Assignment: Modeling and Forecasting

For the following questions, make sure to state your testable hypothesis, present the model that you have developed to test your hypothesis, the appropriate statistical test for your hypothesis and the significance level of the test, and interpret the results of your test. Using the dataset Australia you wish to examine whether Australian bond rates (IAU) are influenced by US bond rates (IUS). You want to use the Exchange Rate (E) as a control variable.

a. First, present the plot of the IAU series versus time. What does the time series plot suggest about the stationarity of the bond rate series?

b. Run an Augmented Dickey-Fuller test of the null hypothesis of a unit root process using 3 lags. Do you fail to reject the null hypothesis of a unit root?

c. Assuming that you fail to reject the null, first difference the IAU series and Re-run the Augmented Dickey-Fuller test with 3 lags. What do you now conclude?

d. Run the Augmented Dickey-Fuller test on IUS and E.

e. Run the OLS model on levels or first differences as necessary to included stationary variables, if you include levels instead of differences, you should also include the first lag of IUS and E.

You now need to test whether your assumption of no serial correlation is correct. We can use the Bruesch-Godfrey test to examine this hypothesis. Basically, you are concerned that errors yesterday explain errors today, so you need to test whether today's errors are a function of yesterday's errors, while controlling for the independent variables.

f. Do you fail to reject or reject the null hypothesis of no serial correlation?

g. In your final regression, correct for stationarity and serial correlation as needed, what can you say about the impact of US bond rates on Australian bond rates?

Attachment:- Australia.rar