Alexander Enterprises manufactures plastic parts for the automotive industry. Its sales (in thousands of dollars) for 2003Q1 through 2007Q4 are as follows:

You are asked to forecast sales for 2008Q1 through 2008Q4.
a. Begin by preparing a time-series plot of sales. Does it appear from this plot that a linear trend model might be appropriate? Explain.
b. Use a bivariate linear regression trend model to estimate the following trend equation:
SALES = a + b (TIME)
Is the sign for b what you would expect? Is b significantly different from zero? What is the coefficient of determination for this model? Is there a potential problem with serial correlation? Explain.
c. Based on this model, make a trend forecast of sales (SALESFT) for the four quarters of 2008.
d. Given that actual sales (SALESA) for the four quarters of 2008 are:
2008Q1 4,667.1
2008Q2 4,710.3
2008Q3 4,738.7
2008Q4 4,789.0
calculate the root-mean-squared error for this forecast model in the historical period (2003Q1-2007Q4) as well as for the forecast horizon (2008Q1-2008Q4). Which of these measures accuracy and which measures fit?

Transcribed Image Text:

46389860 400 94 2 3 4 5 5 6 NA NA NA NA 09236163 123412341234 578614041585 668699368019 16913724 3 88901112223 333444444444 123412341234