Sources:

- “Corporate Finance”, 4th Global Edition, Jonathan Berk & Peter DeMarzo
- “Financial Economics”, 2nd Edition, Zvi Bodie, Robert C. Merton, David L. Cleeton
- “Quantitative Financial Economics: Stocks, Bonds & Foreign Exchange”, 2nd Edition, Keith Cuthbertson & Dirk Nitzsche
- “Investments”, 10th Edition, Zvi Bodie, Alex Kane, Alan J. Marcus
Simple (discrete-time estimation) and Log-returns
(continuous-time estimation).

The simple average is the actual return over n

observations, most of the times multiplied by 100 to
generate the %return.

The Log-returns do not follow the aforementioned

method of “mean” estimation
Investment management is easier when returns are
normal.
 Standard deviation is a good measure of risk when returns are
symmetric.
 If security returns are symmetric, portfolio returns will be, too.
 Future scenarios can be estimated using only the mean and the
standard deviation.
 Skewness Kurtosis

 _ 3
  _ 4
 R − R   
 R − R  
skew = average  ^   kurtosis = average  ^   − 3
   
 σ  σ
3 4
 
   
If return distributions are not asymmetric, then:

 We should look at negative outcomes separately

 Because an alternative to a risky investment is a risk-free
investment, we should look at deviations of returns from the
risk-free rate rather than from the sample average
 Fat tails should be accounted for
If an investor buys an asset, a, at time 0 for $P0, receives a
cash-flow at time 1 of $D1 and sells it at time 1 for $P1, the
simple return on the asset is defined as Ra,1 = (P1 + D1 – P0)/P0
At time 0, the actual value of Ra,1 is unknown.
Suppose there are 3 different possible outcomes, corresponding
to three different states of economy: states 1, 2 and 3.
Ra,1 could be equal to 3%, 7% or 10% with probabilities of 0.3,
0.2 and 0.5 for the three different states.
The mean or expected value of Ra,1 is computed as 0.3(3%) +
0.2(7%) + 0.5(10%) = 7.3%
The expected value is a kind of average outcome, where the
different possible outcomes are weighted by the
probabilities.
E(Ra) is the way we represent the expected return on asset a.
Here is the general formula for expected return, where pR is
the probability of a return R.
An asset has three possible return outcomes over the
next year: 30% (probability 0.3), 7% (probability
0.2), and 10% (probability 0.5). The expected return
is:
 8.2%
 6.6%
 You can’t compute the expected return
 The expected return doesn’t depend on the possible outcomes.
It is what each person wants to get from that investment.
Consider another asset, b, which has returns of
1%, 7% and 11.2% in the three different states.
The expected return of asset b is 0.3(1%) +
0.2(7%) + 0.5(11.20%), which is also 7.3%.
However, asset b’s returns are said to be more
volatile than those of asset a, meaning roughly
that the range of possible outcomes is greater, or
that more extreme outcomes – both positive and
negative – are more likely.
The variance of returns is computed as the mean squared
deviation from the expected return.
The squared deviation is a measure of how extreme a return is.
Taking the mean computes the average value of these
deviations.
The square root of the variance is called the standard deviation
or volatility.
Variance is measured in percent-squared.
Standard deviation is measured in the same units as returns,
viz. percent.
 Probabilities Return (Asset a) Deviation from Squared
Mean Return Deviation
0.3 3 -4.3 18.49
0.2 7 -0.3 0.09
0.5 10 2.7 7.29

We see that the variance of returns for asset a is 9.21, computed

as 0.3(18.49) + 0.2(0.09) + 0.5(7.29) and the standard
deviation is 3.035%
Probabilities Return (Asset b) Deviation from Squared
Mean Return Deviation

0.3 1 -6.3 39.69

0.2 7 -0.3 0.09
0.5 11.2 3.9 15.21

o The variance of returns for asset b can be computed as 39.69

(0.3) + 0.09 (0.2) + 15.21 (0.5) = 19.53; the standard deviation
is the square root of 19.53, i.e. 4.419%.

o We see - as expected – that asset b, which has more extreme

outcomes, also has a higher volatility.
σa is used to represent the standard deviation of returns
on asset a, while Var(Ra) or σa2 is used to represent the
variance of returns on asset a.
The probability distributions that we have considered
only have three possible outcomes; actual return
distributions may have many, even an infinite number 0f
outcomes.
One particular distribution with an infinite number of
outcomes is called a normal distribution. If we order
observations by value, group them into categories and
plot a frequency distribution, the resulting plot looks like
a bell curve as we make the categories smaller and
smaller.
The probability curve is comparable to a figure you might
draw for a discrete distribution with the values of the
random variable on the X-axis and the probabilities on
the Y-axis.
The height of the normal distribution probability curve at
a specific X-point can be thought of roughly as
proportional to the probability that the random variable
will take a value in the neighborhood of that point.
For any range of values on the X-axis, the associated
probability is found by drawing a vertical line
corresponding to the two end-points of the range. The
area under the curve between those two lines is the
required probability.
The total area under the curve is, thus, one.
Up to now, we started out with probability distributions
and used them to compute expected returns and
standard deviations.
Where do we get these probability distributions?
We might use them to represent our subjective
evaluations of what we think will happen in the future.
This might work well for returns on new projects or
assets without any price history.
However, if we have assets with a price history, we would
like to use this price history to inform our assumptions
regarding the probability distributions of their returns.
We already know how returns are computed:

If we have quarterly returns, we can combine them to get

annual returns, as follows:

However, we can combine returns from periods of

different lengths as well. We can see this in the following
example.
Suppose we start out with the following data:
We can use the formulas to obtain returns for each of
the periods. Thus the return for the period between
12/31/98 and 2/2/99, e.g., is computed as the sum of
the dividend (0.5) and the end of period price
(89.44) divided by the price at 12/31/98, $71.56:

The returns for the different periods are combined to

get the total return for 1999 and 2004 as:
Here are some return data for different assets and asset classes:
We can use this data to compute average returns using the
following formula, where T is the number of years of data:

Using this formula, we can compute the average return over

the nine years as 11.39% for the S&P 500 investment and 3.7%
for the 3-month T-bill investment (Table before).
If we now wanted to use this information to come up with an
estimate for the expected value of the underlying probability
distribution of returns on the S&P 500 investment and the 3-
month T-bills, we would use exactly these values as estimates.
Thus: if I believe that the return over the next year on the S&P
500 came from the same probability distribution as over the
nine years of my data, my best estimate of this return would
be 11.39%.
How do we use this data to come up with an estimate of
the variance and standard deviation of the underlying
probability distribution of returns on my asset classes?
The formula for this variance estimate is given by:

The estimated standard deviation is simply the square

root of this estimated variance.
For our data, we would then compute our estimate of the
underlying variance and standard deviation as:
Again, if we believe that the underlying probability
distribution that generated the returns in the years 1996-
2004 is the same as the distribution from which returns
will be generated in the future, we will have been able to
establish information regarding future returns:
 Our estimate of the expected value of future annual returns is 11.39%
and the standard deviation is 20.6% p.a.
Even though the average annual return is estimated to be
11.39%, we also know that in any given year, the actual
return could be quite different. Well, how different could
it be?
The standard deviation turns out to be useful to answer
this question.
We can answer this question by making two
assumptions:
 One, the different returns that we are using to compute our
average are independent or unrelated – that is, there is no
pattern across the observations. For example, it cannot be the
case that high return years are followed by high (or low) return
years.
 Two, the plot of returns looks bell-shaped or normal.
If we can make these assumptions, then statistical
theory allows us to say that future returns will be
within a specific range (or confidence interval)
approximately 95% of the time.
This confidence interval is computed as the expected
return estimate ± ~2.086*the standard deviation (s.d.).
Let’s round it up, to 2 (for a 2-sided 95% H0)
For the S&P 500, the s.d. is 20.6% and our confidence
interval is 11.39 ± ~2(20.6), which works out to
(-29.81, 52.59). Clearly, this is not very precise!
Fortunately, we can be much more certain regarding the
average return over a number of years.
The confidence interval for the average return over T years
is computed as the expected return estimate ± twice the
s.d. of the average return, which is the s.d. of the
individual returns divided by √T. The standard deviation
of the average is also called the standard error (s.e.).
For the S&P 500, if we hold it for say 20 years, the s.e.
is 20.6/√20 = 4.61% and our confidence interval is
11.39 ± 2(4.61), which works out to (2.18,20.6).
The same computation can be done for our confidence
regarding the underlying expected return over the
sample period 1996-2004. The confidence interval for
the expected return over that period is 11.39 ±
2(20.6/√9) or (-2.343,25.123).
As we have more data to estimate the expected return,
our confidence increases. However, as we use longer
periods of data, we are less confident that the future
will be similar to the past!
Using historical data, if we compute average returns and
volatility (standard deviation) of returns for different asset
classes over 1926-2004, we find a relationship:

This might lead us to

believe that average
returns are related to
standard deviation.
However, before we conclude that the expected return for an
asset (GM stock) or an asset class (S&P 500) is positively
related to its return volatility, look at the picture below
The figure includes data on 500 individual stocks ranked by
size and updated quarterly over the same period (1926-2004)
Why is there a difference between asset classes and single
assets? Can we come up with a single theory to explain
both?
To do this, we need to realize that the numbers defining
portfolios are not simply the average of the numbers for
the assets making up the portfolio.
In fact, when we put together single assets in a portfolio,
the volatility of the portfolio diminishes, the greater is
the number of assets in the portfolio.
The simple return (but not the Log-returns!) on a
portfolio of assets is indeed the average return on the
different assets making up the portfolio.
 Thus, if we have a portfolio invested equally over 3 assets earning
2%, 4% and 9% respectively, the return on the portfolio is (2+4+9)/3
= 5%
With respect to standard deviation, the answer is different.
We said earlier, that if the different observations used to compute an
average are independent, then the standard deviation for the average,
s.e. equals (s.d.)/√n, where n is the number of observations and s.d. is
the standard deviation of the observations.
So if the return on a portfolio is an average, can we say that if a
portfolio consists of 100 assets, the standard deviation of the returns
on the portfolio is one-tenth of the standard deviation of the returns
on an individual asset?
No, because while returns on an asset over different periods are
mostly uncorrelated, the returns on the different assets in a portfolio
are correlated.
In fact, we know that the prices of most assets in an economy move in
the same direction as the overall market! How then can we say that
they are unrelated? We can’t.
We will use the concept of diversification.
The risk that the main factory of a company will be
shut down due to a tornado.

 Idiosyncratic Risk
 Systematic Risk
The risk that expansion into the Chinese market next
year will not materialize for a firm, whose stock you
are holding.

 Idiosyncratic
Risk
 Market Risk
For individual assets, the relevant volatility is that part
of their volatility that cannot be diversified (it’s not bi
times the volatility of the market portfolio).
An asset’s bi is called beta (written β) and its average
return in a market where assets are properly priced will
be proportional to its beta risk.
If we plot the betas of single assets against their
average returns, they should form a straight line
The model that describes this relationship is called the
Capital Asset Pricing Model – the CAPM.
The mimicking factor of an asset is its bi (beta)
(recall Ri = ai + biRm + ei).
Hence the average return on a stock is β∗E(Rm) + (1-
β)*rf. This can be rewritten as rf + β[E(Rm)-rf], where
E(Rm)-rf is the market risk premium.
The required rate of return on a portfolio, therefore
is also related to its beta risk and equals rf +
βpr[E(Rm)-rf], where βpr is the beta of the portfolio.
We just noted that an asset’s beta is simply its
market mimicking factor.
Hence we can estimate a stock’s beta as the slope
coefficient in a regression of the stock return on the
market portfolio return because the slope coefficient
in a regression is the extent to which the LHS
variable changes when the RHS variable changes.
Any uncertainty in an asset’s return that is market
related is called non-diversifiable risk and all other
uncertainty is called diversifiable risk and, as we
have already seen, is irrelevant for pricing purposes.
Cisco’s returns tend to move in the same direction, but with greater
amplitude, than those of the S&P 500.
Beta corresponds to
the slope of the
best-fitting line.
Beta measures the
expected change in
Cisco’s excess
return per 1%
change in the
market’s excess
return.

Deviations from the

best-fitting line
correspond to
diversifiable, non-
market-related risk.
Betas can be estimated by regressing the historical return on
the stock against the historical return on the market portfolio,
which is often proxied by an index e.g., the S&P 500.
Further, if the beta of a firm measures the likelihood of a firm
to move with the market, then betas must be similar for all
firms that are similar, e.g. by being in a given industry.
This is what we see in the next table. The left-hand side
column shows betas for sector/industries, while the right-
hand side shows betas for firms.
Note that consumer goods firms like Anheuser-Busch, Heinz
and Coca-Cola have similar betas – and they are all less than
one.
Note, however, that these numbers are only estimates and
have their own standard error or confidence intervals.
According to the CAPM, a lottery ticket has a beta of:

 Zero because it’s return is unrelated to the market portfolio.

 A beta greater than zero because it is risky.
 A negative beta because you can use it to diversify.
Individuals have differing preferences for risk. A utility
function is a convenient way of describing individual’s
preferences for risk.
In economics, a utility function is described over
quantities of goods. That works when we know the
consumption of a good with precision.
However, if we are to describe preferences over uncertain
quantities of goods, then we use a concept invented by
von Neumann, called Expected Utility, which can be used
provided individuals’ preferences satisfy some basic
conditions, such as transitivity, i.e. if an individual
prefers a choice A to B and B to C, then he must prefer A
to C.
A commonly used utility function is U(r) = E(r) – 0.005Aσ2(r),
where returns are specified in percent per annum.
The constant risk aversion parameter, is called A; higher
values of A represent a preference for less risk and r is
measured in percentages.
The utility score can also be used as a certainty equivalent
rate of return.
That is, if the investor is indifferent between two assets or
portfolios, one with an uncertain return of r0 and another
with a certain return of r1, then: U(r0)=E(r0) – 0.005Aσ2(r0) =
U(r1) = E(r1) = r1
Thus, if an asset/portfolio has an expected return of 15% and
a std. dev. of 5% and A=3 for a given individual, then she
would be indifferent between that asset and another one
with a certain return of 15-(0.005)3(5)2, or 14.625%
The utility function U is a good way of describing investors’
preferences under two conditions – one that return
distributions are normal (i.e. that only mean and standard
deviation of returns are necessary to describe the distribution)
– or that investor preferences are quadratic.
While some individuals might have approximately quadratic
utility functions, it still represents a strong restriction on
individual preferences.
We know now (particularly post the 2008 crisis) that return
distributions probably are not normal. Hence we must be
careful in using our utility function.
However, it still represents a very powerful way of describing
risk preferences and we can derive a lot of useful results with
that assumption.
If we decide that a mean-variance utility function is appropriate, then
what we need for this is to figure out the investor’s A-number.
One way to do this is by using questionnaires, asking investors which
one of a pair of investment choices they would prefer.
This can then be used to construct indifference curves and finally these
indifference curves can be used to figure out the investor’s A value.
Alternatively, we could observe individuals’ decisions when confronted
with risk.
Finally, we could observe how much people are willing to pay to avoid
risk.
Risk free rate of return :
rate of return which can be earned with certainty (i.e σ = 0), is the
3 months T-bill
Risk premium :
expected return in excess of the risk free rate i.e. (ERp – rf)
Risk aversion :
 measures the reluctance by investors to accept (more) risk
 ‘High number’ : risk averse
 ‘Low number’ : less risk averse
Example : ERp - rf = 0.005 A σ2p
with A = (ERp - rf) / (0.005 σ2p)
 Indifference curve

Asset Q

ERp
Asset P

σp σ
Suppose we have a random variable representing an
asset, a wealth / return for a period Wi , with ‘n’
possible outcomes Wi , each one assigned a
probability pi
The utility from any wealth outcome is denoted as
U(Wi). The expected utility is:
E[U(W)] = ΣpiU(Wi)
 Investment A Investment B Investment C
Outcome Prob Outcome Prob Oucome Prob

20 3/15 19 1/5 18 ¼

18 5/15 10 2/5 16 ¼

14 4/15 5 2/5 12 ¼

10 2/15 8 ¼

6 1/15
Assume the following utility function :
U(W) = 4W – (1/10) W2

 If the outcome is 20, U(W) = 80 – (1/10) 400 = 40

 …

 Expected Utility
Investment A : E(UA) = … = 36.3
Investment B : E(UB) = … = 26.98
Investment C :
E(UC) = 39.6(1/4) + 38.4(1/4) + 33.6(1/4) + 25.6(1/4) = 34.3
 45

40

35

30

25
U(W)

20

15

10

0
0 5 10 15 20 25

W
Ranking of investment remains unchanged if
 a constant is added to the utility function
 the utility function is scaled by a constant

Example :
a + b*U(W) gives the same ranking as U(W)
A fair lottery is defined as one that has an expected value
of zero.
“Risk aversion” implies that an individual would not
accept a ‘fair lottery’.
Example :
 tossing a coin with $1 for WIN (heads) and -$1 for LOSS (tails).
x = k1 with probability p
x = k2 with probability 1-p
E(x) = pk1 + (1-p)k2 = 0
k1/k2 = -(1-p)/p or p = -k2/(k1-k2)
 Tossing a coin : p = ½ and k1 = -k2 = $ 1
Example :

 Tossing a (fair) coin (i.e. p = 0.5 for head)

 gamble of receiving £ 16 for a ‘head’ and £ 4 for ‘tails’
 EW = 0.5 (£ 16) + 0.5 (£ 4) = £ 10
 If the cost of ‘gamble’ is = £ 10, then  EW – Cost = 0
 How much is an individual willing to pay for playing the
game ?
Assume the following utility function

U(W) = W1/2

Expected return from gamble is:

E[U(W)] = 0.5 U(WHead) + 0.5 U(WTails)
= 0.5 (16)1/2 + 0.5(4)1/2 = 3
 Utility
U(16) = 4
BE point U(W)= W1/2
U(EW) = 101/2 =3.162 x
π
E[U(W)] = 3 A
=0.5 (16)1/2 + 0.5(4)1/2

U(4) = 2

0
4 EW=10 16 Wealth
(W–π) = 9
Utility U(W) Risk Neutral
Risk Averter

U(16)
U(10)

U(4) Risk Lover

4 10 16 Wealth
 Risk Averter
Expected Return

Risk Lover

Risk Neutral

Risk, σ
U(W) = W(1-γ) / (1-γ) γ > 0, γ ≠ 1 (power)
As γ  1, logarithmic utility is a limiting case of power utility i.e.,
U(W) = ln(W)
U(W) = W – (b/2)W2 b > 0 (quadratic)
U(W) = a – be-cW c > 0 (negative exponential)
Investors maximise expected utility at the end-of-period
wealth
It is proven that the above principle implies maximising a
function of expected asset/portfolio returns and variance
providing
 Either the utility is quadratic, or
 Asset/Portfolio returns are normally distributed (and the utility is
concave)

The wealth vis-à-vis the returns are: W = W0(1 + Rp)

So, the utility is: U(W) = U[W0(1 + Rp)]
Expanding U(Rp) in Taylor series around mean of Rp (=µp)
U(Rp) = U(µp) + (Rp – µp) U’(µp)
+ (1/2)(Rp – µp)2 U’’(µp)
+ higher order terms

And taking expectations

E[U(Rp)] = U(µp) + (1/2) σ2p U’’(µp) +E(higher-terms)

 Hence, the E[U(Rp)] is only a function of the mean and

of the variance
Assuming we know the mean and variance of the risky part of
the investor’s asset (or portfolio of assets), we compute the
proportion of the it that would go into the risky part and the
riskless part, respectively.
 Let y=portion allocated to the risky part, P
 (1-y)=portion to be invested in risk-free asset, F

Assume the following data:

rf = 7% σrf = 0%
E(rp) = 15% σp = 22%
The investor must choose one optimal asset (or portfolio) C, from the set of
feasible choices, for different values of y, knowing that the asset expected
return is
E ( rc ) = rf + y  E ( rP ) − rf 

And its variance: σ C2 = y 2σ P2

Now let’s assume another utility function e.g., U(r) = E(r) – 0.01Aσ2(r)

Substituting these expressions into the utility function and maximizing it

with respect to “y” (partial derivation finding the maximum), we find:

In our example, this works out to be: (15-7)/0.01(4*222)= 41.32%, for A=4
Note that there is an optimal value of y, which can be seen
graphically in the next slide.
We can also examine the optimal
choice of asset/portfolio in the (E(r), σ)
plane.

The indifference curves are straight

lines in the (E(r), σ2) plane, and convex
in the (E(r), σ) plane.

But, the total depiction is 3D, in (y,

E(r), σ2) or (y, E(r), σ) space, so the 2D
we see, is a “projection” of y-planes
cutting off through the 2D plane

We first view the shape of the utility

function in this plane and then
superimpose it upon the investment
opportunity set, viz. the CAL.
Assume from utility theory, that an investor uses a quadratic function e.g.,
E(Rm) - rf = (A/0.85) σ2m
The main implication of the CAPM is that the market is mean-variance efficient, and
therefore that the expected-return beta relationship holds for the market portfolio.
But, the true market portfolio is not identifiable (because it includes all risky assets).
Also, even if we accept the validity of the linear risk-return relationship in our empirical
tests, we may not have identified a true ex-ante efficient portfolio that can be used for
expected return computations.
Practically, empirical tests of the CAPM use diversified portfolios of stocks as proxies
for the market portfolio.
Still, empirically we find that if excess returns on stocks are regressed on the excess
market return, the intercept is significantly higher than the zero value predicted by the
CAPM!
Usually, the alpha (or the beta-adjusted excess return) is positive for low-beta securities
and negative for high-beta securities.
So if the linear expected return-beta relationship does not hold up, is it still true that the
observed market portfolio is ex-ante mean-variance efficient?
Since most mutual funds are usually not able to outperform the expected return based
on the observed market portfolio (e.g., S&P500), we can argue that the observed market
portfolio is mean-variance efficient, and use the CAPM with respect to the observed
market portfolio.
We first define portfolio weights
 The fraction, xi of the total investment in the portfolio held in
investment ‘i’; the portfolio weights must add up to 1.00 or
100%.
Value of investment i
xi =
Total value of portfolio
Then the return on a portfolio, Rp , is the weighted
average of the (simple) returns, Ri, on the investments
in the portfolio, where the weights correspond to
portfolio weights.
RP = x1 R1 + x2 R2 + ⋯ + xn Rn =  xRi i i
 The expected return of a portfolio is the weighted average
of the expected returns of the investments within it.

E [ RP ] = E   i xi Ri  =  E[x R ]
i i i =  x E [R ]
i i i

The next step is to see how forming a portfolio affects the

volatility of returns.
Combining Risks
Table: Returns for Three Stocks, and Portfolios of Pairs of
Stocks
Covariance
 The expected product of the deviations of two returns from their means
 Covariance between Returns Ri and Rj

Cov(Ri ,R j ) = E[(Ri − E[ Ri ]) (R j − E[ R j ])]

 Estimate of the Covariance from Historical Data
1
Cov (Ri ,R j ) =
T − 2
 t
(Ri ,t − Ri ) (R j ,t − R j )
If the covariance is positive, the two returns tend to move together.
If the covariance is negative, the two returns tend to move in opposite directions.
Correlation
 A measure of the common risk shared by stocks that does not
depend on their volatility, and importantly is metric-free

Cov(Ri ,R j )
Corr (Ri ,R j ) =
SD(Ri ) SD(R j )
 The correlation between two stocks will always be between –1
and +1.
 For a two-security portfolio:

Cov(P1 ,P2 ) = Cov(x1 R1 + x2 R2 , x1 R1 + x2 R2 )

= x1 x1Cov(R1 ,R1 ) + x1 x2Cov(R1 ,R2 ) + x2 x1Cov(R2 ,R1 ) + x2 x2Cov(R2 ,R2 )

 The Variance of a 2-Stock Portfolio

Var (RP ) = x12Var (R1 ) + x22Var (R2 ) + 2 x1 x2Cov(R1 ,R2 )

Equally Weighted Portfolio
 A portfolio in which the same amount is invested in each stock

Variance of an Equally Weighted Portfolio of n

Stocks

1
Var ( RP ) = (Average Variance of the Individual Stocks)
n
 1
+ 1 −  (Average Covariance between the Stocks)
 n
For a portfolio with arbitrary weights, the standard deviation is
calculated as:
Security i’s contribution to the
volatility of the portfolio

SD( RP ) =  i
xi × SD( Ri ) × Corr ( Ri ,R p )
↑ ↑ ↑
Amount Total Fraction of i’s
of i held Risk of i risk that is
common to P

Unless all of the stocks in a portfolio have a perfect positive correlation

of +1 with one another, the risk of the portfolio will be lower than the
weighted average volatility of the individual stocks:

SD( RP ) =  x SD( R ) Corr (R ,R

i i i i p ) <  x SD( R )
i i i
Efficient Portfolios with Two Stocks
 Consider a portfolio of Intel and Coca-Cola

Table: Expected Returns and Volatility for Different Portfolios of Two

Stocks
Efficient Portfolios with Two Stocks
 Consider investing 100% in Coca-Cola stock. As shown in on
the previous slide, other portfolios—such as the portfolio with
20% in Intel stock and 80% in Coca-Cola stock—make the
investor better off in two ways:
 It has a higher expected return
 It has lower volatility
 As a result, investing solely in Coca-Cola stock is inefficient.
Correlation has no effect on the expected return of a
portfolio. However, the volatility of the portfolio will differ
depending on the correlation.
The lower the correlation, the lower the volatility we can
obtain. As the correlation decreases, the volatility of the
portfolio falls.
The curve showing the portfolios will bend to the left to a
greater degree.
SD(RP ) = x12Var (R1 ) + x22Var (R2 ) + 2 x1 x2ρ12SD ( R1 ) SD(R2 )
We now see what happens if we allow short positions. What
is a short position and a long position?
Long Position
 A positive investment in a security (buy)

Short Position
 A negative investment in a security (sell)
 In a “naked” short sale, you sell a stock that you do not own and then
buy that stock back in the future.
 Short selling is an advantageous strategy if you expect a stock price to
decline in the future.
vs.
Consider adding Bore Industries to the 2-stock portfolio:

Although Bore has a lower return and the same volatility as Coca-
Cola, it still may be beneficial to add Bore to the portfolio for the
diversification benefits.
We will see what happens if we invest in combinations of Bore and a
50-50 portfolio invested in Intel and Coke
And finally, we see what happens if can vary the proportions for all 3
stocks.
The efficient portfolios, those offering the highest
possible expected return for a given level of
volatility, are those on the northwest edge of the
shaded region, which is called the efficient frontier
for these three stocks.
In this case none of the stocks, on its own, is on the
efficient frontier, so it would not be efficient to put all
our money in a single stock.
Risk can also be reduced by investing a portion of a portfolio
in a risk-free investment, like T-Bills. However, doing so will
likely reduce the expected return.
On the other hand, an aggressive investor who is seeking
high expected returns might decide to borrow money to
invest even more in the stock market.
Consider an arbitrary risky portfolio and the effect on risk
and return of putting a fraction of the money in the portfolio,
while leaving the remaining fraction in risk-free Treasury
bills.
The expected return would be:
E [RxP ] = (1 − x)rf + xE[RP ] = rf + x (E[RP ] − rf )
 The standard deviation of the portfolio would then be
calculated as:

SD[RxP ] = (1 − x) 2Var (rf ) + x 2Var (RP ) + 2(1 − x)xCov(rf ,RP )

= x 2Var (RP ) Zero (0)

= xSD(RP )
 Note: The standard deviation is only a fraction of the volatility
of the risky portfolio, based on the amount invested in the risky
portfolio.
To go past point P, it is necessary to buy stocks on margin.
What is buying on margin? It is borrowing money to
invest in a stock.
It is similar to short-selling however, in this case, we’re not
short-selling the stock. Rather, we’re short-selling a risk-
free security.
A portfolio that consists of a short position in the risk-free
investment is known as a levered portfolio. Margin
investing is a risky investment strategy.
Note, however, that portfolio P is not the best portfolio to
use for our margin strategy.
To earn the highest possible expected return for any level
of volatility we must find the portfolio that generates the
steepest possible line when combined with the risk-free
investment.
The tangent portfolio provides the best risk and return trade-off
available to an investor.
 This means that the tangent portfolio is efficient and that all efficient portfolios are
combinations of the risk-free investment and the tangent portfolio.
 Every investor should invest in the tangent portfolio independent of her taste for
risk (risk-appetite).

An investor’s preferences will determine only how much to invest

in the tangent portfolio versus the risk-free investment.
 Conservative investors will invest a small amount in the tangent portfolio.
 Aggressive investors will invest more in the tangent portfolio.
 Both types of investors will choose to hold the same portfolio of risky assets, i.e.,
the tangent portfolio which is the efficient portfolio.
The Capital Asset Pricing Model (CAPM) allows us to
identify the efficient portfolio of risky assets without
having any knowledge of the expected return of
each security.
The CAPM uses the optimal choices investors make
to identify the efficient portfolio as the market
portfolio, the portfolio of all stocks and securities in
the market.
Three Main Assumptions
 Assumption 1
Investors can buy and sell all securities at competitive market prices
(without incurring taxes or transactions costs) and can borrow and lend
at the risk-free interest rate.

 Assumption 2
Investors hold only efficient portfolios of traded securities—portfolios
that yield the maximum expected return for a given level of volatility.
 Assumption 3
Investors have homogeneous expectations regarding the volatilities,
correlations, and expected returns of securities (EMH).
Homogeneous Expectations means that:
 All investors have the same estimates concerning future investments
and returns.
When the CAPM assumptions hold, an optimal
portfolio is a combination of the risk-free investment
and the market portfolio.
 When the tangent line goes through the market portfolio, it is
called the portfolio capital market line (CML).

The expected return and volatility of a capital market

line CML-portfolio are:
E [RxCML ] = (1 − x)rf + xE[RMkt ] = rf + x(E [RMkt ] − rf )

SD(RxCML ) = xSD(RMkt )
Market Risk and Beta
 Given the efficient market portfolio (Mkt=Eff), the expected
return of any asset is:

E[Ri ] = ri = rf + βiMkt (E[RMkt ] − rf )

Risk premium for security i

 The βeta is:

Volatility of i that is common with the market

SD(Ri ) × Corr (Ri ,RMkt ) Cov(Ri ,RMkt )

β i
Mkt
≡ βi = =
SD(RMkt ) Var (RMkt )
There is a linear relationship between a stock’s beta
and its expected return.
The portfolio security market line (SML) is graphed
as the line through the risk-free investment and the
market.
According to the CAPM, if the expected return and
beta for individual securities are plotted, they should
all fall along the SML.
PANEL (a)

The CML depicts

portfolios combining the
risk-free investment and
the efficient portfolio,
and shows the highest
expected return that we
can attain for each level
of volatility.
According to the CAPM,
the market portfolio is
on the CML and all
other stocks and
portfolios contain
diversifiable risk and lie
to the right of the CML,
as illustrated for Exxon
Mobil (XOM).
PANEL (b)

The SML shows the

expected return for each
security as a function of
its beta with the market.
According to the CAPM,
the market portfolio is
efficient, so all stocks
and portfolios should lie
on the SML.
Beta of a Portfolio

 The beta of a portfolio is the weighted average of the betas of

the assets in the portfolio.

Cov(RP ,RMkt ) (
Cov  i xi Ri ,RMkt ) Cov(Ri ,RMkt )
βP =
Var (RMkt )
=
Var (RMkt )
=  i xi Var (RMkt )
=  xβ
i i i
Let us look at the simplest case of putting two arbitrarily
correlated assets in a portfolio.
Let us call the two assets, a bond, D, and a stock (equity), E.
Then, we can write out the following relationship:

rp = wr
D D
+ wEr E
rP = Portfolio Return
wD = Bond Weight
rD = Bond Return
wE = Equity Weight
rE = Equity Return
We know the relationship:

σ p2 = wD2 σ D2 + wE2 σ E2 + 2wD wE Cov (rD , rE )

where Cov(rD, rE) represents the covariance between the

returns on assets D and E.
If we use ρDE to represent the correlation coefficient
between the returns on the two assets, then
Cov(rD,rE) = ρDEσDσE
The formula for portfolio variance can be written either
with covariance or with correlation.
Given that wD + wE =1 we have
σ p2 = wD2 σ D2 + (1 − wD ) 2 σ E2 + 2 wD (1 − wD )Cov ( rD , rE )
The minimum variance portfolio of risky assets D, E can be found by
taking the 1st derivative of the portfolio variance with respect to wD and
setting it equal to zero:
dσ p2
=0
dwD
Solving it gives the following formula:
σ E2 − Cov(rD , rE )
wmin ( D) = 2 ; wmin ( E ) = 1 − wmin ( D)
σ D − σ E − 2Cov(rD , rE )
2

Similarly, the formula for the tangency portfolio can be shown to be:
 Advanced portfolio optimization
 If a portfolio comprises >2 assets with various assigned weights,
and involves “conditions/restrictions” regarding the expected
return whilst minimising its volatility, then we use the Var-Covar
matrix, we set up the Lagrangian function with its parameters
apply FOCs/SOCs, and in general solve a complex system using
matrix algebra
In practice, with the data and methods that are
available to us to measure market beta, it is not
sufficiently useful to compute required rates of
return and expected returns or to discover mispriced
securities.
Multifactor models are useful in this context.
These models introduce uncertainty stemming from
multiple sources
Whereas the CAPM, in principle, limits risk to one
source – covariance with the market portfolio.
We saw previously that the expected return on any
marketable asset s can be written as a function of the
expected return on an efficient portfolio.

What if, we wish to invest in several (alternative)

portfolios that are themselves not efficient, but can be
combined to form an efficient portfolio?
Suppose the efficient portfolio can be formed by
combining two portfolios F1 and F2, called factor
portfolios.
We regress the excess return (return in excess of the risk-free
rate) of the asset s, on the 2 factor portfolios (multivariate
regression)

(1)

To do this, consider a portfolio P, where you first buy the

asset s then sell a fraction βsF1 in factor portfolio 1, and a
fraction βsF2 in factor portfolio 2, and then invest the
proceeds in the risk-free asset.
The return on this portfolio would be:

(2)
If we substitute Rs from (1), into (2), we have:

Now the uncertain part (random residual) of this return εs,

must be uncorrelated with the factor portfolios F1 and F2
and hence with the efficient portfolio.
Consequently, the uncertain part of the return εs, needs no
compensation and does not have any risk premium.
But the expected return on the portfolio P must simply be
the risk-free rate because we invested in it, therefore, αs = 0.
Now if we go back to equation (1) and take the expected
value of both sides, we see that:
The following description assumes (2) two factors, but unrelated to the Efficient
(market) portfolio e.g., different macroeconomic factors, e.g. business cycle
uncertainty , interest rates, inflation etc.

Assume that returns can be generated as in a 2-factor model:

Ri = ai + bi1F1 + bi2F2 + error.

Note that we are no longer assuming any connection between these factors and
the market portfolio.

If there are enough assets that are sufficiently different from each other, it
should be possible to create three well-diversified portfolios that had no
uncertainty other than their exposure to F1 and F2 (i.e. the error term would be zero).

Then, it will be true that the expected return on these three portfolios can be
described by a linear combination of their b’s: E(R) = c0 + c1 bi1 + c2 bi2.

The problem is that the underlying model could change at any time; we have no
guidance as to what generates the underlying model!
Suppose these three portfolios – A, B and C – were
described by the following parameters:
Portfolio Expected Return bi1 bi2
A 15 1.0 0.6
B 14 0.5 1.0
C 10 0.3 0.2

Then, the APT says: E(R) = c0 + c1 bi1 + c2 bi2

That is, 15 = c0 + 1c1 + 0.6c2
14 = c0 + 0.5c1 + 1.0c2 and
10 = c0 + 0.3c1 + 0.2c2.
Solving this system of three equations for c0, c1 and c2, we
find c0 = 7.75, c1 = 5 and c2 = 3.75.
The claim is that this pricing equation (E(R) = 7.75 + 5 bi1
+ 3.75 bi2) can be used to price any security in the
economy, based on many factors.
The figure shows the percentage of firms that use CAPM, multifactor models, the historical
average return etc., or other methods to calculate the return on equity i.e., the RE part, and
thus the WACC.