TOPIC 4: VARIABLE-INCOME MARKETS

1. Harry Markowitz assumptions

Suppose that we have several assets, A1, A2, …, An. Usually, these assets are stocks.

OUR PROBLEM: We have a capital C> 0. How do we distribute it among the different assets?

Thus, we consider portfolios x=(x 1 , x 2 , … , x n) , represented by n dimensional vectors.

 x 1percentage invested on asset 1 (A1)

As we have € C , the total invested will be:

 x 1 X € C in asset 1 (A1)
 x 2 X € C in asset 2 (A2)
 x nX € C in asset n (An)

The problem we are going to face is to compute x so the investment is optimal. What does it mean OPTIMAL
INVESTMENT?

- The return is as great as possible.

This is called RATIONAL CONDUCT of the investor. This theory was born in 1952 with the work of Harry Markowitz:
Portfolio selection, Journal of Finance.

2. Expected return and variance of a portfolio of assets: Covariance matrix

HOW DOES MARKOWITZ MEASURE THE PROFITABILITY OF A PORTFOLIO? With the mathematical expectation of the
return that the investor expects to obtain in the future.

HOW DOES MARKOWITZ MEASURE THE RISK OF A PORTFOLIO? With the standard deviation.

That is why it is called a 2-dimensional investment model or a mean-variance decision model.

RETURN OF A FINANCIAL ASSET AND A PORTFOLIO

What is the return of a portfolio, x ? In order to answer, we need to talk about the return of each asset, A1, A2, ..., An
which are represented by r1, r2, …, r𝑛.

Then, the return, r𝑖, of each asset A𝑖, will be given by the formula

Where:

- P0 (A𝑖) is the initial Price of A𝑖

P0 (A𝑖) is known, but P𝑓 (A𝑖) is RANDOM, so r𝑖 is also RANDOM. That is, we don’t Know its exact value. And the return
of the portfolio? Let's deduce it.
 1. Continuing with the previous reasoning, if we invest a capital C at zero, at 𝑓 it will have become
C f =C(1+ r) as we can see in the graph below.
So we would have that:

2. We have invested: So we obtain:

We add those amounts and get the total:

Also, thanks to step 1, we know that:

So, the return of a portfolio is given by the following formula:

As each r𝑖 is RANDOM (its final value is unknown at the initial moment), will also be RANDOM. A RANDOM VARIABLE
CAN NOT BE OPTIMIZED, SO WE CANNOT MAXIMIZE. BUT WE CAN MAXIMIZE ITS AVERAGE 

Notice that the average is LINEAR:

- The mean of the sum is the sum of the means: E ( x + y + z )=E ( x )+ E ( y )+ E ( z )

PORTFOLIO’S RISK

So far, we have seen that Markowitz proposes to maximize the expected return. To measure the risk, he proposes
the standard deviation. Why? Because it assumes that the r 𝑖 have symmetric distributions, and then he
demonstrates that minimizing the standard deviation, the investor's utility is maximized (Second Order Stochastic
Dominance Principle).

See the following graphs where two distributions are shown, one symmetric and the other asymmetric.
Actually, asset returns usually do not follow symmetric, but asymmetric distributions. For example: financial
derivatives (such as futures, forwards, and financial options). So the variance (or standard deviation) does not serve
to measure the risk in these cases, but there are other risk measures such as the CVaR and other risk measures.

How we compute the risk of a portfolio?

This is another way to compute the Risk of a portfolio:

Let us deduce both formulas. As we have for each r𝑖 we have a sigma σ,

Moreover,

represents the variance and covariance matrix between the different assets. This matrix is symmetric (that is, the
values are the same on each side of the main diagonal).

So risk is defined as: and

Where,
2
σ 1 is the variance of r1, that is, the variance of the return of asset 1.
2
σ 2 is the variance of r2, that is, the variance of the return of asset 2.
2
σ nis the variance of rn, that is, the variance of the return of asset n.
Moreover, the following expression applies

where, σ ijrepresents the covariance between returns of assets i and j. It measures the ratio of dependence (positive
or direct / null / negative or inverse) between two variables. The covariance measures in squared units.

ρij represents the correlation coefficient of ri and rj. It is computed as

Moreover, ρij satisfies -1≤ ρij ≥ 1and it does not have units.

From the previous expression,

Another expression for the risk, using the standard deviation is the following:

The correlation matrix is:

Which is also symmetric. On the main diagonal there are always ones since the correlation coefficient of an asset
with itself is equal to one.

In general, the formula below applies:

3. Efficient Portfolios and Efficient Frontier

Once we know how to compute the expected return and the standard deviation, we take a step further: we look for
portfolios with maximum return at a given risk or portfolios with minimum risk at a given return. This way we
obtain the efficient frontier:

That is, we search for the PARETO’s OPTIMUM.

WHICH ARE, INTUITIVELY, THE PARETO’S OPTIMUM?

(1) The red portfolio is NOT efficient because there is another portfolio with MORE EXPECTED RETURN AND LESS
RISK
(2) The red portfolio is NOT efficient because there is another portfolio with LESS RISK and EQUAL EXPECTED
RETURN
(3) The red portfolio is NOT efficient because there is another portfolio with MORE EXPECTED RETURN AND EQUAL
RISK.
(4) This portfolio is EFFICIENT, because there is no other that offers higher returns and equal risk, or lower risk and
equal returns. It can be concluded that it is a PARETO OPTIMUM.

HOW WE CALCULATE THE EFFICIENT FRONTIER?

In order to compute this frontier, we have two options:

1. To maximize the portfolio’s expected return at a given level of risk. When we change the values of σ0, we
obtain different portfolios on the efficient line.

2. To minimize portfolio’s standard deviation (risk) at a given level of expected return. When we change the
values of E (r) (we obtain different portfolios on the efficient line).

ALL THE PORTFOLIOS OF THE EFFICIENT LINE WILL BE VERY DIVERSIFIED

Every convex optimization problem (the function is convex) leads to diversified results. So we would not obtain 𝑥1 = 1
and 𝑥2 = 0 or 𝑥1 = 0 and 𝑥2 = 1, but, for instance, 𝑥1 = 0,5 and 𝑥2 = 0,5 or 𝑥1 = 0,3 and 𝑥2 = 0,7.
In practice, the problem solved is: or

We will only solve the equality, since otherwise the K-K-T conditions would be involved, and the problem solution
would be more tedious.

4. Introduction of the risk-free asset (RFA) in the Harry Markowitz model

In this model we will consider a risk-free asset (RFA) which pays a return r f and whose standard deviation (risk)
equals zero. So, we are going to diversify between a risky portfolio and the RFA. We will invest:

X  Risky portfolio ( E(r), σ ) and (1 – x)  RFA (r f , 0¿

Given the previous information, the new return will be:

As r f has standard deviation and correlation equal to zero, we have.

Those would be the expected return and the risk of a portfolio which combines a risky asset with a risk-free one.
The expression for the risk is deduced from:

The previous information can be represented as follows 

 LOAN: To invest (to buy) in a risky portfolio and the RFA; to invest only in the RFA (X = 0); to invest only in the
risky portfolio (X = 1).
 DEBT: You borrow money to the bank at the risk-free rate, r f , in order to buy the risky portfolio.

5. The Market Portfolio (M)

We consider the "Umbrella" of Markowitz, represented in the following

So once you incorporate the RFA, you improve all the efficient line and it
becomes a straight line from the RFA, passing through the Market Portfolio
(M). This new straight line represents the possible combinations of the with
the Market Portfolio.

 LOAN: we diversify between the M and the RFA.

ABOUT THE MARKET PORTFOLIO (M):

When we introduce the RFA, all the portfolios are beaten but one, which is the Market Portfolio. The M is a very
diversified portfolio, so it contains all the stocks of the market. That is, we can
approximate it by a financial index. The M is characterized by the tangency, so
it maximizes the slope of the straight line. This slope is called Sharpe Ratio. The
M MAXIMIZES the Sharpe Ratio.

The M resolves: or
Every portfolio has a Sharpe Ratio and the is the one having the greater one.

6. Capital Market Line (CML)

The following graph shows the line containing those portfolios with the same Sharpe Ratio as the market. These
portfolios are called efficient portfolios, and the line that contains them is called the Capital Market Line (CML).

The CML is given by the equation below:

where represents the Sharpe Ratio of the Market Portfolio.