UNIVERSITY OF BRISTOL


MAY / JUNE 2019



School of Economics, Finance and Management



Examination for the MSc degree in
Accounting and Finance; Economics and Finance; Finance and Investment




EMPIRICAL FINANCE
(Module No. EFIMM0046)




Time allowed: TWO hours



Answer FOUR questions from the following FIVE.


All questions carry equal marks.


Approved calculators may be used.

Question 1

The Fama-MacBeth approach is mainly used to assess whether an asset pricing model
explains the cross-section of stock returns. In its first step, the approach estimates the
quantities of risk any given test asset carries (i.e., the betas with risk factors). This is usually
done with time-series regressions. Below you have two Stata outputs that show time-series
regressions to obtain the betas for the value-weighted Health industry portfolio (No: 10) and
the Finance/Money industry portfolio (No: 11). We use the Fama-French four factor model.

Health industry portfolio (No: 10)



Finance/Money industry portfolio (No:11)



a. What are the risk factors of the Fama-French four-factor model? Briefly discuss what
they reflect. (25 marks)

b. Discuss the time-series regression output of the Health industry (No: 10) in terms of
exposure to the Fama-French risk factors, statistical significance and economic/practical
significance. (30 marks)

c. Compare the exposure of the Health industry portfolio (No: 10) against the
Finance/Money industry portfolio (No:11) to the High-minus-Low (HML) risk factor.
(15 marks)

d. The Efficient Capital Market (ECM) hypothesis comes with assumptions. Briefly discuss
the Rational Information Processing assumption. (30 marks)

(Total 100 marks)
Question 2

You work for a large pension fund and are responsible for evaluating the performance of
several mutual fund managers.

a. Fund SPDR is an ETF tracking the S&P 500 index. What performance measures would
you use to evaluate this fund’s manager? (15 marks)

b. The Jupiter India Fund is an actively managed fund investing in companies that operate
in India. What performance measures would you use to evaluate this fund’s manager?
(15 marks)

You estimate the Jensen alpha of the Jupiter India Fund using iShares MSCI World ETF, an
ETF that tracks the MSCI World Index, a broad global equity index.


Note: r_jupiter_e: monthly excess return of Jupiter India Fund; r_msci_world_e: monthly
excess return of iShares MSCI World ETF. Numbers are in percent form, e.g. 0.5 means
0.5%.


c. Interpret the estimated coefficients. (25 marks)

d. Test the null hypothesis that the estimated alpha of Jupiter is equal to 0 against the 2-
sided alternative (state the null and alternative hypotheses, state the used significance
level, show how the t-statistic is computed, compare t-statistic to critical value and
decide whether to reject the null or not reject). (20 marks)

e. Ignoring its statistical significance, discuss the likely reasons for Jupiter’s alpha to be
large and positive in this specification. (25 marks)

(Total 100 marks)
Question 3

You are interested in forecasting the combined earnings of all S&P 500 companies (simply
referred to as earnings from now on).

a. Discuss why earnings are expected to be non-stationary. (15 marks)

The following table gives the results of an ADF test (“le” is the variable name of log earnings,
it is available at the quarterly frequency):



b. Interpret the results of the ADF test. (10 marks)

Suppose you want to use an AR(2) model to generate a forecast. Since you strongly suspect
log earnings to contain a unit root, you first difference it and estimate an AR(2) model, which
gives the following:


Note: first differencing a series in logs approximates the growth rate of the original series.

c. Based on the output, is there any indication that the AR(2) model is a useful forecast
tool in this case? (20 marks)

d. Suppose that earnings growth was 5% during both the first and the second quarter.
Based on the estimated AR(2), what is the expected earnings growth for the third and
the fourth quarter? (20 marks)



You want to test whether log earnings and log prices (log of the S&P 500 index) are
cointegrated. You run the Johansen test, which gives the following output:




e. Discuss why earnings and prices could be cointegrated. (15 marks)

f. Interpret the results of the Johansen test. (20 marks)

(Total 100 marks)
Question 4

a. Discuss the concept of timing the market and its implications for firms’ financing
decisions. Propose ONE testable hypothesis on the determinants of capital structure
based on this concept. (40 marks)

b. Give at least two estimation approaches for testing your proposed hypothesis in (a)
above. Compare and contrast your approaches. (40 marks)

c. Below is the error message that you received after running xtreg command in Stata.
Please identify and solve the problem. (20 marks)


(Total 100 marks)


Question 5

a. Explain what daily ‘abnormal returns’ (ARs) and ‘cumulative abnormal returns’ (CARs) in
event studies are. Why do researchers normally focus on CARs rather than ARs?
(25 marks)

The Eventus output below presents CARs around share repurchase announcements.



b. What is the level and significance of the immediate market reaction to repurchases? In
your answer present the null hypothesis in the Portfolio Time-Series and Generalized
Sign significance tests. (20 marks)

c. Explain what ‘271:240)’ (third raw, fourth column) means. (10 marks)

d. Explain why we may be interested in the stock price behaviour before or after the event
(for example, in windows (-60,-2) and (+2,+60)) in the test above). Referring to any
event study of your choice, give one example of a research question which can be
answered using the mean CAR in such windows. (30 marks)

The sample of 511 firms announcing share repurchases analysed above is split into firms
announcing large (Group 1) and small repurchases (Group 0), measured with the
percentage of shares the firm seeks to repurchase. Stata output from one of the follow-on
tests is presented below.



e. What is the aim of this test? What is its main result? (15 marks)

(Total 100 marks)





























































































































































































学霸联盟