BMAN71111 ASSET PRICING
January 2019
Section A
Instructions: Attempt all questions. Each question carries 3 marks. No marks will be deducted for incorrect answers.
FOR EACH QUESTION IN THIS SECTION, PLEASE PLACE A CROSS IN THE BOX NEXT TO THE ANSWER THAT YOU BELIEVE IS CORRECT. IN EACH QUESTION, ONLY ONE ANSWER IS CORRECT.
Question 1. A given portfolio p is mean-variance efficient when:
(A) It is NOT possible to find any other portfolio such that the condition SR[Rp] ≥ SR[Rother] holds, where SR[R] denotes the Sharpe ratio and Rother is the return on any other portfolio different from p.
(B) It is possible to find another portfolio such that the condition E[Rp] ≥ E[Rother] and Var[Rp] ≤ Var[Rother] holds with at least one inequality being strict, where E[R] denotes expected returns, Var[R] denotes the variance of returns, and Rother is the return on a portfolio different from p.
(C) It is NOT possible to find any other portfolio such that the condition E[Rp] ≥ E[Rother] and Var[Rp] ≤ Var[Rother] holds with at least one inequality being strict, where E[R] denotes expected returns, Var[R] denotes the variance of returns, and Rother is the return on any other portfolio different from p.
(D) It is NOT possible to find any other portfolio such that the condition E[Rp] ≤ E[Rother] and Var[Rp] ≥ Var[Rother] holds with at least one inequality being strict, where E[R] denotes expected returns, Var[R] denotes the variance of returns, and Rother is the return on any other portfolio different from p.
Question 2. Consider the fundamental pricing equation derived in the present value, dividend- discount model (also called Gordon’s model):
where the symbols have the same interpretation as in lecture notes. If the constant dividend growth rate g exceeds the required rate of return k, then:
(A) The resulting stock price is negative; obviously, this creates short-sale opportunities that show that Gordon’s model is invalid.
(B) The resulting stock price is negative; obviously, this creates the buy and long opportunities for high trading profits.
(C) The model does not admit a sensible solution; equivalently, Gordon’s model defines the stock price P as an infinite product of interest rates that – because g > k – fails to satisfy some mathematical conditions.
(D) The model does not admit a finite solution; equivalently, Gordon’s model defines the stock price P as an infinite sum of discounted future dividends that – because g > k – fails to converge to a finite value.
Question 3. Given the following implications of any asset pricing model studied in the lectures, which combination of the implications will be tested when we empirically test the CAPM?
(i) Beta is the only variable that can explain cross-sectional expected stock returns.
(ii) The correlation between stock returns and aggregate consumption growth determines cross-sectional expected stock returns.
(iii) The pricing errors are equal to zero on average.
(iv) Idiosyncratic risks do not matter for cross-sectional expected stock returns.
(v) The factors SMB (small minus big) and HML (high minus low) are the two key determinants of expected asset returns.
(A) (i), (ii) and (iv).
(B) (i), (iii) and (iv).
(C) (iii), (iv) and (v).
(D) (i), (ii) and (iii).
Question 4. Campbell and Shiller’s linearized rational valuation formula,
implies that:
(A) A high price-dividend ratio today must be caused by either forecasts of low dividend growth rates in the future or by forecasts of high future returns; moreover, given the current dividend, an increase in the price-dividend ratio today implies an increase in the stock price and therefore an increase in the current realized return; then future higher (lower) expected returns are associated with current higher (lower) realized returns.
(B) A high price-dividend ratio today must be caused by either forecasts of high earning growth rates in the future or by forecasts of low future long-term bond returns; moreover, given the current dividend, an increase in the price-dividend ratio today implies an increase in the stock price and therefore a decrease in the current realized return; then future lower (higher) expected returns are associated with current higher (lower) realized returns.
(C) A high price-dividend ratio today must be caused by either forecasts of high dividend growth rates in the future or by forecasts of lower future expected returns; moreover, given the current dividend, an increase of the price-dividend ratio today implies an increase in the stock price and therefore an increase in the current realized return; then future lower (higher) expected returns are associated with current higher (lower) realized returns.
(D) A high price-dividend ratio today must be caused by either forecasts of high dividend growth rates in the future or by forecasts of lower future expected returns; moreover, given the current dividend, an increase in the price-dividend ratio today implies an increase in the stock price and therefore an increase in the current realized return; then future lower (higher) expected returns are associated with current lower (higher) realized returns.
Question 5. Which one of the following statements regarding the expectation of short-term interest rates is correct?
(A) Suppose that the pure expectation hypothesis holds for the term structure of interest rates. An upward sloping yield curve clearly indicates that short-term interest rates are expected to decrease.
(B) Suppose that the liquidity premium hypothesis holds for the term structure of interest rates. An upward sloping yield curve must only indicate that short-term interest rates are expected to increase.
(C) Suppose that the liquidity premium hypothesis holds for the term structure of interest rates. An upward sloping yield curve can be consistent with the fact that short-term interest rates are expected to decrease.
(D) Suppose that the liquidity premium hypothesis holds for the term structure of interest rates. The yield curve must only be either monotonically increasing or monotonically decreasing.
Question 6. You have estimated the classical CAPM (regression) equation, Ri = Rf + βi [RM – Rf] + εi, on a sample of 300 stocks indexed by i over a sample of monthly observations for the period 1980-2010 and obtained estimates of 300 beta coefficients, . You then perform a second-step cross-sectional regression using data for each of the 300 stocks, with the model specification:
Ri = a1 + a2 + a3Var(ε(ˆ)i ) +ηi ,
where Ri is the sample mean of returns, Var(ε(ˆ)i ) is the variance of the residuals from each of the
300 initial, first-step regressions, and ηi is a white noise error. You find that the estimates in the
second step regression are: a(~)1 = 0.01 , a(~)2 = -0.20 , and a(~)3 = -0.84. The three estimates are highly
statistically significant. You then conclude that the CAPM is:
(A) Rejected by this sample because a negative estimate of a2 is inconsistent with any plausible empirical estimate of the market risk premium, and also because the significant estimate of a3 indicates that other factors besides market (systematic) risk are priced in the cross-section of 300 stocks.
(B) Not rejected by this sample because the estimate of the intercept a1 is consistent with the fact that we normally expect Rf (the risk-free rate) to be positive and small in any sample, because the estimate of a2 is consistent with the empirical fact that the market risk premium is generally negative, and because the significant estimate of a3 indicates that other factors besides market (systematic) risk are priced in the cross-section of 300 stocks.
(C) Accepted by this sample because all the estimates turn out to be highly statistically significant, as they should be if the CAPM holds.
(D) Neither rejected or accepted by this sample because there is insufficient information to draw any final conclusion; in particular one would need information on the sign and distribution of the estimated time series betas before making a conclusion on the validity of the CAPM.
Question 7. With reference to the following plot of historical returns on the NYSE value-weighted stock index, which one of the following statements is correct regarding the variance ratio statistic?
(A) The variance ratio statistic is defined as VR(k) = (12/k)[Var(Rk)/Var(R12)] where Rk denotes the return measured over a period of k months. With reference to the plot above, we expect that VR(k)<1 when k is a multiple of 12 months and that VR(k) is increasing in k, because returns display negative serial correlation.
(B) The variance ratio statistic is defined as VR(k) = (k/12)[Var(R12)/Var(Rk)] where Rk denotes the return measured over a period of k months. With reference to the plot above, we expect that VR(k)<1 when k is a multiple of 12 months and that VR(k) is decreasing in k, because returns display positive serial correlation.
(C) The variance ratio statistic is defined as VR(k) = (12/k)[Var(Rk)/Var(R12)] where Rk denotes the return measured over a period of k months. With reference to the plot above, we expect that VR(k)>1 when k is a multiple of 12 months and that VR(k) is increasing in k, because returns display mean reversion.
(D) The variance ratio statistic is defined as VR(k) = (12/k)[Var(Rk)/Var(R12)] where Rk denotes the return measured over a period of k months. With reference to the plot above, we expect that VR(k)<1 when k is a multiple of 12 months and that VR(k) is decreasing in k, because returns display mean reversion.
Question 8. Which one of the following observations is NOT among the main findings of Fama and French (1992) "The Cross-Section of Expected Stock Returns", Fama and French (1993) "Common risk factors in the returns on stocks and bonds" and Fama and French (1996) "Multifactor explanations of asset pricing anomalies"?
(A) Fama and French find that when one allows for variation in betas that is unrelated to size, the relationship between betas and average returns is almost flat, even when beta is the only explanatory variable.
(B) Fama and French find that the estimate of the risk premium on book-to-market equity is positive and statistically significant when book-to-market equity is the only explanatory variable. This finding implies that portfolios with low book-to-market ratios tend to have low average returns while portfolios with high book-to-market ratios tend to have high average returns.
(C) In a three-factor time series regression with the market factor, small-minus-big (SMB) factor and high-minus-low (HML) factor, Fama and French find that for 25 portfolios sorted according to size and book-to-market, the slopes on HML increase monotonically from high book-to-market to low book-to-market quintiles. This captures the book-to-market effect on cross-sectional expected returns.
(D) In a three-factor time series regression with the market factor, small-minus-big (SMB) factor and high-minus-low (HML) factor, Fama and French find that for 25 portfolios sorted according to size and book-to-market, the slopes on SMB increase monotonically from big size to small size quintiles. This captures the size effect on cross-sectional expected returns.
Question 9. Which one of the following statements is NOT consistent with the implications of Campbell and Cochrane's (1999) habit formation model?
(A) In good times, aggregate consumption experiences a series of positive innovations and thus the habit builds up gradually. In bad times, aggregate consumption experiences a series of negative innovations and thus the habit declines gradually.
(B) The surplus consumption ratio is countercyclical, which leads to procyclical risk aversion. A high surplus consumption ratio implies a low degree of risk aversion. This usually occurs in recessions when consumption falls toward the habit level and investors become much less willing to tolerate further declines in consumption.
(C) The habit formation model implies high variation in the stochastic discount factor, and the high volatility of the stochastic discount factor is mainly driven by time-varying surplus consumption ratios.
(D) Both the Sharpe ratio and equity premium are countercyclical in the model because in recessions investors are very risk averse and thus require high expected returns on stocks.
Question 10. With reference to the excess volatility puzzle, the following plot can be interpreted to show that:
(A) The actual, historical real stock price has been persistently more volatile than the prices computed from the rational valuation formula under perfect foresight (on future dividends) and both constant and time- varying real discount rates; therefore actual real stock prices appear to have been excessively volatile.
(B) The prices computed from the rational valuation formula under perfect foresight (on future dividends) and both
constant and time-varying real discount rates have been persistently more volatile than the actual, historical real stock price; therefore actual real stock prices do not appear to have been excessively volatile.
(C) The actual, historical real stock price has been negatively correlated with the prices computed from the rational valuation formula under perfect foresight (on future dividends) and both constant and time-varying real discount rates; therefore actual real stock prices appear to have been excessively volatile.
(D) The efficient market hypothesis does not hold because actual, real stock prices have significantly drifted away from the prices computed under perfect foresight of future dividends and a constant discount rate.
Section B
Instructions: Answer ONE question in all its parts. All parts together are worth 35 marks. Marks for all parts are shown below.
PLEASE USE THE SPACE PROVIDED ON THE QUESTION PAPER TO ANSWER ALL SECTIONS. The space provided is sufficient to give a concise and pertinent answer to each question and/or sub-point of a question. Extra answer booklets will not be provided – please use the spaces provided only. Answers written outside the spaces or on other booklets will not be marked.
Question 11 (Mean-Variance Algebra)
Suppose that your investment menu has two risky assets and a risk-free asset. The risk-free rate is 0.01. The first risky asset has a mean return of 0.06 and a standard deviation of 0.12. The second risky asset has a mean return of 0.05 and a standard deviation of 0.18. The correlation coefficient between returns on the two risky assets is 0.2.
11a. (8 marks) Assume (for this part only) that only the two risky assets are available for investment. Suppose that the expected return of a portfolio of the two risky assets is 0.08. Calculate the variance of this portfolio.
Consider another portfolio with portfolio weights (-52.83%,152.83%) on the first and second risky assets respectively. Calculate the variance and expected return for this portfolio. Compare the two portfolios using the mean-variance criterion.
11b. (15 marks) Assume that the investor wants to maximize the mean-variance objective function
by choosing the fractions of wealth to invest in the three assets. Here μ and σ are the mean and standard deviation of the investor’s portfolio, and Y is the risk aversion coefficient. Of the optimal amount of investment in the two risky assets, what fraction is optimal to invest in the first risky asset? Equivalently, find the optimal risky portfolio.
11c. (12 marks) The optimal risky portfolio obtained in part (b) is the tangency portfolio. (i) Illustrate the unique property of the tangency portfolio in the presence of a risk-free asset. (ii) Formulate an alternative optimization problem from which you can obtain the same optimal risky portfolio as in part (b). (iii) Solve the problem and verify that the resulting optimal portfolio is the same as the optimal risky portfolio in part (b).
Question 12. (The APT and Its Applications in Forming No-Arbitrage Portfolios)
Suppose assets A, B, C, and D are correctly priced and span the asset market, i.e., no arbitrage opportunities are possible by investing in A, B, C, and D. Each of these four assets is subject to three risk factors, 1, 2, and 3. No assets or portfolios have any idiosyncratic risk components. You have been given the following information about the expected returns and factor loadings of assets A, B, C, and D.
Asset
|
Expected Return (%)
|
bi1
|
bi2
|
bi3
|
A
|
14
|
1
|
0.6
|
0.4
|
B
|
12
|
0.4
|
0.9
|
0.3
|
C
|
8
|
0.5
|
0.3
|
0.2
|
D
|
6
|
0.2
|
0.1
|
0.05
|
12a. (12 marks) Compute (i) the rate of return for the risk-free asset that is not subject to any risk factor, i.e., λ0 , and (ii) factor risk premiums on the three risk factors, i.e., λ1,λ2 and λ3 . Find the equation for the expected return of any asset if the arbitrage pricing theory (APT) holds and express the equation using factor loadings and the factor risk premiums implied by the table given above. It is known that the risk premium on factor 1, λ1 = 0.2588 .
12b. (11 marks) Suppose that portfolio F has weights 35%, 45%, and 20% on assets B, C, and D respectively. It is known that the actual expected return on portfolio F is 10%. Discuss whether arbitrage opportunities exist or not. Create a portfolio strategy that generates arbitrage profits if such opportunities exist. Note: in constructing an arbitrage portfolio, you are allowed to trade only assets A, B, and D, portfolio F and the risk-free asset,i.e., trading asset C is not permitted.
12c. (12 marks) Suppose that portfolio G has factor loadings bG1 = 0.6, bG2 = 0.5 and bG1 = 0.3 . It is known that the actual expected return of portfolio G is 7%. Discuss whether arbitrage opportunities exist or not. Create a portfolio strategy that generates arbitrage profits if such opportunities exist. Note: in constructing an arbitrage portfolio, you are allowed to trade only assets A, B, and D, portfolio G and the risk-free asset,i.e., trading asset C is not permitted.