You are on page 1of 4

Lab Exercise 2-2 Finance Model Analysis

Portfolio Formation Model

Problem
An individual plans to invest in a diversified portfolio consisting of four assets with different rates of
return and riskiness. Before investing, he has collected data of monthly average price movements of
those four assets over the past 12 months. After conducting some statistical analysis, he has come up
with the mean and standard deviation of monthly log rate of return from each of the four assets. Their
covariances are also derived statistically in the variance-covariance matrix. Assuming that his
personal fund is allocated equally among these four assets, he is interested in knowing what would be
the expected rate of return from holding this portfolio and how much portfolio risk he would bear.

Asset Allocation
Portfolio formation problems deal with how to optimally allocate limited funding resources to
different asset classes. From an individual investor’s micro-perspective, these problems often involve
the dynamic tradeoffs between rate of return expected of each asset and it accompanying risk
exposure. Some speculative investors prefer the highest possible rate of return regardless of the risk
levels. Other conservative investors tend to prefer the lowest possible level of risk given some levels
of expected rate of return. From the market’s macro-perspective, the search for a common ground on
which all investors’ return utility and risk preference functions lie is the main focus of equilibrium and
arbitrage asset pricing theories wherein the so-called market portfolio is studied.
We shall not concern ourselves with the market portfolio formation for the time being. Rather, we
shall investigate how a certain individual with a given risk-return tradeoff function allocates his wealth
across many risky assets. We start with the derivation of an expected rate of return on a portfolio
E[R]:
E[R] = Σ wi ri Equation 1
where: R = rate of return on a portfolio
ri = rate of return on an individual asset i
wi = holding weight of the individual asset i
The expected rate of return on a portfolio is the sum of rates of return on each asset weighted by its
investment proportion. The rate of return on each individual asset used in the equation comes from
either (1) a rate of dividend yield from cash flows or (2) a rate of capital gain/loss from price changes.
The former is the ratio between the expected dividend per share and the current asset price; the latter is
the ratio between the expected change in asset price and the current asset price.
We will concentrate on the latter representation of individual asset’s rate of return, i.e., capital gain
from price movements, as it can be statistically derived from time-serial data of historical asset prices.
The capital-gain rate of return on a discrete change in asset price is given by:
ri = ΔP / P0 Equation 2
where: ΔP = one-period discrete change in asset price
P0 = current-period asset price
When more historical price data are collected, the pattern of changes in asset prices become more
smooth and continuous. Due to the limited liability legal feature of a public corporation issuing the
asset such as common stocks, asset price will never go below zero. This feature results in the
distribution of asset prices that is lognormal. Therefore, it is more appropriate to express the
continuous capital-gain rate of return in its logarithmic form:
ri = ln[(P0 + ΔP)/P0] Equation 3
This serves two major purposes. First, this log rate of return (or “log return” for short) will have a
normal distribution, which is easier to manipulate than the lognormal distribution. Second, when this
rate of return is known we can derive the asset price in the future period from its current price using a
continuous compounding factor:

1
Lab Exercise 2-2 Finance Model Analysis

exp(ri)P0 = (P0 + ΔP) = P1 Equation 4


With a series of log returns derived from Equation 3, we can now calculate for the average value of
log rate of return (or “mean log return” for short) and the standard deviation of log rate of return (or
“SD log return” for short). However, an Excel spreadsheet model used for deriving these two
statistical parameters is left to the students as their exercise. For simplicity, we assume that they are
readily available for all risky assets we intend to form in our portfolio.
Our next task is to derive a portfolio’s variance and standard deviation. The formula below is used for
calculating the variance of a portfolio V[R]:
V[R] = Σ wi2 σi2 + 2(Σi Σj wi wj σij) Equation 5
where: σi = variance of return of an individual asset i
2

σij = covariance between returns of asset i and asset j


Statistically, the covariance measure can be derived from the products of the standard deviations of
assets i and j and their correlation coefficient ρij.
σij = ρij σi σj Equation 6
With an alternative representation of covariance, the variance of a portfolio can also be rewritten as:
V[R] = Σ wi2 σi2 + 2(Σi Σj wi wj ρij σi σj) Equation 7
The standard deviation of a portfolio SD[R] is simply the square root of the variance of a portfolio
V[R]:
SD[R] = V[R]0.5 = [Σ wi2 σi2 + 2(Σi Σj wi wj ρij σi σj)]0.5 Equation 8
This is the cross-sectional measure of portfolio’s risk, which will become an objective function for a
portfolio optimization problem to be discussed later.

Input Variables
Suppose there are four risky assets that we are currently holding. Their parameters, including the
mean monthly log return, the SD monthly log return, and the covariance log return, are provided in the
two tables below as direct input values.

Mean Log SD Log Holding


Asset Class
Return Return Weight
Asset 1 0.5142% 5.9462% 25.00%
Asset 2 1.5575% 4.8984% 25.00%
Asset 3 3.2399% 4.7830% 25.00%
Asset 4 2.6886% 6.3034% 25.00%

VCV
Asset 1 Asset 2 Asset 3 Asset 4
Matrix
Asset 1 0.001287 0.000739 0.000898
Asset 2 0.001287 0.000599 0.001269
Asset 3 0.000739 0.000599 -0.000276
Asset 4 0.000898 0.001269 -0.000276

The screen shot of the Excel spreadsheet for our Portfolio Formation Model is also shown below.

2
Lab Exercise 2-2 Finance Model Analysis

Calculating Portfolio Expected Return


In cell E6, enter =D6*B6. Then, copy-paste it down to cells E7:E9. Let us name the range E6:E9 as
“ExpectedReturns” so that we can do the summation in cell G14 as =SUM(ExpectedReturns). This
gives us the expected rate of return from a portfolio ΣE[R] pursuant to Equation 1 above.

Calculating Portfolio Variance and Standard Deviation


We first start with the derivation of asset variances. In cell B14, enter =C6^2. This represents the
square of SD monthly log return we have as the input variable. Repeat the process in cells C15, D16,
and E17 with =C7^2, =C8^2, and C9^2, respectively. Notice that the downward diagonal cells in the
VCV Matrix are all the values of asset variances.
In the second step, we will move to cell F6. In there, enter =D6^*B14. This represents the weighted
variance of each asset. We cannot copy-paste it down since the values of asset variances lie
diagonally in the VCV Matrix table. Thus, we have to complete all the remaining cells in turn. In cell
F7, enter =D7^2*C15; cell F8, enter =D8^*D16; and cell F9, enter =D9^2*E17. We then sum F6:F9
in cell F10 to obtain ΣV[ri]. And let us name cell F10 as “SumAssetVariances”.
Our third step to derive the portfolio variance is to deal with the covariances between the six pairs of
four risky assets. We move to cell G6 and enter =D6*D7*B15. This represents the product of the two
weight values and the covariance between Asset 1 and Asset 2. Use the same logic to repeat this
process for Assets 1 & 3, Assets 1 & 4, Assets 2 & 3, Assets 2 & 4, and Assets 3 & 4. Then, in cell
G12 so the summation of G6:G11. Let us name cell G12 as “SumAssetCovariances”.
The fourth step is simply the combination of the results from the second and third steps. In cell G15,
enter =SumAssetVariances+2*SumAssetCovariances. This will give us the value of portfolio
variance pursuant to Equation 5 above.
Finally, we can derive the value of portfolio standard deviation by taking the square root of portfolio
variance. So, in cell G16, enter =G15^0.5. This is done in the same fashion as in Equation 6.
The following screen shot shows all the results obtained from all the above step-wise procedures.

3
Lab Exercise 2-2 Finance Model Analysis

Obtaining Portfolio Return, Variance, and Standard Deviation with Matrix


An alternative way to derive all the three parameters we have just calculated is to use the matrix
operations and the array functions provided by Excel. Through this method, we can bypass all the
tedious pair-wise calculations. Imagine if we have forty asset classes instead of four. Using matrix
calculation would shorten our formulation as well as calculation time manifold.
We start by going to cell B20, and enter the following nested VLOOKUP function:
=B14/(VLOOKUP(B$19,ReturnRiskArray,3,0)*VLOOKUP($A20,ReturnRiskArray,3,0))
where “ReturnRiskArray” is the named range of A6:C9. We can then copy-paste this cell to the rest
of the Correlation Matrix table. Notice that the downward diagonal cells all have a unity value. This
simply means that the correlation coefficient between an asset and itself is a perfect one.
Now, let us go to cell H14, which is the cell for E[R]. Enter the following array function:
{=MMULT(TRANSPOSE(PortWeights),AssetDrifts)}
Do not forget to press Shift+Ctrl+Enter simultaneously before exiting the cell formulation. This is
necessary because the formula represents a matrix multiplication that utilizes Excel array function.
To obtain the value of portfolio variance V[R] through our Correlation Matrix table in cell H15, just
enter the following array function:
{=MMULT(TRANSPOSE(PortWeights),MMULT(VCVMatrix,PortWeights))}
Again, do not forget to press Shift+Ctrl+Enter before leaving. We will see that the result is identical
to the one we had previously done using the pair-wise method.
There is no complex array function to derive SD[R] in cell H16 unless we want it to be. It should
follow that SD[R] is the square-root value of V[R] of the cell above it.

You might also like