Homework answers / question archive / Time Series Assignment 1  UCD’s late submission policy will be applied (https://www

Time Series Assignment 1  UCD’s late submission policy will be applied (https://www

Statistics

Share With

Time Series Assignment 1 

• UCD’s late submission policy will be applied

(https://www.ucd.ie/governance/resources/policypage-latesubmissionofcoursework/).

• Only PDF uploads will be accepted. You can hand write if you like, but must come as PDF and not JPEG, DOCX etc
• You MUST also provide the do file(s) or commands embedded in your document to show how you came by your answers. You should also annotate the do files to show that you understand what you did. If you are using R or any other software, the equivalent command files will be sufficient.
• No late work will be accepted once the solutions are posted ( a week following the deadline). Work submitted after the deadline and up until the solutions are posted will automatically be docked one grade point (A+ → A). 
• If you have extenuating circumstances for late submission, communicate these early.

No requests for extensions the night before. 

• Answers will be checked for originality: 
  • • Don’t just copy from the book: Rewrite in your own words to illustrate your understanding. 
    • Don’t just copy from other sources: Use proper citation. No block quotations.
    • Don’t copy from each other – identical answers will both be marked zero. 
• All questions (and parts of questions) carry equal marks

The file sales.dta contains 157 weekly observations on sales revenues (sales) and advertising expenditure (adv) in millions of dollars for a large US department store for 2005-2007. The weeks are from December 28, 2004, to December 25, 2007. 

1. 1. Graph the series for sales. Does it appear to be stationary, trend stationary or stochastic?
   2. Formally test for the presence of a unit root. Explain what version of the test you used and why you used it.
   3. Examine the autocorrelation function and propose one or two possible models that may have generated the series.
   4. Estimate the model(s) proposed in part (c).
   5. Assess your model(s) using if possible formal tests and/or an information criterion. 
   6. Estimate the model you deem most appropriate and estimate it again on the first 100 observations. 
   7. Generate forecasts for the next 57 observations, and plot forecasted and actual observations. Assess the forecasting power of your model.
   8. Estimate a VAR model of sales and advertising. Be sure to justify your choice of lag length.
   1. Do sales Granger cause advertising or vice-versa of both? Carefully state the null hypothesis of any tests you employ and carefully state the conclusion.
   10. Generate a set impulse response functions and discuss their economic interpretation.  
   11. Generate a forecast of sales from the VAR and compare it to the univariate forecast of part g).