1

END OF CHAPTER EXERCISES - ANSWERS

Chapter 16 : Predicting Stock Returns

Q1 Does the EMH imply that you should never change the proportions in which you hold
stocks in your portfolio?

A1
Mean-variance portfolio theory shows that the optimal proportions held in each risky asset
depends on the variances and the correlations between all asset returns. If forecasts of these
“inputs” change through time, then so do the optimal proportions. Hence even if the EMH is
correct, you will alter the proportions held in each risky asset, over time. If tax laws change
this may also lead to new optimal proportions.

Q2 If the EMH were true, would there ever be any takeovers?

A2
This is a rather tricky question to answer. In an efficient market, all quoted share prices
reflect the present value of future earnings (dividends) from that firm, taken as a single entity.
If the managers of a firm-A are ineffective, then expected dividends will be low and the current
share price will also be low. If another firm-B thinks it can take control of firm-A’s physical
(and intangible) assets and increase profits then it may try to purchase firm-A’s shares. As it
does so, firm-A’s shares will rise to reflect the expected higher future earnings, if the firm is
taken over. Hence, the EMH does not rule out takeovers. If there is a takeover bid in the
offing then the market price of the stock of firm-A will rise to reflect in part, the increased value
of the combined firm, should the takeover bid be successful.

Q3 What is a random walk for stock prices? What does this imply for stock returns? Is a
random walk a useful statistical representation of stock price movements?

A3
A random walk is like a coin flip (using a fair coin) – if “heads” came up at the last flip, this
tells you nothing about what will happen on the next (or any future) flip. Applied to stock
prices it means that today’s price is of no use in predicting tomorrow’s price. Hence the
change in the stock price (ie. loosely speaking, the stock return) is unpredictable.

The random walk is a useful empirical representation of stock prices, particularly over short
horizons (e.g. 1-day, 1-month), although it does not capture all features of stock price data.
Changes in stock prices over one-day or even one-month are largely unpredictable from past
price changes – consistent with the random walk model of stock prices.

Q4 How can you test whether stock prices are mean reverting?

A4
The easiest way to test for mean reversion is to use regression analysis :

[1] Rt,t+N = α + β Rt-N,t + εt

where Rt,t+N is the return from t to t+N and Rt-N,t is the return between t-N and t. For example
for N = 2, equation [1] states that the stock return between say 1996 and 1998 influences the
subsequent return between 1998 and 2000. We can run a separate equation for different

© K. Cuthbertson and D. Nitzsche (2008) ‘Investments’, J. Wiley, 2nd edition

 2

values of N (i.e. for 1-year returns, 2-year returns, etc.) and see if mean reversion (i.e. β < 0)
occurs over specific investment horizons.

Q5 Where does the balance of the evidence lie on whether you can successfully “pick
winners” using regression equations?

A5
It looks as if there are certain empirical regularities, which enable you to predict stock returns
to some degree. It may be that the source of these ‘regularities’ change over time (e.g. in
some periods the dividend-price ratio has predictive power, in other periods it does not).
Hence ‘picking winners' is difficult and a risky business.

© K. Cuthbertson and D. Nitzsche (2008) ‘Investments’, J. Wiley, 2nd edition