CHAPTER 9

Answers to Problems
1. Consider the following data for two risk factors (1 and 2) and two securities (J and L):
λ0 = 0.05 λ1 = 0.02 λ2 = 0.04
bJ1 = 0.80 bJ2 = 1.40
bL1 = 1.60 bL2 = 2.25
a. Compute the expected returns for both securities.
In general for the APT, E(Rq) = 0 + 1bq1 + 2bq2
For security J: E(RJ) = 0.05 + 0.02 x 0.80 + 0.04 x 1.40 = .122 or 12.2%
For Security L: E(RL) = 0.05 + 0.02 x 1.60 + 0.04 x 2.25 = .172 or 17.2%
b. Suppose that Security J is currently priced at $22.50 while the price of Security L is $15.00.
Further, it is expected that both securities will pay a dividend of $0.75 during the coming
year. What is the expected price of each security one year from now?
Total return = dividend yield + capital gain yield

For security J, the dividend yield is $0.75/$22.50 = 0.033 or 3.33%

For security L, the dividend yield is $0.75/$15.00 = 0.05 or 5%

For security J, the expected capital gain is therefore 12.20% – 3.33% = 8.87%
For security L the expected capital gain is therefore 17.20% - 5.00% = 12.20%
Therefore:
The expected price for security J is $22.50x(1.0887) = $25.50
The expected price for security L is $15.00x(1.1220) = $16.83

2. Earlier in the text, it was demonstrated how the Fama-French three-factor model could be used
to estimate the expected risk compensation for a set of equities (INTC, JPM, and WFMI).
Specifically, using return data from the 1996–2000 period, the following equations were
estimated:
INTC: [E(R) – RFR] = (0.615)(λm) + (–0.640)(λSMB) + (–1.476)(λHML)
JPM: [E(R) – RFR] = (1.366)( λm) + (–0.387)(λSMB) + (0.577)(λHML)
WFMI: [E(R) – RFR] = (1.928)( λm) + (0.817)(λSMB) + (1.684)(λHML)
Using the estimated factor risk premia of λm = 11.50%, λSMB = –1.44%, and λHML = –5.40%, it was
then shown that the expected excess returns for the three stocks were 15.96%, 13.15%, and
11.90%, respectively.
a. Exhibit 9.8 lists factor risk prices calculated over two different time frames:
(1) 1981–2000: λm = 9.09%, λSMB = –1.10%, and λHML = 4.48%
(2) 1928–2000: λm = 7.02%, λSMB = 3.09%, and λHML = 4.39%.
Calculate the expected excess returns for INTC, JPM, and WFMI using both of these
alternative sets of factor risk premia.
For 1981-2000:
RINTC = 0.615 x 9.09 - 0.64 x (-1.1) - 1.476 x 4.48 = -0.32%
RJPM = 1.366 x 9.09 - 0.387 x (-1.1) + 0.577 x 4.48 = 15.43%
RWFMI = 1.928 x 9.09 +0.817 x (-1.1) + 1.684 x 4.48 = 24.18%
For 1928-2000:
RINTC = 0.615 x 7.02 - 0.64 x 3.09 - 1.476 x 4.39 = -4.14%
RJPM = 1.366 x 7.02 - 0.387 x 3.09 + 0.577 x 4.39 = 10.92%
RWFMI = 1.928 x 7.02 +0.817 x 3.09 + 1.684 x 4.39 = 34.71%
 b. Do all of the expected excess returns you calculated in Part a make sense? If not, identify
which ones seem inconsistent with asset pricing theory and discuss why
The excess return on Intel (INTC) in the 1981-2000 period seems reasonable, in that it is very
close to zero. INTC’s excess return in the 1928-2000 period is unusual in that it is negative.
The excess returns of J. P. Morgan (JPM) and Whole Foods (WFMI) appear to be extremely
large. This is partly due to the fact that we are using out-of-sample numbers to do the
estimating. That is, the regressions were estimated using 1996-2000 data, but the estimates
were made using data from much longer periods
c. Would you expect the factor betas to remain constant over time? Discuss how and why these
coefficients might change in response to changing market conditions.
No, we wouldn’t expect the factor betas to remain constant over time. The sensitivity of a
particular company’s return to a specific factor will change as the character of the firm
changes. For instance, growth companies don’t remain growth companies forever, but tend
to mature. Thus their factor betas would change to reflect the slowdown in growth but
increased stability of earnings.

3. You have been assigned the task of estimating the expected returns for three different stocks:
QRS, TUV, and WXY. Your preliminary analysis has established the historical risk premiums
associated with three risk factors that could potentially be included in your calculations: the
excess return on a proxy for the market portfolio (MKT), and two variables capturing general
macroeconomic exposures (MACRO1 and MACRO2). These values are:
λMKT = 7.5%, λMACRO1 = –0.3%, and λMACRO2 = 0.6%.
You have also estimated the following factor betas (i.e., loadings) for all three stocks with
respect to each of these potential risk factors:

FACTOR LOADING
Stock MKT MACRO1 MACRO2
QRS 1.24 -0.42 0.00
TUV 0.91 0.54 0.23
WXY 1.03 -0.09 0.00

a. Calculate expected returns for the three stocks using just the MKT risk factor. Assume a
risk-free rate of 4.5%.
RQRS = 4.5 + 7.5 x 1.24 = 13.8%
RTUV = 4.5 + 7.5 x 0.91 = 11.325%
RWXY = 4.5 + 7.5 x 1.03 = 12.225%
b. Calculate the expected returns for the three stocks using all three risk factors and the
same 4.5% risk-free rate.
RQRS = 4.5 + 7.5 x 1.24 + (-0.3) x (-0.42) + 0.6 x 0.00 = 13.926%
RTUV = 4.5 + 7.5 x 0.91 + (-0.3) x (0.54) + 0.6 x 0.23 = 11.301%
RWXY = 4.5 + 7.5 x 1.03 + (-0.3) x (-0.09) + 0.6 x 0.00 = 12.252%
c. Discuss the differences between the expected return estimates from the single-factor
model and those from the multifactor model. Which estimates are most likely to be
more useful in practice?
Assuming that the factor loadings are significant the three factor model should be more
useful to the extent that the non-market factors pick up movements in returns not
captured by the market return
d. What sort of exposure might MACRO2 represent? Given the estimated factor betas, is it
really reasonable to consider it a common (i.e., systematic) risk factor?
 Because the factor loadings on MACRO2 are zero for two of the stocks, it appears that
MACRO2 is not a systematic factor, i.e., one that generally affects all stocks. It may
represent industry- or firm-specific factors.

4. Consider the following information about two stocks (D and E) and two common risk factors (1
and 2):

Stock bi1 bi2 E(Ri)

D 1.2 3.4 13.1%
E 2.6 2.6 15.4%
a. Assuming that the risk-free rate is 5.0%, calculate the levels of the factor risk premia
that are consistent with the reported values for the factor betas and the expected
returns for the two stocks.
E(RD) = 5.0 + 1.2 x 1 + 3.4 x2 = 13.1%
E(RE) = 5.0 + 2.6 x1 + 2.6 x2 = 15.4%
Solving the second equation for 1 in terms of 2 , we get: 1 = [(10.4 – 2.6)/2.6]2
Substituting that into the first equation: 1.2[(10.4 – 2.6)/2.6]x2 + 3.4x2 = 8.1
Solving for 2, we find 2 = 1.5. Using this value we can determine from either equation
that 1 is equal to 2.5.

b. You expect that in one year the prices for Stocks D and E will be $55 and $36,
respectively. Also, neither stock is expected to pay a dividend over the next year. What
should the price of each stock be today to be consistent with the expected return levels
listed at the beginning of the problem?
Because neither stock pays a dividend, the total return is all due to price appreciation.
Therefore:
for stock D: P0x(1.131) = $55 P0 = $55/1.131 = $48.63
for stock E: P0x(1.154) = $36 P0 = $36/1.154 = $31.20
c. Suppose now that the risk premium for Factor 1 that you calculated in Part a suddenly
increases by 0.25% (i.e., from x% to (x +0.25)%, where x is the value established in Part
a). What are the new expected returns for Stocks D and E?
From part (a), the risk premium for factor 1 was 2.5%. The new risk factor is thus 2.5% +
0.25%, or 2.75%. The new expected returns are:
E(RD) = 5.0 + 1.2 x  + 3.4 x = 13.4%
E(RE) = 5.0 + 2.6 x  + 2.6 x = 16.05%
d. If the increase in the Factor 1 risk premium in Part c does not cause you to change your
opinion about what the stock prices will be in one year, what adjustment will be
necessary in the current (i.e., today’s) prices?
D: PD0(1 + 0.134) = $55 PD0 = $55/1.134 PD0 = $48.50
E: PE0(1 + .1605) = $36 PE0 = $36/(1.1605) PE0 = $31.02
5. Suppose that three stocks (A, B, and C) and two common risk factors (1 and 2) have the
following relationship:
E(RA) = (1.1) λ1 + (0.8) λ2
E(RB) = (0.7) λ1 + (0.6) λ2
E(RC) = (0.3) λ1 + (0.4) λ2
a. If λ1 = 4% and λ2 = 2%, what are the prices expected next year for each of the stocks?
Assume that all three stocks currently sell for $30 and will not pay a dividend in the next
year.
E(RA) = 1.1x0.04 + 0.8x0.02 = 0.06 or 6% E(Price A) = $30(1.06) = $31.80
E(RB) = 0.7x0.04 + 0.6x0.02 = 0.04 or 4% E(Price B) = $30(1.04) = $31.20
E(RC) = 0.3x0.04 + 0.4x0.02 = 0.02 or 2% E(Price C) = $30(1.02) = $30.60
 b. Suppose that you know that next year the prices for Stocks A, B, and C will actually be
$31.50, $35.00, and $30.50. Create and demonstrate a riskless, arbitrage investment to
take advantage of these mispriced securities. What is the profit from your investment?
You may assume that you can use the proceeds from any necessary short sale.
Note
A : 31.80 related to 31.50 expected to be undervalued I should be short know
B : 31.20 related to 35.00 expected to be overvalued I should be long know
A : 30.60 related to 31.50 expected o be undervalued I should be short know
In order to create a riskless arbitrage investment, an investor would be short 1 share of
A and one share of C, and buy 2 shares of B.
The weights of this portfolio are: WA = -0.5, WB = +1.0, and WC = -0.5.
The net investment is:
Short 1 share A = +$30
Buy 2 shares B = - $60
Short 1 share C = +$30
Net investment = $ 0
The risk exposure is:
Risk Exposure Factor 1 Factor 2
A (-0.5)x1.1 (-0.5)x0.8
B (+1.0)x0.7 (+1.0)x0.6
C (-0.5)x0.3 (-0.5)x0.4
Net Risk Exposure 0 0

At the end of the period the profit is given by:

Profit = ($30 - $31.50) + 2x($35 – 30) + ($30 - $30.50) = -$1.50 + $10 - $0.50 = $8

Problems 6–7 refer to the data contained in Exhibit 9.12, which lists 30 monthly excess returns
to two different actively managed stock portfolios (A and B) and three different common risk
factors (1, 2, and 3). (Note: You may find it useful to use a computer spreadsheet program (e.g.,
Microsoft Excel) to calculate your answers.)

6.
a. Compute the average monthly return and monthly standard return deviation for each
portfolio and all three risk factors. Also state these values on an annualized basis. (Hint:
Monthly returns can be annualized by multiplying them by 12, while monthly standard
deviations can be annualized by multiplying them by the square root of 12.)
Using a spreadsheet program we find for monthly returns:
Portfolio A: RA = 0.0187 or 1.87% Standard deviation A = 0.0559 or 5.59%
Portfolio B: RB = 0.01497 or 1.497% Standard deviation B = 0.0465 or 4.65%
Factor 1 average = 0.01148 or 1.148% Standard deviation 1 = 0.053 or 5.3%
Factor 2 average = 0.00035 or 0.035% Standard deviation 2 = 0.069 or 6.9%
Factor 3 average = -0.01287 or –1.287% Standard deviation 3 = 0.0497 or 4.97%
At annual rates:
Portfolio A: RA = 22.44% Standard deviation A = 19.36%
Portfolio B: RB = 17.96% Standard deviation B = 16.11%
Factor 1 average = 13.78% Standard deviation 1 = 18.36%
Factor 2 average = 0.42% Standard deviation 2 = 23.90%
Factor 3 average = -15.44% Standard deviation 3 = 17.22%
 b. Based on the return and standard deviation calculations for the two portfolios from Part a, is
it clear whether one portfolio outperformed the other over this time period?
It is not clear from the numbers in part a whether one portfolio outperformed the other
because A has higher return and higher risk.
c. Calculate the correlation coefficients between each pair of the common risk factors (i.e., 1 &
2, 1 & 3, and 2 & 3).
r(1,2) = 0.2207 r(1,3) = -0.5505 r(2,3) = -0.7531
d. In theory, what should be the value of the correlation coefficient between the common risk
factors? Explain why.
In theory the correlations should be zero, because we want the factors to be independent of
each other.
e. How close do the estimates from Part b come to satisfying this theoretical condition? What
conceptual problem(s) is created by a deviation of the estimated factor correlation
coefficients from their theoretical levels?
Factors 1 and two are not highly correlated, but it appears that there is significant correlation
between factors 1 and 3 and factors 2 and 3. Statistically speaking, this leads to problems of
multicollinearity, which would affect regression estimates

7.
a. Using regression analysis, calculate the factor betas of each stock associated with each of the
common risk factors. Which of these coefficients are statistically significant?
Using a basic regression package, we get the following results (t-statistics given in
parentheses):
RA = .0057 + 0.99x(Factor 1) - 0.201x(Factor 2) - 0.133x(Factor 3)
(2.92*) (22.27*) (-4.61*) (-1.90) adj. R2 = .967

RB = .0080 + 0.969x(Factor 1) + 0.059x(Factor 2) + 0.344x(Factor 3)

(2.85*) (14.67*) (0.91) (3.29*) adj. R2 = .900

*Significant at 5% level of significance

b. How well does the factor model explain the variation in portfolio returns? On what basis can
you make an evaluation of this nature?
 The adjusted R2’s in both regressions are very high (.967 and .900, respectively.) This means
the regressions account for over 90% of the variation of each portfolio’s returns around its
respective mean
c. Suppose you are now told that the three factors in Exhibit 9.12 represent the risk exposures
in the Fama-French characteristic-based model (i.e., excess market, SMB, and HML). Based
on your regression results, which one of these factors is the most likely to be the market
factor? Explain why.
Factor 1 is the most likely candidate for the market factor, because it has a large, significant,
and positive effect on both portfolios. Factors 2 and 3 have different signs in the two
equations.
d. d. Suppose it is further revealed that Factor 3 is the HML factor. Which of the two portfolios
is most likely to be a growth-oriented fund and which is a value-oriented fund? Explain why.
A positive HML factor loading indicates a value stock or portfolio. Portfolio B has a positive
loading on this factor, and therefore is the more likely candidate for the valueoriented
portfolio. A negative HML factor loading indicates a growth stock or portfolio. Portfolio A has
a negative loading on this factor, and therefore is more likely to be a growth-oriented
portfolio.