Dummy variables and controlling for seasonal effects

Dummy variables and controlling for seasonal effects - Transcript

ECON50001 Business Forecasting
Computer Exercise 6 Dummy variables and controlling for seasonal effects
Object Estimate models with dummy variables and interpreting checking stability and including time trends Data Files pub wf1 Question 1 Prior to February 1 1966 pubs in the state of Victoria Australia closed at 6 00pm On that date the closing time was extended until 10 00pm Some groups within the community expressed concern that the longer opening hours would lead to a greater consumption of alcohol Relevant data quarterly data 44 observations running from 1958 3 to 1969 2 are provided in the data file pub wf1 Q Quantity consumed defined as retail sales of beer wine and spirits per head of population divided by an index of Melbourne beer prices Y Income P Price S1 S2 S3 S4 Seasonal Dummy variables representing March June Sept and Dec quarters respectively a Consider the model Qt 0 1Pt 2 Yt 1 S1t 2 S2t 3 S3t et

Estimate the above model using least squares and report the results You will need to create the dummy variables for the seasons i How would you interpret the coefficients of the variables S1 S2 and S3 ii Do the coefficients have the expected signs Comment on the significance of the individual coefficients iii Create a dummy variable which is equal to 0 for all the quarters prior to the pubs opening longer hours and equal to 1 for all the quarters in which the pubs were opening longer hours including the quarter where it was open longer hours for 2 3 months

iv Did extended pub hours increase alcohol consumption

An alcohol industry representative suggests that the whole relationship may have changed in 1966 Q1 Include an interaction dummy variable for each of the other explanatory variables in the model And discuss the result vi Use a Chow breakpoint test to see whether the rep s claim is valid

v

vii Using the model from iii create a time trend variable and include it in your model as well as the other explanatory variables we are assuming there is some type of trend associated with the Quantity consumed defined as retail sales of beer wine and spirits per head of population Assume the trend takes the following two functional forms a Linear b Quadratic vii Which form do you think is the best at explaining the relationship of time with Quantity consumed defined as retail sales of beer wine and spirits per head of population What can you use to help you make this decision Note To generate seasonal dummy variables in Eviews select Quick Generate Series and use the equation s1 seas 1 Similarly for s2 and s3 To generate time trend variables in Eviews select Quick Generate Series and use the equation time trend