Chapter Six

Modern Portfolio Theory

6.1 The Concept of Portfolio

An investment portfolio is a collection of investment assets, such as stocks, bonds, mutual funds,
derivatives, real estate and commodities. Investment portfolios should be constructed after
carefully assessing the investor's goals, objectives and portfolio, such as age, growth and income
requirements, risk tolerance and liquidity needs. These investments are chosen generally on the
basis of different risk-reward combinations: from 'low risk, low yield' (gilt edged) to 'high risk,
high yield' (junk bonds) ones; or different types of income streams: steady but fixed,
or variable but with a potential for growth.

6.2 Definition of Risk and Its Alternative Measures

Although there is a difference in the specific definitions of risk and uncertainty, for our purposes
and in most financial literature the two terms are used interchangeably. In fact, one way to define
risk is the uncertainty of future outcomes. An alternative definition might be the probability of an
adverse outcome. Subsequently, in our discussion of portfolio theory, we will consider several
measures of risk that are used when developing the theory.

One of the best-known measures of risk is the variance, or standard deviation of expected
returns. It is a statistical measure of the dispersion of returns around the expected value whereby
a larger variance or standard deviation indicates greater dispersion. The idea is that the more
disperse the expected returns, the greater the uncertainty of future returns.

Another measure of risk is the range of returns. It is assumed that a larger range of expected
returns, from the lowest to the highest return, means greater uncertainty and risk regarding future
expected returns. Instead of using measures that analyze all deviations from expectations, some
observers believe that when you invest you should be concerned only with returns below
expectations, which means that you only consider deviations below the mean value. A measure
that only considers deviations below the mean is the semi-variance. Extensions of the semi-
variance measure only computed expected returns below zero (that is, negative returns), or
returns below some specific asset such as T-bills, the rate of inflation, or a benchmark. These
measures of risk implicitly assume that investors want to minimize the damage from returns less
than some target rate.

Assuming that investors would welcome returns above some target rate, the returns above a
target return are not considered when measuring risk. Although there are numerous potential
measures of risk, we will use the variance or standard deviation of returns because (1) this
measure is somewhat intuitive, (2) it is a correct and widely recognized risk measure, and (3) it
has been used in most of the theoretical asset pricing models.

6.3 Markowitz Portfolio Theory

In the early 1960s, the investment community talked about risk, but there was no specific
measure for the term. To build a portfolio model, however, investors had to quantify their risk
variable. The basic portfolio model was developed by Harry Markowitz, who derived the
expected rate of return for a portfolio of assets and an expected risk measure. Harry M.
Markowitz is credited with introducing new concepts of risk measurement and their application
to the selection of portfolios. He started with the idea of risk aversion of average investors and
their desire to maximize the expected return with the least risk. Markowitz model is thus a
theoretical framework for analysis of risk and return and their inter-relationships. He used the
statistical analysis for measurement of risk and mathematical programming for selection of assets
in a portfolio in an efficient manner. His framework led to the concept of efficient portfolios. An
efficient portfolio is expected to yield the highest return for a given level of risk or lowest risk
for a given level of return.

The Markowitz model is based on several assumptions regarding investor behavior:

1) Investors consider each investment alternative as being represented by a probability
distribution of expected returns over some holding period.
2) Investors maximize one-period expected utility, and their utility curves demonstrate
diminishing marginal utility of wealth.
3) Investors estimate the risk of the portfolio on the basis of the variability of expected
returns.
 4) Investors base decisions solely on expected return and risk, so their utility curves are a
function of expected return and the expected variance (or standard deviation) of returns
only.
5) For a given risk level, investors prefer higher returns to lower returns. Similarly, for a
given level of expected return, investors prefer less risk to more risk.

Undertheseassumptions,asingleassetorportfolioofassetsisconsideredtobeefficientif
nootherassetorportfolioofassetsoffershigherexpectedreturnwiththesame(orlower)risk,
orlowerriskwiththesame(orhigher)expectedreturn.Diversification of securities is one method by
which the above objectives can be secured. The unsystematic and company related risk can be
secured. The unsystematic and company related risk can be reduced by diversification into
various securities and assets whose variability is different and offsetting or put in different words
which are negatively correlated or not correlated at all.

6.4 Portfolio Expected Rates of Return

The expected rate of return for an individual investment is computed as shown in chapter two.
The expected return for an individual risky asset with the set of potential returns and an
assumption of equal probabilities used in the example would be 11 percent.
The expected rate of return for a portfolio of investments is simply the weighted average ofthe
expected rates of return for the individual investments in the portfolio. The weights are
theproportion of total value for the investment.The expected rate of return for a hypothetical
portfolio with four risky assets is shown below. Theeffect of adding or dropping any investment
from the portfolio would be easy to determinebecause you would use the new weights based on
value and the expected returns for each of theinvestments. This computation of the expected
return for the portfolio [E(Rport)] can be generalized as follows:
n
E ( R Por )=∑ W i E ( Ri )
t =1

Where:
Wi=thepercentoftheportfolioinasseti
E(Ri)=theexpectedrateofreturnforasseti

Computationoftheexpectedreturnforanindividualriskyasset
Probability Possible Rate of Return (Percent) Expected Return (Percent)
 0.25 0.08 0.020
0.25 0.10 0.025
0.25 0.12 0.030
0.25 0.14 0.035
E(R) = 0.110

Computationoftheexpectedreturnforaportfolioofriskyassets
Weight (Wi) Expected Security Return, E(Ri) Expected Portfolio Return [Wi*E(Ri)]
(Percent of Portfolio)
0.20 0.10 0.020
0.30 0.11 0.033
0.30 0.12 0.036
0.20 0.13 0.026
E(RPort) = 0.115

6.5 Variance (Standard Deviation) of Returns for a Portfolio

As noted, we will be using the variance or the standard deviation of returns as the measure of
risk (recall that the standard deviation is the square root of the variance). Here, after discussing
some other statistical concepts, we will consider the determination of the standard deviation for a
portfolio of investments. The variance, or standard deviation, is a measure of the variation of
possible rates of return, Ri, from the expected rate of return [E(Ri)]. Two basic concepts in
statistics, covariance and correlation, must be understood before we discuss the formula for the
variance of the rate of return for a portfolio.

Covariance of Returns: It is a measure of the degree to which two variables “move together”
relative to their individual mean values over time. In portfolio analysis, we usually are concerned
with the covariance of rates of return rather than prices or some other variable.A positive
covariance means that the rates of return for two investments tend to move in the same direction
relative to their individual means during the same time period. In contrast, a negative
covarianceindicates that the rates of return for two investments tend to move in different
directions relative to their means during specified time intervals over time. The magnitude of the
covariance depends on the variances of the individual return series, as well as on the relationship
between the series.The covariance statistic provides an absolute measure of how they moved
together over time. For two assets, iand j, the covariance of rates of return is defined as:

Cov ij=∑ {[ Ri −E( Ri ) ] [ R j −E( R j ) ] }

Example: Computation of Covariance of Returns for Coca-Cola and Home Depot: 2001

As can be seen, if the rates of return for one stock are above (below) its mean rate of return
during a given period and the returns for the other stock are likewise above (below) its mean rate
of return during this same period, then the product of these deviations from the mean is positive.
If this happens consistently, the covariance of returns between these two stocks will be some
large positive value. If, however, the rate of return for one of the securities is above its mean
return while the return on the other security is below its mean return, the product will be
negative. If this contrary movement happened consistently, the covariance between the rates of
return for the two stocks would be a large negative value.

According to the above monthly rates of return during 2001 for Coca-Cola and Home Depot one
might expect the returns for the two stocks to have reasonably low covariance because of the
differences in the products of these firms. The expected returns E(R) were the arithmetic mean of
the monthly returns:

Interpretation of a number such as 6.37 is difficult; is it high or low for covariance? We know the
relationship between the two stocks is generally positive, but it is not possible to be more
specific.

Correlation Coefficient of Returns: Covariance is affected by the variability of the two

individual return series. Therefore, a number such as the 6.37 in our example might indicate a
weak positive relationship if the two individual series were volatile but would reflect a strong
positive relationship if the two series were very stable. Obviously, you want to “standardize” this
covariance measure taking into consideration the variability of the two individual return series,
as follows:

Where:
rij=the correlation coefficient of returns
ri= the standard deviation of Rit
rj=the standard deviation of Rjt

Standardizing the covariance by the individual standard deviations yields the correlation
coefficient (rij), which can vary only in the range –1 to +1. A value of +1 would indicate a
perfect positive linear relationship between Ri and Rj, meaning the returns for the two stocks
movetogether in a completely linear manner. A value of –1 indicates a perfect negative
relationshipbetween the two return series such that when one stock’s rate of return is above its
mean, theother stock’s rate of return will be below its mean by the comparable amount.

To calculate this standardized measure of the relationship, you need to compute the standard
deviation for the two individual return series. We already have the values for Rit – E(Ri) and Rjt –
E(Rj) in last example. Then, we can square each of these values and sum them as shown in the
next exhibition to calculate the variance of each return series.

The standard deviation for each series is the square root of the variance for each, as follows:

Thus, based on the covariance between the two series and the individual standard deviations, we
can calculate the correlation coefficient between returns for Coca-Cola and Home Depot as
Obviously, thisformulaalsoimpliesthat
Covij=rijσiσj(.108) (5.80)(10.17)=6.37
As noted, a correlation of +1.0 would indicate perfect positive correlation, and a value of –1.0
would mean that the returns moved in a completely opposite direction. A value of zero would
mean that the returns had no linear relationship, that is, they were uncorrelated statistically. That
does not mean that they are independent. The value of rij = 0.108 is quite low. This relatively low
correlation is not unusual for stocks in diverse industries (i.e., beverages and building materials).
Correlation between stocks of companies within some industries approaches 0.85.

Portfolio Standard Deviation Formula Now that we have discussed the concepts of covariance
and correlation, we can consider the formula for computing the standard deviation of returns for
a portfolio of assets, our measure of risk for a portfolio. One might assume it is possible to derive
the standard deviation of the portfolio in the same manner, that is, by computing the weighted
average of the standard deviations for the individual assets. This would be a mistake. Markowitz
derived the general formula for the standard deviation of a portfolio as follows:

σPort = thestandarddeviationoftheportfolio
wi=theweightsoftheindividualassetsintheportfolio,whereweightsaredeterminedbythe
proportionofvalueintheportfolio
r2=thevarianceofrates of returnforassetsi
i

Covij=thecovariancebetweentheratesofreturnforassetsiandj,where Covij=rijrirj

This formula indicates that the standard deviation for a portfolio of assets is a function of the
weighted average of the individual variances (where the weights are squared), plus the weighted
covariances between all the assets in the portfolio. The standard deviation for a portfolio of
assets encompasses not only the variances of the individual assets but also includes the
covariances between pairs of individual assets in the portfolio. Further, it can be shown that, in a
portfolio with a large number of securities, this formula reduces to the sum of the weighted
covariances.

Although most of the subsequent demonstration will consider portfolios with only two assets
because it is possible to show the effect in two dimensions, we will demonstrate the
computations for a three-asset portfolio. Still, it is important at this point to consider what
happens in a large portfolio with many assets. Specifically, what happens to the portfolio’s
standard deviation when you add a new security to such a portfolio? As shown by the formula,
we see two effects. The first is the asset’s own variance of returns, and the second is the
covariance between the returns of this new asset and the returns of every other asset that is
already in the portfolio. The relative weight of these numerous covariances is substantially
greater than the asset’s unique variance; and the more assets in the portfolio, the more this is
true. This means that the important factor to consider when adding an investment to a portfolio
that contains a number of other investments is not the investment’s own variance but its average
covariance with all the otherinvestments in the portfolio.

In the following examples, we will consider the simple case of a two-asset portfolio. We do these
relatively simple calculations to demonstrate the impact of different covariances on the total risk
(standard deviation) of the portfolio.

6.6 Two-Asset Portfolio

Equal Risk and Return—Changing Correlations: consider first the case in which both assets
have the same expected return and expected standard deviation of return. As an example,let us
assume:

E (R1) = 0.20, σ1 = 0.10, E(R2) = 0.20, σ2 = 0.10

To show the effect of different covariances, assume different levels of correlation between the
two assets. Consider the following examples where the two assets have equal weights in the
portfolio (W1 = 0.50; W2 = 0.50). Therefore, the only value that changes in each example is the
correlation between the returns for the two assets.
Recall that: Covij=rijσi σj
Consider the following alternative correlation coefficients and the covariances they yield. The
covariance term in the equation will be equal to r1,2 (0.10)(0.10) because both standard deviations
are 0.10.

a. r1,2 = 1.00; Cov1,2 = (1.00)(0.10)(0.10) = 0.010

b. r1,2 = 0.50; Cov1,2 = (0.50)(0.10)(0.10) = 0.005
c. r1,2 = 0.00; Cov1,2 = 0.000(0.10)(0.10) = 0.000
d. r1,2 = –0.50; Cov1,2 = (–0.50)(0.10)(0.10) = –0.005
e. r1,2 = –1.00; Cov1,2 = (–1.00)(0.10)(0.10) =–0.01

Nowletusseewhathappenstothestandarddeviationoftheportfoliounderthesefiveconditions.
Recallthat:

When this general formula is applied to a two-asset portfolio, it is

In this case, where the returns for the two assets are perfectly positively correlated (r1,2 = 1.0), the
standard deviation for the portfolio is, in fact, the weighted average of the individual standard
deviations. The important point is that we get no real benefit from combining two assets that are
perfectly correlated; they are like one asset already because their returns move together.Now
consider Case b, where r1,2 equals 0.50:
The only term that changed from Case a is the last term, Cov 1,2, which changed from 0.01 to
0.005. As a result, the standard deviation of the portfolio declined by about 13 percent, from 0.10
to 0.0868. Note that the expected return did not change because it is simply the weighted average
of the individual expected returns; it is equal to 0.20 in both cases. You should be able to confirm
through your own calculations that the standard deviations for Portfolios c and d are as follows:

c. 0.0707 d. 0.05

The final case where the correlation between the two assets is –1.00 indicates the ultimate
benefits of diversification:

Here, the negative covariance term exactly offsets the individual variance terms, leaving an
overall standard deviation of the portfolio of zero. This would be a risk-free portfolio.

The following exhibition illustrates a graph of such a pattern. Perfect negative correlation gives a
mean combined return for the two securities over time equal to the mean for each of them, so the
returns for the portfolio show no variability. Any returns above and below the mean for each of
the assets are completely offset by the return for the other asset, so there is no variability in total
returns, that is, no risk, for the portfolio. This combination of two assets that are completely
negatively correlated provides the maximum benefits of diversification—it completely
eliminates risk.
Time Patterns of Returns for Two Assets with Perfect Negative Correlation:

The above graph shows the difference in the risk-return posture for these five cases. As noted,
the only effect of the change in correlation is the change in the standard deviation of this two-
asset portfolio. Combining assets that are not perfectly correlated does not affect the expected
return of the portfolio, but it does reduce the risk of the portfolio (as measured by its standard
deviation). When we eventually reach the ultimate combination of perfect negative correlation,
risk is eliminated.

6.7 Three-Asset Portfolio

A demonstration of what occurs with a three-asset class portfolio is useful because it shows the
dynamics of the portfolio process when we add additional assets to a portfolio. It also shows the
rapid growth in the computations required.

The formula for portfolio of containing three assets is as follow:

 p   A2 wA2   B2 wB2   C2 wC2  2wA wB  A, B A B  2wB wC  B ,C B C  2wA wC  A,C A C

Let’s see the portfolio of three asset classes including of stock, bond and cash equivalent:
Thecorrelationsareasfollows:

Given the weights specified, the E(Rp) is:

When we apply the generalized formula to the expected standard deviation of a three-asset class,
it is as follows:

It is important to keep in mind that the results of this portfolio asset allocation depend on the
accuracy of the statistical inputs. In the current instance, this means that for every asset (or asset
class) being considered for inclusion in the portfolio, you must estimate its expected returns and
standard deviation. In addition, the correlation coefficient among the entire set of assets must
also be estimated. The number of correlation estimates can be significant—for example, for a
portfolio of 100 securities, the number is 4,950 (that is, 99 + 98 + 97 + . . .). The potential source
of error that arises from these approximations is referred to as estimationrisk.

6.8 Efficient Frontier

If we examined different two-asset combinations and derived the curves assuming all the
possibleweights, we would have a graph like that in exhibition. The envelope curve that contains
the best of all these possible combinations is referred to as the efficient frontier. Specifically,the
efficient frontier represents that set of portfolios that has the maximum rate of return forevery
given level of risk, or the minimum risk for every level of return. Every portfolio that lies on the
efficient frontier has either ahigher rate of return for equal risk or lower risk for an equal rate of
return than some portfoliobeneath the frontier. Thus, we would say that Portfolio A in exhibit
dominates Portfolio C because it has an equal rate of return but substantially less risk. Similarly,
Portfolio B dominates Portfolio C because it has equal risk but a higher expected rate of return.
Because of the benefits of diversification among imperfectly correlated assets, we would expect
the efficient frontier to be made up of portfolios of investments rather than individual securities.
Two possible exceptions arise at the end points, which represent the asset with the highest return
and that asset with the lowest risk.

Numerous Portfolio Combinations of Available Assets

Efficient Frontier for Alternative Portfolios

As an investor, you will target a point along the efficient frontier based on your utility function
and your attitude toward risk. No portfolio on the efficient frontier can dominate any other
portfolio on the efficient frontier. All of these portfolios have different return and risk measures,
with expected rates of return that increase with higher risk.

6.9 Efficient Frontier and Investor Utility

The curve above shows that the slope of the efficient frontier curve decreases steadily as you
move upward. This implies that adding equal increments of risk as you move up the efficient
frontier gives you diminishing increments of expected return.

Selecting an Optimal Risky Portfolio on the Efficient Frontier

An individual investor’s utility curves specify the trade-offs he or she is willing to make between
expected return and risk. In conjunction with the efficient frontier, these utility curves determine
which particular portfolio on the efficient frontier best suits an individual investor. Two
investors will choose the same portfolio from the efficient set only if their utility curves are
identical.

The above curve shows two sets of utility curves along with an efficient frontier of investments.
The curves labeled U1 are for a strongly risk-averse investor (with U3 U2 U1). These utility
curves are quite steep, indicating that the investor will not tolerate much additional risk to obtain
additional returns. The investor is equally disposed toward any E(R), σ combinations along a
specific utility curve, such as U1.The curves labeled U1′ (U3′ U2′ U1′) characterize a less-risk-
averse investor. Such an investor is willing to tolerate a bit more risk to get a higher expected
return. The optimal portfolio is the portfolio on the efficient frontier that has the highest utility
for a given investor. It lies at the point of tangency between the efficient frontier and the curve
withthe highest possible utility. A conservative investor’s highest utility is at point X in the above
curve where the curve U2 just touches the efficient frontier. A less-risk-averse investor’s highest
utility occurs at point Y, which represents a portfolio with a higher expected return and higher
risk than the portfolio at X.