FINS5513 Lecture 2B

Forming Efficient
Portfolios
Lecture Outline
❑ 2.4 Introduction to Modern Portfolio Theory (MPT)
➢ Portfolios of Two Risky Assets
• Return, risk, covariance and correlation
➢ Multiple Risky Assets
➢ Diversification Benefit
• Perfect correlation
• Portfolio diversification benefit
❑ 2.5 Optimal Risky Asset Portfolio Construction
➢ Minimum Variance Frontier

➢ Efficient Frontier
• Global Minimum Variance Portfolio (GMVP)
• Optimal portfolio: no risk-free asset

2
2.4 Introduction to
Modern Portfolio Theory
(MPT)

FINS5513
Portfolios of Two Risky Assets

FINS5513 4
Forming Portfolios
❑ What is our objective when we make an investment?
To achieve the optimal investment outcome that maximises our utility
Utility from an investment depends on its expected return and risk
❑ The task is therefore to construct the optimal portfolio from all available assets, which has the
preferred combination of expected return and risk, therefore providing the highest utility
(among all possible portfolios)
❑ So, we first need to compute the expected return and risk of a portfolio directly from its
component assets
➢ Then we need to weight the portfolio optimally

❑ Let’s start simply with two assets

5
Portfolios of Two Risky Assets: Return
❑ Portfolio returns are simply a weighted average of individual asset classes in the portfolio.
For example, consider Bonds and Equities asset classes:
𝑟𝑝 = 𝑤𝐵 𝑟𝐵 + 𝑤𝐸 𝑟𝐸
𝑟𝑝 = Portfolio return
𝑤𝐵 = Bond weight
𝑟𝐵 = Bond return
𝑤𝐸 = Equity weight
𝑟𝐸 = Equity return

❑ Similarly, portfolio expected return is a weighted average of expected returns of the two asset
classes in the portfolio:
𝐸(𝑟𝑝) = 𝑤𝐵 𝐸(𝑟𝐵 ) + 𝑤𝐸 𝐸(𝑟𝐸 )

6
Portfolios of Two Risky Assets: Risk
❑ Calculating portfolio risk on the other hand is not simply a weighted average. That is, it is
not simply a case of taking the individual risks of the portfolio’s component assets
❑ We must also take into account the covariance between asset classes (or individual assets)
within the portfolio:

𝜎𝑝2 = 𝑤𝐵2 𝜎𝐵2 + 𝑤𝐸2 𝜎𝐸2 + 2𝑤𝐵 𝑤𝐸𝐶𝑜𝑣(𝑟𝐵 , 𝑟𝐸 )

𝜎𝑝2 = Portfolio variance
𝜎𝐵2 = Bond variance
𝜎𝐸2 = Equity variance
𝐶𝑜𝑣(𝑟𝐵 , 𝑟𝐸 ) = Covariance between Bonds and Equities

7
Covariance
❑ The covariance between two variables refers to their tendency to be higher or lower than their
respective mean values at the same time
➢ It is the sum of deviations from the mean of two assets in different states

σ𝑛
𝑡=1 𝑟𝐵𝑡 − 𝑟𝐵 𝑟𝐸𝑡 − 𝑟𝐸
For time series data (ex-post): 𝐶𝑜𝑣 𝑟𝐵 , 𝑟𝐸 =
𝑛−1
𝐶𝑜𝑣 𝑟𝐵 , 𝑟𝐸 = Covariance of returns between Bonds and Equities
𝑟𝐵𝑡 − 𝑟ഥ𝐵 = Deviation from the mean for Bonds at time 𝑡
𝑟𝐸𝑡 − 𝑟ഥ𝐸 = Deviation from the mean for Equities at time 𝑡
𝑛 = Number of observations
❑ It is useful to note that variance is a special case of covariance. The covariance of a variable
with itself is simply its variance

8
Correlation
❑ Correlation scales the covariance to 1
❑ Covariance of returns between Bonds and Equities can also be expressed in terms of their
correlation:
𝐶𝑜𝑣 𝑟𝐵 , 𝑟𝐸 = 𝜎𝐵𝐸 = 𝜌𝐵𝐸𝜎𝐵𝜎𝐸
𝐶𝑜𝑣 𝑟𝐵 ,𝑟𝐸
Therefore: 𝜌𝐵𝐸 =
𝜎𝐵𝜎𝐸

𝜌𝐵𝐸 = Correlation coefficient of returns between Bonds and Equities

❑ Range of values for 𝜌𝐵𝐸 : − 1.0 ≤ 𝜌𝐵𝐸 ≤ +1.0

9
Example: Portfolio Risk Reduction
❑ Example 2B1: Balanced Fund calculates that the statistical characteristics of the asset
classes equities and bonds are as follows (assume a risk-free rate of 2.5%):
Asset Expected Risk Premium Risk Sharpe Covariance
Class Return (rf = 2.5%) σ ratio
Equities 11.1% 8.6% 21.7% 0.396
-0.01309
Bonds 3.5% 1.0% 8.5% 0.118

Calculate the correlation between the assets. Assume Balanced Fund constructs a portfolio
comprising 60% Equities / 40% Bonds. Calculate the portfolio return, risk and Sharpe ratio
➢ Correlation= -.01309 / (.217 x .085) = -.7095
➢ 𝐸(𝑟𝑝) = .6 x .111 + .4 x .035 = 8.06% → Excess return = 8.06% – 2.50% = 5.56%
➢ 𝜎𝑝2 = .62 x .2172 + .42 x .0852 + 2 x .6 x .4 x -.01309 = 0.0118 and 𝜎𝑝 = 0.0118 = 10.88%
➢ Sharpe ratio = .0556 / .1088 = 0.511
❑ Sharpe ratio improves from the benefit of negative correlations (diversification benefit)
Excel 2BE1: “2B - Portfolio Return, Risk, Covariance and Correlation” 10
Multiple Risky Assets

FINS5513 11
Portfolios of Multiple Risky Assets: Return
❑ The return on a portfolio of more than two assets, 𝑟𝑝 is also simply the weighted average of
the returns of the assets that make up the portfolio:
𝑛

𝑟𝑝 = 𝑤1 𝑟1 + 𝑤2 𝑟2 + ⋯ + 𝑤𝑛 𝑟𝑛 = ෍ 𝑤𝑖 𝑟𝑖
𝑖=1

𝑤𝑖 = portfolio weight of asset 𝑖 (i.e. the fraction of portfolio value invested in asset 𝑖)
❑ Similarly, the expected return on a portfolio is a weighted average of the expected returns on
its component assets: 𝑛

𝐸 𝑟𝑝 = ෍ 𝑤𝑖 𝐸(𝑟𝑖 )
𝑖=1

12
Portfolios of Multiple Risky Assets: Risk
❑ For 𝑛 assets (or asset classes), portfolio risk can be calculated with a covariance matrix,
consisting of:
➢ 𝑛 variances

➢ (𝑛2 – 𝑛)/2 covariances

❑ The generalised result for portfolio variance (risk) is given by:
𝑛 𝑛

𝜎𝑝2 = 𝐶𝑜𝑣(෍ 𝑤𝑖 𝑟𝑖 , ෍ 𝑤𝑖 𝑟𝑖 )
𝑖=1 𝑖=1

❑ We can set up a variance-covariance matrix from this equation

13
Portfolios of Multiple Risky Assets: Risk
❑ For a 3 stock portfolio:
𝑤1 𝑟1 𝑤1 𝑟1 𝑤12 𝜎12 + 𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2 + 𝑤1 𝑤3 𝐶𝑜𝑣 𝑟1 , 𝑟3
𝑤2 𝑟2 𝑤2 𝑟2
= 𝑤22 𝜎22 + 𝑤2 𝑤1 𝐶𝑜𝑣 𝑟2 , 𝑟1 + 𝑤2 𝑤3 𝐶𝑜𝑣 𝑟2 , 𝑟3
𝑤3 𝑟3 𝑤3 𝑟3 𝑤32 𝜎32 + 𝑤3 𝑤1 𝐶𝑜𝑣 𝑟3 , 𝑟1 + 𝑤3 𝑤2 𝐶𝑜𝑣 𝑟3 , 𝑟2

𝑤12 𝜎12 + 𝑤22 𝜎22 + 𝑤32 𝜎32 + 2[𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2 + 𝑤1 𝑤3 𝐶𝑜𝑣 𝑟1 , 𝑟3 + 𝑤2 𝑤3 𝐶𝑜𝑣 𝑟2 , 𝑟3 ]

❑ For a 5 stock portfolio

𝑤1 𝑟1 𝑤1 𝑟1 𝑤12 𝜎12 + 𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2 + 𝑤1 𝑤3 𝐶𝑜𝑣 𝑟1 , 𝑟3 + 𝑤1 𝑤4 𝐶𝑜𝑣 𝑟1 , 𝑟4 + 𝑤1 𝑤5 𝐶𝑜𝑣 𝑟1 , 𝑟5
𝑤2 𝑟2 𝑤2 𝑟2 𝑤22 𝜎22 + 𝑤2 𝑤1 𝐶𝑜𝑣 𝑟2 , 𝑟1 + 𝑤2 𝑤3 𝐶𝑜𝑣 𝑟2 , 𝑟3 + 𝑤2 𝑤4 𝐶𝑜𝑣 𝑟2 , 𝑟4 + 𝑤2 𝑤5 𝐶𝑜𝑣 𝑟2 , 𝑟5
𝑤3 𝑟3 𝑤3 𝑟3 𝑤32 𝜎32 + 𝑤3 𝑤1 𝐶𝑜𝑣 𝑟3 , 𝑟1 + 𝑤3 𝑤2 𝐶𝑜𝑣 𝑟3 , 𝑟2 + 𝑤3 𝑤4 𝐶𝑜𝑣 𝑟3 , 𝑟4 + 𝑤3 𝑤5 𝐶𝑜𝑣 𝑟3 , 𝑟5
=
𝑤4 𝑟4 𝑤4 𝑟4 𝑤42 𝜎42 + 𝑤4 𝑤1 𝐶𝑜𝑣 𝑟4 , 𝑟1 + 𝑤4 𝑤2 𝐶𝑜𝑣 𝑟4 , 𝑟2 + 𝑤4 𝑤3 𝐶𝑜𝑣 𝑟4 , 𝑟3 + 𝑤4 𝑤5 𝐶𝑜𝑣 𝑟4 , 𝑟5
𝑤5 𝑟5 𝑤5 𝑟5 𝑤52 𝜎52 + 𝑤5 𝑤1 𝐶𝑜𝑣 𝑟5 , 𝑟1 + 𝑤5 𝑤2 𝐶𝑜𝑣 𝑟5 , 𝑟2 + 𝑤5 𝑤3 𝐶𝑜𝑣 𝑟5 , 𝑟3 + 𝑤5 𝑤4 𝐶𝑜𝑣 𝑟5 , 𝑟4

𝑤12 𝜎12 + 𝑤22 𝜎22 + 𝑤32 𝜎32 +𝑤42 𝜎42 + 𝑤52 𝜎52 + 2[𝑤1 𝑤2 𝐶𝑜𝑣 𝑟1 , 𝑟2 + 𝑤1 𝑤3 𝐶𝑜𝑣 𝑟1 , 𝑟3
+ 𝑤1 𝑤4 𝐶𝑜𝑣 𝑟1 , 𝑟4 + 𝑤1 𝑤5 𝐶𝑜𝑣 𝑟1 , 𝑟5 + 𝑤2 𝑤3 𝐶𝑜𝑣 𝑟2 , 𝑟3 + 𝑤2 𝑤4 𝐶𝑜𝑣 𝑟2 , 𝑟4
+ 𝑤2 𝑤5 𝐶𝑜𝑣 𝑟2 , 𝑟5 + 𝑤3 𝑤4 𝐶𝑜𝑣 𝑟3 , 𝑟4 + 𝑤3 𝑤5 𝐶𝑜𝑣 𝑟3 , 𝑟5 + 𝑤4 𝑤5 𝐶𝑜𝑣 𝑟4 , 𝑟5 ]
14
Diversification Benefit

FINS5513 15
Portfolio Risk: Perfect Correlation
❑ Recall portfolio variance for a 2-asset portfolio:
𝜎𝑝2 = 𝑤12 𝜎12 + 𝑤22 𝜎22 + 2𝑤1 𝑤2𝐶𝑜𝑣(𝑟1 , 𝑟2 )
Which can be restated in terms of correlation as:
𝜎𝑝2 = 𝑤12 𝜎12 + 𝑤22 𝜎22 + 2𝑤1 𝑤2 𝜌12 𝜎1 𝜎2
❑ If asset 1 and 2 are perfectly correlated ρ = 1, we have:
𝜎𝑝2 = 𝑤12 𝜎12 + 𝑤22 𝜎22 + 2𝑤1 𝑤2 𝜎1 𝜎2 = (𝑤1 𝜎1 + 𝑤2 𝜎2 )2
𝜎𝑝 = 𝑤1 𝜎1 + 𝑤2 𝜎2
❑ In other words, the standard deviation of a portfolio with perfectly correlated assets is the
weighted average of the component asset standard deviations
➢ For perfectly correlated assets, both portfolio risk and return are a weighted average

➢ There is no risk reduction benefit in combining perfectly correlated assets

16
Portfolio Risk: Perfect Correlation
❑ For perfectly correlated assets A and B, we can restate the risk and return equations as
follows:
Risk Return

❑ So graphically, 𝑤𝐵 tells us where we are along

a line connecting point A (a portfolio consisting
of 100% in A i.e. 𝑤𝐵 = 0) and point B (a
portfolio consisting of 100% in B i.e. 𝑤𝐵 = 1)
❑ In other words, for perfectly correlated
assets the risk-return relationship is linear

17
Portfolio Risk: Diversification
❑ For lower values of 𝜌 i.e. imperfect correlation, the portfolio standard deviation must be lower:
𝜎𝑝2 = 𝑤12 𝜎12 + 𝑤22 𝜎22 + 2𝑤1 𝑤2 𝜌12 𝜎1 𝜎2 < (𝑤1 𝜎1 + 𝑤2 𝜎2 )2
𝜎𝑃 < 𝑤1 𝜎1 + 𝑤2 𝜎2
❑ By combining low/negative correlated assets, we can optimise portfolio outcomes
❑ Example 2B2: Balanced Fund wants to determine portfolio diversification benefits at different
portfolio weights and different assumed stock-bond correlation levels. The correlation it
derived earlier was -0.71 (negative correlation). Suppose it wishes to determine the optimal
weightings between stocks and bonds assuming correlations of +1.0, +0.7, 0, -0.71
(estimated correlation from earlier), and -1.0.
➢ To calculate this, we replace various weighting levels (say between 0-100% in increments of
10%) and calculate the expected return at different weightings. Then replace in the various
correlations above into the two-asset portfolio risk formula (from earlier) and calculate
portfolio risk at different weighting levels (say between 0-100% in increments of 10%)

Excel 2BE1: “2B - Portfolio Return, Risk, Covariance and Correlation”

18
 Diversification Benefit
Diversification Benefit ❑ Portfolio weights change as we move from
12%
E to B. The lower the correlation, the more
Portfolio Expected Return %

E
portfolio risk drops as we move from E to B
10%
at each return level
Diversification benefit
8%
❑ The curvature is the diversification benefit.
100% invested
in Equities The lower the correlation the bigger the
6%
bulge (diversification benefit) - same
4%
portfolio return for lower portfolio risk
B 100% invested ❑ The diversification benefit depends on the
in Bonds
2%
correlation:
➢ If 𝜌 = +1.0, no risk reduction is possible
0%
0% 5% 10% 15% 20% 25%
➢ If 𝜌 = 0, 𝜎𝑝 will be less than the standard
Portfolio Standard Deviation %
deviation of either component asset
Correlation = 1 Correlation = .7 Correlation = 0

Correlation = -.71 Correlation = -1

➢ If 𝜌 = -1.0, a riskless hedge is possible

Video 2BV1: RB “The benefits of combining less than perfectly correlated assets”
19
Diversification Benefit
❑ Combining two or more imperfectly correlated assets in one portfolio is called diversification
➢ The risk reduction is called a diversification benefit

➢ This concept is at the heart of portfolio theory

➢ Sophisticated version of “not putting all your eggs in one basket”

➢ The lower the correlation the better (same portfolio return for lower portfolio risk)

➢ We typically want to combine many assets with low (ideally negative) correlations to
maximize the diversification benefit
❑ Why is there a diversification benefit?
➢ Low correlated assets are unlikely to return below their respective means at the same time
(cancels out some portfolio risk)
➢ If correlations are negative, they even work as insurance for each other (a riskless hedge
may be possible)

Mini Case Study: “Bridgewater’s All Weather Fund: Dalio’s E=MC2 Moment”
20
Portfolio Diversification

❑ In this example, as we add more assets to our portfolio the portfolio standard deviation drops
from ~50% to ~19% - this is the diversification benefit
➢ However, the diversification benefit increases rapidly at first, then diminishes significantly

Video 2BV2: “Ray Dalio breaks down his ‘Holy Grail’"

21
2.5 Optimal Risky Asset
Portfolio Construction

FINS5513
Minimum Variance Frontier

FINS5513 23
What is an Optimal Portfolio?
❑ Imagine an investment universe with assets characterized by the risk-return profile as plotted
below:

❑ So how do we form an optimal portfolio out of these 10 individual assets

24
Minimum Variance Frontier
❑ By combining risky assets in a portfolio by weighting each individual asset appropriately we
can construct portfolios with risk-return profiles which are on the red line below
❑ The red line is known as the minimum variance frontier MVF – the lowest risk portfolio at
each level of return. The derivation steps are:
1. Set the target expected portfolio return eg 10%
2. Optimise portfolio weights to minimise variance at this level of return
3. Repeat at different return levels till we have plotted the frontier

25
Minimum Variance Frontier
❑ From the mean-variance criterion, we know that any portfolio or asset combination below the
MVF will be dominated by a portfolio on the MVF
❑ Similarly, any portfolio below the turning point of the MVF will be dominated by one above the
turning point
➢ The turning point is the portfolio (asset combination) that has the lowest possible risk level

➢ It is known as the Global Minimum Variance Portfolio (GMVP)

Portfolio combinations
or individual assets
which are dominated
GMVP
g

26
Efficient Frontier

FINS5513 27
Global Minimum Variance Portfolio (GMVP)
❑ We can see that not every point on the MVF is optimal. Those points below the turning point
of the frontier are dominated by points above the turning point (higher return for the same
level of risk)
❑ The GMVP is the turning point, and it is the portfolio combination which provides the lowest
possible risk (defined by standard deviation). It is a unique set of portfolio asset weightings
which results in the lowest possible standard deviation:
❑ For a portfolio with just 2 risky assets (or asset classes), the GMVP portfolio weights are
given by:
𝜎22 − 𝐶𝑜𝑣(𝑟1 ,𝑟2 )
𝑤1 =
𝜎12 + 𝜎22 −2𝐶𝑜𝑣(𝑟1 ,𝑟2 )

𝑤2 = 1 - 𝑤1

28
Efficient Frontier
❑ As each portfolio on the MVF above the GMVP dominates each portfolio below the GMVP, we
discard the portion below the GMVP
❑ This is called the Efficient Frontier
➢ To plot the efficient frontier we would need to identify the GMVP first
➢ All portfolio combinations on the MVF below the GMVP are discarded as they are all
dominated by the GMVP
➢ Risk averse investors should only choose portfolios on the efficient frontier

E(r)

GMVP Efficient assets are those sitting

on the Efficient Frontier (lowest
risk at each level of return)
mvp

σ
29
Optimal Point on the Efficient Frontier?
❑ So, where along the efficient frontier should we invest?
❑ We should pick the asset portfolio weighting combination which provides the highest utility
➢ Highest utility is represented by the highest attainable indifference curve

➢ The portfolio marked by the star gives the highest possible utility for this investor
➢ It lies on the tangent point between the efficient frontier and the highest attainable
indifference curve

E(r)

Optimal risky
portfolio

σ 30
Optimal Portfolio: No Risk-free Asset
❑ Although all investors face the same efficient frontier, they will each have different utility
functions (due to different risk aversion coefficients A)
➢ They will therefore have different indifference curves
➢ So the optimal risky portfolio may differ between investors
E(r)

Optimal risky portfolio

Investor A Optimal risky portfolio
Investor B

𝝈
❑ Portfolios marked by stars are optimal portfolios when we only have risky assets available
❑ Adding a risk-free asset can improve our allocation options and enhance our optimal portfolio
31
Next Lecture
❑ BKM Chapter 6 and 7

❑ 3.1 Introducing the Risk-Free Asset

❑ 3.2 Deriving Optimal Portfolio Weights

32