Your report should be submitted to the school office in the Darwin Building by the deadline given. Also, you should upload your report, log file and any other files that you feel would be appropriate, e.g. graphs etc. to the reassessment dropbox on KLE.
Note that there are severe penalties for collusion. All the preparation and production of the answers should be your own work. You are advised to NOT copy other students work OR to let other students copy your work. Both actions are equally CRIMINAL. Examiners will actively check the scripts.
INSTRUCTIONS: Your answers should be written on the form of a report of an empirical analysis. It is an exercise in correctly reporting and interpreting the results. At each stage you should comment on the model you have estimated. Some examples of things that should be included are: ? Interpret the coefficients (do they have the signs you expected; are they of a reasonable magnitude; ie. do they make sense? Remember to use the units of measurement in your answers) ? Test your coefficients to see if they are significantly different from zero at the 5% and/or 1% level of significance using the appropriate critical-t from a t-table. ? Possibly compute the elasticities for each regressor, and comment on their relative magnitudes ? Assess the goodness of fit of your model: o Interpret the adjusted R2 and “overall” F statistic (i.e. do an F test). o If possible conduct a joint hypothesis test of your model. What you choose to test is up to you, but it must be a joint hypothesis test that uses the F statistic. This test must be something other than an overall test of your equation (i.e. all slope coefficients equal to zero) Include the output from STATA as an appendix to your report. Include any graphs as an appendix or within the body of the report. Marks will be awarded for clarity of discussion and comprehensive analysis but not for repetition so do not laboriously give details of the t-test for every coefficient etc.
The data is found in the file reassessment data-webdecs.dta and the exercise is based on the type work you should have completed in the labs. The file contains data from a company called WebDecs. WebDecs manufacture decorations and party items and are faced with a very variable demand for their products throughout the year. It is a competitive market and the firm wishes you to help to develop a model to explain their sales to help future production schedule. Quarterly data is available for the previous 7½years on the prevailing market price and the volume of sales.
salesvol sales volume for the quarter (000’s)
price price index for goods sold (2010q4 =100)
year year of observation
quarter indicator variable for quarter (=1 for quarter 1; 2 for quarter 2 etc)
trend integer taking 1 for first observation, 2 for second, 3 for third etc.
a) Explore the data characteristics
Graph the sales volume against price; sales and price against trend; and comment on patterns observed; find any means, standard deviations, correlations etc that are interesting and/or useful in the analysis.
b) Do prices affect the volume of sales?
Estimate the regression of sales volume on price. In a second regression include dummies to see if sales are affected by the quarters. Review the regressions along the lines indicated in the instructions to answer this question. A useful statistic to calculate might be price elasticity.
c) Is the relationship linear?
Run the regression of ln(salesvol) on ln(price). Review the revised model and comment on the non-linear relationship between sales and price. Explore whether the quarterly dummies are better included in this ln-ln model or the levels model as estimated in b).
d) Explore interactive dummy variables?
Construct some new variables that show the interaction between price and the quarters, e.g. create 3 new variables such as prq1=price*q1 etc and include in the regression. Review the revised regression and compare it with previous models.
e) Best model?
Outline which you think is the best model and your reasons why it is preferred. As part of the discussion produce a graph showing the prediction for sales volume from a suitable model(s)
WORD LIMIT: 800-1000 word (not including equations, tables, appendices etc). This is a report so marks will be allocated on the content and presentation of the whole entity rather than on each section.