown that affected the banking

industry more than any other and whose impact is felt across the globe up till this time.

Portfolio diversification refers to mixing up your investments (i.e. making them more

diverse) and such a [good] practice guarantees an investor at least two main advantages as

follows:

Portfolio diversification helps an investor in minimizing the chances of losing all

his/her investment capital; if a given stock, sector or industry performs poorly in the

market.

It also helps an investor increases his chances of being in the right place at the right

time; if a given stock, sector or industry performs pretty well in the market.

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 103

Let us now show how a (unique/firm-specific/non-systematic/diversifiable) risk can be

somehow eliminated using the strategy of portfolio diversification. Now notice that equation

1 above can be re-written as:

222

111,

[Sum of asset variances Sum of asset covariances]

nnn

Pii ijij

iijji

www

8

3.1.1 Correlated Assets

Two assets i and j are said to be correlated if

0.. 0

ij ij

ie

, and suppose that the

investment funds are shared equally among the portfolio’s constituent assets; that is, let

1 1,2,...,

i

wni n be a weight associated with asset i. Now the portfolio risk in equation

5 above can be re-written as:

2

2

22

111,

2

111,

1

1

1

1

1

nnn

ij

i

P

iijji

nnn

ij

i

iijji

nn

n

nn n nn

n

Average Variance Average Covariance

nn

9

Taking limits on both sides of equation 9 above leads to obtaining:

2

2

1

1

P

nn

P

n

n

lim lim Average Variance Average Covariance

nn

lim Average Covariance

10

The above relation can be graphically represented as in Figure 1 below:

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 104

3.1.2 Uncorrelated Assets

Assets i & j are said to be uncorrelated if

0 . . 0, , 1,2,..., |

ij ij ij ij i j

ie i j n i j since

.

Again, if we let

1 1,2,...,

i

wni n

, the portfolio risk (equation 8 above) takes the form:

222

1

2

2

1

2

1

1

1

n

Pii

i

n

i

i

n

i

i

w

n

nn

Average Variance

n

11

Therefore, by taking limits:

2

2

1

0

P

nn

P

n

lim lim Average Variance

n

lim

12

Similarly, the above relation can be pictorially represented as shown in Figure 2 below:

Number of stocks

Diversifiable (non-Market) risk

Total Portfolio risk

0

Nondiversifiable (Market) risk

Figure 2

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 105

Now concerning the decision to take on how many securities are needed to constitute a

portfolio that will be able to eliminate most or all of the non-systematic risk, opinion differs

among researchers. For instance, Statman [11] argue that: “a well diversified portfolio of

randomly chosen stocks must include at least 30 to 40 stocks.” However, this contradicts the

stand of Evans and Archer [10] in which they raise doubts concerning the economic

justification of increasing a portfolio size beyond 10 or so securities; as according to them:

“virtually all economic benefits of diversification are exhausted, provided the portfolio

contains ten or so stocks”.

4. Conclusions

From the above statistical analysis, it can easily be observed that, it is possible to diversify

away a portion of Total Portfolio risk. The other portion of the Total Portfolio risk that

remains even after such diversification is called the Market risk – a risk attributable to

market-wide sources.

It can also be observed that, the degree of relationship (correlation) existing among various

securities in the market plays a vital role in portfolio’s risk reduction; in the sense that if the

assets (that makes up a given portfolio) are correlated, the (diversifiable) risk cannot be

reduced lower than the Average Covariance. On the other hand, if portfolio’s constituent

assets are uncorrelated, the (diversifiable) risk can be completely eliminated, and the only

risk an investor or portfolio manager has to contend with, is the non-diversifiable, otherwise

known as the Market risk. In view of the foregoing analyses and discussions, we advice

investors and portfolio managers to consider uncorrelated securities in constructing their

portfolios as it is proven to be the most intelligent investment decision any investor/portfolio

manager can make.

References

[1] A. F. Perold, Large-Scale Portfolio Optimization, Management Science, 30 (1984), 1143–

1160.

[2] A. Yahaya, Computational aspects of alternative portfolio selection models in the

presence of discrete asset choice constraints. An unpublished MSc Dissertation,

Department of Mathematical Sciences, Brunel University, Uxbridge, Middlesex, United

Kingdom, (2004).

[3] B. R. Feiring, W. Wong, M. Poon and Y. C. Chan, Portfolio selection in downside risk

optimization approach: Application to the Hong Kong stock market, International Journal

of Systems Science, 25(11) (1994), 1921 – 1929.

[4] E. Ballestero, Mean-Semivariance Efficient Frontier: A Downside Risk Model for

Portfolio Selection, Applied Mathematical Finance, 12(1) (2005), 1 – 15.

[5] H. Konno and H. Yamazaki, Mean-Absolute Deviation Portfolio Optimization Model and

Its Applications to Tokyo Stock Market, Management Science, 37 (1991), 519 – 531.

[6] H. M. Markowitz, P. Todd, G. Xu and Y. Yamane, Computation of mean-semivariance

efficient sets by the Critical Line Algorithms, Annals of Operations Research, 45 (1993),

307– 317.

[7] H. M. Markowitz, Portfolio Selection, Journal of Finance, 7 (1952), 77 – 91.

Int. J. Pure Appl. Sci. Technol., 7(2) (2011), 98-106. 106

[8] H. M. Markowitz, Portfolio Selection: Efficient Diversification of Investments, 2

nd

edition, Cambridge, MA, Basil Blackwell, 1959.

[9] J. Estrada, Mean-Semivariance Optimization: A Heuristic Approach, Journal of Applied

Finance – Spring/Summer (2008), 57 – 72.

[10] J. L. Evans and S. H. Archer, Diversification and the Reduction of Dispersion: An

Empirical Analysis, Journal of Finance, 23(5) (1968), 761 – 767.

[11] M. Statman, How many stocks make a Diversified Portfolio? The Journal of Financial

and Quantitative Analysis, 22(3) (1987), 353 – 363.

[12] T. J. Chang, N. Meade, J. E. Beasley and Y. M. Sharaiha, Heuristics for cardinality

constrained portfolio optimization, Computers & Operations Research, 27 (2000), 1271-

1302.

[13] Y. Crama and M. Schyns, Simulated Annealing for Complex Portfolio Selection

Problems, European Journal of Operational Research, 150 (2003), 546 – 571.