Question 1

(50 points). Are the following statements true or false? Provide a short justification for your answer. (You
are evaluated on your justification.) Remember that a statement is false if any part of the statement is false.
A single correct counterexample is sufficient to show that a statement is false. If a statement is true, however,
you will only be given full credit if you provide a formal justification for why the statement is true.

Part 1
(10 points). Suppose that a stock is negatively correlated with the market. Then, its CAPM beta must be
negative.

Part 2
(10 points). Suppose that you are a mean-variance optimizer. The risk-free rate is rf . You can invest in the
risk-free asset, or another two risky assets, with expected returns E [r̃1 ] , E [r̃2 ]. You are not allowed to short
either asset. Suppose both risky assets have expected returns strictly lower than the risk-free rate:

E [r̃1 ] < rf , E [r̃2 ] < rf

A mean-variance investor would invest her entire portfolio in the risk-free asset.

Part 3
(10 points). If you are a mean-variance investor, you would never hold positive amounts of a risky asset
in your portfolio that offers an expected return below the risk-free rate; you could always construct a
mean-variance dominating portfolio by replacing the risky asset with the risk-free asset.

Part 4
(10 points, hard!). The CAPM states that expected returns depend on an asset’s loading on market risk.
Thus, if the CAPM holds, any asset with a standard deviation smaller than the standard deviation of the
market portfolio must have an expected return smaller than the market portfolio, since it is less risky than
the market. If this were not the case, investors would not want to hold the market portfolio, so markets
could not clear.

Part 5
(10 points, hard!). Suppose the CAPM holds. Suppose the variance of the annual market return, Var (r̃MKT ),
is 10%. Suppose the variance of ALZ corp stock is Var (r̃ALZ ), is 20%. The correlation coefficient between
ALZ corp stock and the market return is:

Cov (r̃ALZ , r̃MKT )

Then, the CAPM states that the expected return of ALZ corp must be greater than the expected return on
the market, that is,
E [r̃ALZ ] > E [r̃MKT ]

2
Question 2
(50 points). Tilting the market. There are 2 stocks, A and B, and the market. You believe their expected
returns and covariances are:

E[r̃A ] = 0.14
E[r̃B ] = 0.05
E[r̃m ] = 0.08
rf = 0.02

2 2
3 2 3
A A,B A,m 0.25 -0.01 0.05
⌃=4 2 5 4 0.01 0.0025
A,B B B,m = -0.01
2 0.05 0.002 0.04
A,m B,m m

Part 1
(10 points). Calculate the CAPM ↵’s of asset A and B. Are they positive or negative?

Part 2
(10 points, hard!). Let’s try to beat the market. Form synthetic assets Â, B̂, consisting of A minus its tracking
portfolio and B minus its tracking portfolio. The tracking portfolio should consists of market portfolio and
the risk free asset. Compute the covariance matrix of the synthetic assets  and Â, with each other, and
with the market.

Part 3
(10 points). What is the portfolio of Â, B̂ (NOT including the market) which maximizes your Sharpe ratio?
Does this portfolio beat the market?

Part 4
(10 points). Now, compute the portfolio of synthetic assets Â, B̂, and the market which maximizes your
Sharpe ratio. Does this portfolio beat the market?

Part 5
(10 points). Now, suppose you were wrong about the assets’ expected returns, and the CAPM is correct.
The covariance matrix is still ⌃, however, the expected returns of assets A and B are consistent with the
CAPM. What are the expected returns of the synthetic assets  and B̂ in this case? What is Sharpe ratio of
the portfolio you solved for in part 4? Does this beat the market?

3
Question 3
(40 points). Suppose that the world is described by a three-factor model, so stock returns depend on three
systematic risk factors, F̃1 , F̃2 , and F̃3 , which have mean 0.

r̃1 = 0.04 + 7F̃1 + 4F̃2 + 2F̃3 + ✏˜ 1

Part 1
(10 points). Compute the portfolio weights and expected returns for the pure factor portfolios for F̃1 . What
is the factor risk premium for factor 1?

Part 2
(10 points). Compute the portfolio weights which produce a portfolio with no systematic risk (i.e. 0 beta on
all factors). What is its expected return?

Part 3
(10 points, hard!). Is it possible to solve for the factor risk premia of factors 2 and 3? If not, why not?

Part 4
(10 points, hard!). If an asset has factor beta 3 with factor 1, factor beta 10 with factor 2, and factor beta 5
with factor 3, what should its expected return be under arbitrage pricing theory?

4
Question 4
(60 points, hard!) Suppose you are a hedge fund. You are trying to form a portfolio out of 2 risky assets
and a risk-free asset. You can take arbitrarily large short positions in the risky assets and the riskless asset
at no cost. However, your clients do not want you to take a position larger than 20% in asset B. Thus, your
portfolio is constrained to have wB < 0.2. The risk-free asset’s expected return is rrf = 2%.

Asset Expected Return Standard Deviation

Part 1
(10 points). First, ignore the constraint. Solve for the tangency portfolio.

Part 2
(10 points). Suppose your client wants to maximize returns, but can accept a maximum standard deviation
of 3%. Solve for the optimal portfolio for the client (that is, the weights wA , wB , wrf which maximize the
client’s expected return, conditional on the standard deviation being less than 3%). (Hint: remember that
the value of a constrained optimization problem can never be higher than the value of an unconstrained
optimization problem. So try just ignoring the constraint, and see what you get.)

Part 3
(10 points). Now, in the following 5 parts, suppose your client can accept a larger maximum standard
deviation of 10%. We will solve for the optimal portfolio in two steps. First, draw a diagram of the
unconstrained tangency portfolio, and the capital allocation line through the tangency portfolio. Find the
point on the capital allocation line (that is, the portfolio, described by weights wA , wB , wrf ), where the
weight on asset B becomes exactly 20%.

Part 4
(5 points). What is the standard deviation of returns of this portfolio that you solved for the in the previous
part?

Part 5
(5 points). What is the highest expected return that the client can attain, if she has a maximal standard
deviation which is equal to your answer to 4.?

Part 6
(10 points). Now, set wB = 20%. If we pick wA , and wrf = 1 - wB - wA = 1 - wA - 0.2, what is the client’s
portfolio expected return and standard deviation as a function of wA ? What choice of wA maximizes the
client’s expected return? Is this higher or lower than your answer to 4.?

5
Part 7
(10 points). Calculate the marginal Sharpe ratios (MB / MC) for asset A and B, at your answer to 6. Which
is greater? Explain the result to your client.