ECON 497A: Forecasting

This web page contains mathematical formulas and is therefore best viewed with Internet Explorer, because other web browsers may not correctly display the "Symbol" fonts involved.

Empirical homework assignment 5

The data for the homework assignment is available in EasyReg (former) default format (*) and in Excel CSV format (**). You may treat the time series involved as an annual time series, starting from year 1. This time series is generated as a stationary MA(q) process:

Yt = m + Ut - q1Ut - 1 - q2Ut - 2 - .... - qqUt - q,

where the errors Ut are independent N(0,s2) distributed.

  1. Determine an upperbound of q on the basis of the autocorrelation function.
  2. Determine q on the basis of information criteria.
  3. Take the maximum of the values of q found in parts 1 and 2, and estimate the model for this upperbound via the EasyReg ARIMA module, using all available data.
  4. Check the results in part 2 by testing whether the value of the upperbound of q in the previous model can be reduced.
  5. Re-estimate the model for the actual q using the first 490 observations, and forecast the last 10 observations, one-step ahead as well as recursively.
  6. Comment on the quality of the forecasts in both cases.

(*) Click on the link, and save the file as a text file on your hard disk via the "Save as .." option of your web browser. Note that Internet Explorer will try to save the file as a web page (*.htm format). You have to overrule that.

(**) If you use Internet Explorer and you click on the link the file will be automatically imported in Excel, provided that you have installed Excel. Then use the "Save as .." option of Excel to save the file on your hard disk. Netscape and Mozilla Firefox will display the file as a text file. Therefore, if you use Netscape or Firefox follow the procedure under (*).