SECURITIES MARKETS 1&2

TOPIC 5: ASSET ALLOCATION

 TOPIC 5: THEMES

• Modern Portfolio Theory

• Investment Opportunity Set
• Efficient Frontier
• Optimal Portfolios
• Practical Issues: Estimation and Views
• Black-Litterman Model
• International Investments
• Life-Cycle Investing
• Case: Harvard Management Company

5-2
 TOPIC 5: QUESTIONS

• If you only care about the risk and return of your investments, how much
should you invest in each asset?
• How does your strategy depend on the correlation between two assets?
• Does adding new assets to your portfolio improve your risk-return tradeoff?
• How sensitive is the tangency portfolio to the inputs (E,σ, ρ)?
• How do we determine the inputs in practice?
• Did Harvard make money because it was smart or patient?
• How did Harvard’s endowment evolve over time? Why?
• If you are smart, should you still be diversified?
• Should you change the proportion between bonds and stocks as you get
older?
• Should you short your own company?

5-3
1. ASSET ALLOCATION and MODERN
PORTFOLIO THEORY (MPT)
 ASSET ALLOCATION AND MPT

Main problem: allocate wealth between various assets

Before modern era: “Fundamental analysis”

 Benjamin Graham, Warren Buffett

Modern Portfolio Theory (MPT)

 Also called Mean-Variance portfolio analysis
 Developed by Harry Markowitz, Jim Tobin, and Bill Sharpe in the 1950s
and 1960s
• All three got the Nobel Prize in Economics

Key concept: diversification

5-5
 MPT – TWO-STEP SOLUTION

We are given a set of financial assets and must decide how to allocate our
investment capital among them
 For example, we might have to find the optimal portfolio from a
universe of stocks and bonds

Modern Portfolio Theory (MPT) takes a simplifying approach by assuming

investors care only about the mean and variance of returns
 These are mean-variance investors
 Therefore, MPT focuses on portfolios drawn in (E,σ) coordinates

Two-step solution
 STEP A: Find the set of all feasible (possible) portfolios
• Called the Investment Opportunity Set (the IO Set)
 STEP B: Find the optimal portfolio in the IO Set

5-6
 PORTFOLIOS

Suppose we invest only in 2 assets: GE, KO

Start with $100,000 and invest $60,000 in GE, and $40,000 in KO

Then we have a portfolio P

 The portfolio weights are
wGE = 0.6 = 60%
wKO = 0.4 = 40%
 The weights must add up to 1 = 100%

The return of the portfolio P is a weighted average of individual asset returns

rP = wGE rGE + wKO rKO

5-7
 PORTFOLIOS

The expected return (E) of the portfolio P is

EP = wGE EGE + wKO EKO

The standard deviation (σ) of P is the square root of the variance of P

The variance of P is given by the following formulas

where ρGE,KO is the correlation of GE and KO

5-8
2. INVESTMENT OPPORTUNITY (IO) SET
 1 RISK-FREE + 1 RISKY ASSET

Example: You have a 1-year horizon, and consider splitting your money between
two assets
 An index of stocks, the Standard & Poor 500 (call it X)
 A risk-free asset, a 1-year T-Bill (call it rf)

Solution: Start with some portfolio P, and denote by

 w = the fraction of P invested in X
 1 – w = the fraction of P invested in rf

5-10
 1 RISK-FREE + 1 RISKY ASSET

1. What about the portfolio corresponding to w = 1?

 This is the S&P 500 (X). Its E and σ are

2. What about the portfolio corresponding to w = 0?

 This is the T-Bill (rf). Its E and σ are

3. Consider the portfolio P corresponding to w = 0.50. We have

5-11
 IO SET: CAPITAL ALLOCATION LINE (CAL)

Let's draw all these portfolios in the (E,σ)-plot

 All portfolios are on the same line: the Capital Allocation Line (CAL)

5-12
 SHARPE RATIO

We have just obtained the Capital Allocation Line formula:

 An application of: “The higher the risk, the higher the reward”

The slope of the CAL is the Sharpe Ratio of asset X:

For the S&P 500 (as for any P on CAL) the Sharpe ratio is

5-13
 TWO RISKY ASSETS

Example: You have a 1-year horizon, and consider splitting your money between
two assets
 An index of stocks, e.g., S&P 500 (call it A)
 An index of bonds, e.g., Lehman Index (call it B)

 Assume the correlation between A and B is ρAB = 0

Solution: Start with some portfolio P, and denote by

 w = the fraction of P invested in A
 1 – w = the fraction of P invested in B

5-14
 IO SET: TWO RISKY ASSETS
The IO Set in the (E,σ)-plot is now a hyperbola
P has lower risk than both stocks and bonds + higher expected return than bonds
 This is an example of diversification

5-15
 IO SET and CORRELATION

The shape of IO Set depends on the correlation ρAB between A and B

5-16
 IO SET and CORRELATION

Notice that for ρ = – 1 it is possible to establish a perfectly hedged portfolio with

these 2 securities
 That is, a portfolio that has no risk (σP = 0)
 Example: The resort owner and the umbrella manufacturer

When ρ = 0.5, 0, or – 0.5 the plot shows some of the hedging effect, though not
as much as when ρ = – 1
 This is an example of diversification: combining assets in a portfolio may
reduce the overall risk!

The lower the correlation, the better the diversification. Why?

 Assets with low correlation with other assets provide better insurance
against portfolio downfall

5-17
 MANY RISKY ASSETS

In the case with more than 2 risky assets, the IO Set is a hyperbola together with all
the portfolios to the right of the hyperbola

The frontier of the IO Set is called the Efficient Frontier

 The efficient frontier is formed with the portfolios which for a given
expected return have minimum variance

Efficient portfolio is any portfolio on the efficient frontier

 Any interior portfolio is dominated by an efficient portfolio (the one with
same E but minimum σ)
 Therefore, any optimal portfolio must be efficient

5-18
 MANY RISKY ASSETS – INPUTS

Example: Harvard Management Company invests in 11 asset classes, plus cash

(assumed riskless). The expected real returns, standard deviations, and
correlations are (in %/year):

5-19
 IO SET: MANY RISKY ASSETS
Plot the IO Set and the 11 asset classes

5-20
 ADDING NEW RISKY ASSETS

What happens when you introduce a new asset to the existing IO set?
 The IO set becomes bigger!
• Unless the new asset is redundant, i.e., it is a linear combination of
existing assets

And what happens to the efficient frontier when we introduce a new risky asset?
 It moves to the left
 Now we can achieve the same expected return, but have lower risk

 Diversification makes everyone better off

5-21
 IO SET: REMOVING RISKY ASSETS
Plot the IO set for the Harvard MC, if they invest only in the first 5 or 10 assets

5-22
 1 RISK-FREE + MANY RISKY ASSETs

Many risky assets  Efficient frontier of all risky assets is a hyperbola

With one risk-free asset  Efficient frontier is a line: Capital Allocation Line
(CAL)
CAL has the highest Sharpe Ratio in the IO set  CAL is tangent to the
hyperbola at T = tangency portfolio

5-23
 FINDING THE TANGENCY PORTFOLIO

How do we determine the tangency portfolio T? The weights of T satisfy a 2⨯2

system of equations with the covariance matrix as coefficients and the expected
excess returns as constant terms

Covariance matrix:

Expected excess returns:

5-24
 FINDING THE TANGENCY PORTFOLIO

The equation is

The solution is

The weights must add to 1, so normalize by the sum wA + wB = 5.24

 The tangency portfolio weights are:

 You should invest 58% of your risky money in stocks (A) and 42% in bonds (B)

5-25
3. OPTIMAL PORTFOLIOS
 OPTIMAL PORTFOLIOS and RISK AVERSION

We just finished STEP A: find all the feasible portfolios

STEP B: Determine the optimal portfolio. This can be done in several ways:
 Determine your target risk σ
 Determine your target expected return
• E.g., Harvard has a target E of 6.25%
 Determine your risk aversion
• This is based on your trade-off between E and σ
• It is usually done with a utility function, e.g., quadratic utility

• U is increasing in E, and decreasing in σ

• A is the coefficient of risk aversion

5-27
 THE OPTIMAL PORTFOLIO

STEP A  The optimal portfolio P must be on the efficient frontier. If we can invest
in a risk-free asset, the efficient frontier is the CML (the line between rf and T)
 Let w = fraction invested in T (risky assets), 1 – w in the risk-free asset. Then:

 The optimal portfolio P is the solution to the following problem:

The solution is:

Observations:
 More risk averse investors (with higher A) invest less in the risky asset
 No matter how risk averse an investor is, he should still invest at least a small
fraction in the tangency portfolio (the risky assets)!

5-28
 THE OPTIMAL PORTFOLIO

Example: You have quadratic utility function with coefficient A = 7. You must
choose a portfolio of a risk-free T-bill and the tangency portfolio T

Suppose the tangency portfolio T has 58% stocks and 42% bonds

Solution: The optimal portfolio weight is

 You should invest 75% in the tangency portfolio, and 25% in the T-bill
 Should have 75% x 58% = 43.5% in stocks, and 75% x 42% = 31.5% in bonds.
Optimal portfolio: 43.5% (stocks) + 31.5% (bonds) + 25% (T-bill)

5-29
4. PRACTICAL ISSUES IN ASSET ALLOCATION
 C. Practical Issues – Estimation

What are the inputs to MPT?

 Expected returns
 Covariances
• Standard deviations
• Correlations
In MPT, we assume the inputs are given

In practice, inputs must be estimated

 Hardest and the most important part of asset allocation
 Estimation errors are large especially for expected returns, even with lots
of data
 Moreover, MPT is extremely sensitive in the inputs

5-31
 INPUT ESTIMATION

How do we estimate the inputs?

 Use historical data, although this is backwards-looking
• “Financial forecasting is like driving a car blindfolded with directions
from a passenger who is looking out the back window” (Werner
DeBondt)
 For expected returns we can also use
• A model, such as CAPM or APT
 Or the combination between historical data and a model
 This is called “shrinkage”
• Use Gordon growth formula P = D/(r-g)  r = D/P + g
• Adjust to the fact that expected returns vary with the business cycle
 Return predictability

5-32
RETURN PREDICTABILITY

5-33
 SMART INVESTORS + VIEWS

MPT assumes all investors have the same inputs

But what if some investors are smarter than others?

 Active management

Diversification with views: Black & Litterman model (developed in 1991 at

GSAM)
 If you have no views  Hold the market portfolio
• On average, investors do hold the market portfolio
• Are you smarter than the average investor?
 If you have views  Move away from the market portfolio
• Long-short strategies in the direction of your views
 Deviation proportional to your confidence
• No views about some assets  Same ratio as the market portfolio
 Smart investors should diversify, too!

5-34
5. CASE STUDY: Harvard Management Company
 HARVARD’S ENDOWMENT

Problem: How to allocate Harvard’s endowment of $18.2 billion and get high
returns while keeping risk reasonably low

Jack Meyer arrived in 1990: Policy Portfolio

 Agreed with the Board
 12 asset classes, with relatively fixed weights
 Sometimes tactical allocation bets (change the weights)

Within each asset class, active management

 Avoid directional bets (similar to hedge funds)
 Interesting compensation scheme for each class based on benchmarks and
clawbacks

Compute efficient portfolios for various target expected returns: mean-variance

analysis with constraints
 Also, portfolio stress tests

5-36
 HMC: POLICY PORTFOLIO

The Policy Portfolio in October 2000

5-37
 POLICY PORTFOLIO: EVOLUTION (2000)

The evolution of the Policy Portfolio

5-38
 HMC: MPT ASSUMPTIONS
HMC invests in 11 asset classes + cash. They use Modern Portfolio Theory
(MPT), using as inputs the real expected returns, standard deviations, and
correlations (in %/year):

5-39
 EFFICIENT FRONTIER: 12 ASSETS, CONSTRAINED
NO SHORT SELLING (EXCEPT CASH)

MPT with no-shorting constraints (except -50% cash): 22 portfolios along the efficient frontier

5-40
 EFFICIENT FRONTIER: 12 ASSETS, CONSTRAINED
NEAR POLICY PORTFOLIO
MPT with constraints near the Policy Portfolio: 8 portfolios along the efficient frontier

5-41
 HMC: LONG-SHORT POSITIONS
Examples of Long-Short Positions

5-42
 HMC: LONG-SHORT POSITIONS

Examples of Long-Short Positions (cont’d)

5-43
 HMC: PORTFOLIO STRESS TESTS

Portfolio Stress Tests ($ in Millions)

5-44
 HMC: ENDOWMENT PERFORMANCE (2000)

Endowment Performance (% returns)

5-45
 HMC: ENDOWMENT PERFORMANCE (2009)

Endowment Performance, updated 2009

5-46
IN CRISES CORRELATIONS GO UP!

5-47
 POLICY PORTFOLIO: EVOLUTION (2014)
Policy Portfolio, updated 2014

5-48
 CASE UPDATE: Harvard Management Company

What role did liquidity play in Harvard’s asset allocation?

 In 2008, HMC held an illiquid portfolio:
• 55% in hedge funds, private equity, and real assets
• 15% in emerging-market equities and high-yield bonds
• 30% in developed-world equities and fixed income
 Desperate for cash, HMC tried to sell some of its $1.5 bn private equity
portfolio (including Apollo Investment and Bain Capital), but buyers
demanded huge discounts, of 50%
• Discussion with Jane Mendillo in Vanity Fair
 Harvard had to slash budgets, introduce hiring freezes, postpone the
planned Allston science complex, etc.

5-49
6. OTHER TOPICS
 INTERNATIONAL ASSET ALLOCATION

International assets provide diversification. Moreover, historically they have had

good returns and low correlations with U.S. assets
 Is this correlation stable over time?

There is a home bias puzzle: e.g., the U.S. stock market accounts for about a half of
world stock market, but over 90% of U.S. equity wealth is invested in U.S. stocks!
 It could be due to superior information about home stocks, but cannot be
the whole story

Should one hedge foreign exchange risk? In the short run:

 It usually decreases risk
 But it also usually increases correlations

In the long run, should not hedge: long-run exchange rates are driven by inflation,
and stocks are a good hedge against inflation

5-51
 LIFE-CYCLE INVESTING

When doing your analysis, use real returns to account for inflation!

Specific needs require dedicated specific assets: duration matching

 If you have a liability, e.g., you will pay tuition for your child in 20 years, you could match it
with a bond of the same duration (20-year bond), to eliminate interest rate risk

Recognize your tolerance / capacity for risk

 When you are young and have a decent income, it is a good idea to have more of your
wealth put at risk (≈ 65%)
• Stocks are less risky in the long run
• Can use up to 20% of that to speculate (mad money!)
 When older, keep a good chunk of your wealth in cash and bonds (≈ 10% + 50%)
• Stocks are riskier in the short run
• Still, you should invest some amount in stocks (≈ 25%)

The more your labor income is correlated to the market, the more you should diminish your
holdings of stocks
 If you were a finance professor, should you short banks?

5-52
 TOPIC 5: QUESTIONS

• If you only care about the risk and return of your investments, how much
should you invest in each asset?
• How does your strategy depend on the correlation between two assets?
• Does adding new assets to your portfolio improve your risk-return tradeoff?
• How sensitive is the tangency portfolio to the inputs (E,σ, ρ)?
• How do we determine the inputs in practice?
• Did Harvard make money because it was smart or patient?
• How did Harvard’s endowment evolve over time? Why?
• If you are smart, should you still be diversified?
• Should you change the proportion between bonds and stocks as you get
older?
• Should you short your own company?

5-53