Reference no: EM132395314

Assignment

(You will likely need to replace one of the X variables due to the high XX correlation.  This indicates X variable information overlap and colinearity which will likely not produce a good regression model for the next assignment. The decline in revenue is cycle (change in the revenue growth rate over several years). Population likely does not express seasonality but does express positive trend and cycle.  The revenue ACFs do indicate 4th and 8th lag slight spikes from seasonality in the data.)

Turn this assignment in as a Word document with the Minitab results and plots copied and pasted into the document. In last assignment you use Population and Real Disposable Income (RDI) as my x variables, you need to replace one of these variables due to the high XX correlation in assignment.

Assignment is to determine the foundation for a good company revenue regression model.  This foundation ties the company performance over time to general economic conditions.  First, run the regression for all of the variables (dependent Y variable Company Revenue, both independent X variables) in a single regression model in Minitab.  Be sure to have the company revenue and each of the X macro economic in separate columns of equal length.  That is, each column should have variable data that begins with the same date and ends at 12/30/2018 and have a variable label at the top.

Go to Stat/Regression/Regression/Fit Regression Model to obtain the regression menu.  Under Responses select the company revenue under the left list of data columns. Under Continuous predictors in the regression menu select each of the two macro variables.  Under the Results button on the menu make sure the Durbin-Watson statistic box is checked (we will discuss this in our next session).  Select OK and OK again to run the regression model.

Find the model results on the desktop including the Regression Equation, Coefficients and Analysis of Variance tables. Copy an paste each of these tables in your assignment Word Document submission.  In the coefficient table ensure that the signs of the coefficients are the same as the X variable correlation signs for each X variable in assignment 3.  That is, if the X correlation to Y is negative the coefficient sign should be negative as well.  If the coefficient sign has flipped from the variable correlation with Y replace the X variable with another macroeconomic variable data series.  Re-run the model for consistent coefficient to correlation sign results.

Next, ensure that each of the macro variables (X variables) have a t-value of absolute 1.960 or greater.  (negative t-values are ok as long as they are greater in absolute terms) If not, replace the X variable with one that is significant (t-value of 1.960 or greater).

Ignore the constant term sign and significance at this point.

Finally, examine each X variable VIF (Variance Inflation Factor) for a value of 5 or less. If it is greater and have a single digit t-value (say 1.98) you should find another X variable to replace it.  VIF of 6 or 7 may be acceptable if the t-value is say 15.60 (greater than single digit).

The sign flip check above and the VIF check are to detect multicollinearity in the regression model.  If multicollinearity is present then the model results are useless.  The t-value check above is to determine if variables belong in the model (tests for statistical significance).  Again, if the model includes and insignificant X variable then the results are again useless.

Regardless, find variables that will work together to produce good estimates of the X variable.  The coefficient of determination (R-square adjusted) will indicate the percent of company revenue that is explained (determined by) the X variables in the regression model.  Higher R-square adjusted values are preferred to lower values but the conditions mention above must be met.