John M. Griffin

Empirical Methods in Finance

University of Texas at Austin

The Warm-up. Assignment #1

Goal: For the student to become familiar with some basic properties of returns, volume, price ratios, and exchange rates. In the process the student should gain human capital by learning how to pull Datastream data and to do computations using this data. Recommended software packages for assignment… Stata, SAS, Eviews, ‘R’, using excel is not acceptable.

Prepare a neat and well organized report of your findings with computer output and some exhibits in the back. Attach your programming code as well. This assignment may take longer than you think it is recommended to get started sooner rather than later.

Use Datastream or another data provider. Download the local indexes, volume equivalent (more below), P/E ratio, dividend to price ratio, and dollar/country cross rate for your assigned countries as far back as the data is available on Datastream (generally from December 31, 1975 or later in many countries) to present. (Daily and Monthly). There are many return series to choose from state the indices you use. (To obtain some of the data you need, you may need to obtain the datastream index).

To calculate the following series you might want to use SAS, STATA, or Eviews to begin familiarizing yourself with it. Calculate the following. (Unless explicitly stated, calculate these for the monthly data). ‘?’ indicates you can pick a country of your own. Answer all of the questions.

Each person picks one country set below:

a) US, Japan, Brazil, ? (pick another as well)

b) US, Thailand, UK, ?

c) US, Mexico, Chile, ?

d) US, Germany, China, ?

e) US, Czech Republic, Russia, ?

f) US, Hong Kong, South Korea, ?

g) US, South Africa, Taiwan, ?

1. One page display plots of the monthly local return index levels (RI), turnover, P/E ratio and the exchange rate with respect to the dollar for each country (except the US). (Hint put all countries in the same graph). On a separate page display plots of the changes in the exchange rates, simple returns, as well as absolute value returns. Use color charts.
--In one paragraph simply state your opinion about what are possible explanations for the large variation in P/E ratios through time?

2. Using monthly data, report the mean, variance, skewness and kurtosis for the following:
a) local currency country index returns

b) absolute returns
c) volume (compute turnover, volume/number of shares outstanding)

d) changes in exchange rates

3. Report the correlation matrix for your monthly returns, absolute returns, turnover, and changes in FX rates with all countries included.

What is the relation between returns across markets. What do you learn from in-country relationships between returns, volatility, turnover, and FX rates?

4. What can we learn from the above analysis in 1-3? Is there persistence in these series? How do inferences differ across countries/series. Are returns distributed normally? Can you make some interesting observations from your analysis?

5. Calculate the average monthly and buy and hold returns in local and dollar terms from June 1997 to present. What do local returns correspond to for a US investor? Dollar returns?
Would a US investor investing in these countries be ahead? What about for the Asian tiger countries?

6. (Most important!) Compute autocorrelation coefficients for up to 12 lags for the index returns and changes in exchange rates.

a) What do the ACF& PACF look like?

b) Are these stationary series?

c) Now repeat a) with daily data

d) What are the main differences between monthly and daily series?

e) What do you learn from your analysis here?

7. Fit a simple ARMA model to your daily country returns using the ACF and PACF.
Do the residuals look like white noise? (Kurtosis, skewness?)

8. Compute the autocorrelations for the squared residuals and absolute value of the residuals from 7.
a) Is there any evidence of non-linearity?
b) compute the correlation between these residuals and the absolute returns you have created earlier. Are there large differences between the two series? Explain why or why not?
c) How are the autocorrelations properties of absolute value returns different from those of returns?

9. Report the simple correlation matrix of your returns in the five markets using Daily data. Now calculate the correlations of the markets in months when the US market is up and months when the US markets are down.
Are markets more correlated in up and down months?

10. Bubbles? With your 4 markets. Calculate annual buy and hold returns each year. Map the P/E ratio of the December just prior to the year’s annual returns. Now make ONE scatterplot plot the returns of a market relative over the year on the x-axis and the prior year December P/E ratio on the y-axis.

a) Now refine your strategy by taking the P/E ratio relative to earnings in the previous 3 years (extrapolate from prior P/E ratios).

b) Estimate pooled time-series regressions of annual returns on prior P/E.

c) Now use monthly data with the prior month’s P/E and compare inferences between the annual and yearly data. Which R² is correct.

11. Write a summary paragraph describing the most interesting things you learned about international markets. A very important quality of a research is to learn what is interesting, what is not and why.

Don’t turn in masses of computer output. Make sure and present these results in a neat and well organized presentation. I.e. Part of learning to be a researcher is not to simply produce computer output but to understand this output and be able to convey the usefulness of the information to a reader.