EC7105 Business and Financial Forecasting.

Course work for EC7105 Business and Financial Forecasting. This coursework should be performed individually. There are three exercises below, all should be undertaken. Each piece of work should focus on a written account of the techniques as well as the pure computation involved. The main part of the marks will be awarded for the written description of the work.

You may obtain suitable data from any source, including the course Blackboard site.

1. Using a suitable piece of exchange rate data perform the Box Jenkins identification process and then use the resulting model to forecast the exchange rate 5 periods into the future.

2. Using some data from, at least three points on an interest rate yield curve derive the common factors underlying the original data. How would you choose the appropriate number of factors to use in a forecasting model and how would you build a factor augmented VAR.

3. Using a suitable stock market index build a model of the conditional variance and then forecast this variance for 5 periods into the future.

Question 1 A) Outline the two basic time series forecasting models which are the building blocks of forecasting models (15) B) Give an account of the Wold decomposition and explain why it is important. (15) C) How would you use the autocorrelation function and the partial autocorrelation function to identify the correct form of ARMA model to estimate? (20) [50 marks] Question 2 A) Explain what is meant by a conditional variance and how would you estimate it? (15) B) What is a GARCH in Mean model? (15) C) What is meant by a conditional covariance matrix and how would you estimate it? (20) [50 marks] Question 3 A) What makes a good forecast? (10) B) Explain how you would use the Root Mean Square Error, the Mean Absolute Error and the Theil inequality coefficient? (20) C) Would a combination of individual forecasts be expected to perform better than an individual forecast and how would you derive such a combination? (20) [50 marks] Question 4 A) What is the probability integral transform and how would you use it to asses a density forecast? (25) B) Would you expect a combination of density forecasts to perform better than the individual forecasts and how would you achieve such a combined forecast? (25) [50 marks]