Forecasting Lab - 24 September 2003

Description: This lab focuses on understanding seasonal adjustment and the seasonal component of economic time series data.

Goals: In this lab, you will

  1. Locate data on retail gasoline prices on the internet
  2. Retrieve these data
  3. Read data into a spreadsheet
  4. Estimate seasonal factors using regression models and moving averages
  5. Determine the size of the seasonal change in retail gasoline prices
In this lab, we locate, manipulate and analyze price data. The lab focuses on the data series:
CUUR0000SS47014 - Gasoline, unleaded regular, U.S. city average
This series is an index with a base period of 1982-84=100.

Instructions:

  1. Go to the Bureau of Labor Statistics (BLS) web site wher ethey provide access to Consumer Price Index (CPI) Data (http://www.bls.gov/cpi/home.htm)
  2. Under the section Create Custom tables (one screen) follow the link for Consumer Price Index - All Urban Consumers (current series). This starts a data access applet.
  3. From this menu, select US City Average in Box 1, Gasoline, unleaded regular in Box 2, Not Seasonally Adjusted and Seasonally Adjusted in Box 3. Click on the get data link and then use the More Formatting Options tool.
  4. Get the series from 1976 to 2003 using the Tables option in comma delimited format.
  5. Read this data into Excel. Note there will be two series in the output file.
  6. Clean the data so that only monthly observations remain. To remove the Annual and HALF1 and HALF2 variables:
    1. Sort on the variable Period
    2. Delete the annual and biannual values
    3. Move the remaining records
    4. Sort on the variable Period again, using the month ordering option
    5. Sort on the variable Year
  7. Check to see that the two series have the same number of observations.
  8. Add an Excel formatted date variable to the spreadsheet. Add a new column, and in the first row type "jan-76"; in the second row type "feb-76". Highlight these two cells and copy them to the end of the series.
  9. Graph both the seasonally adjusted and unadjusted series. What method does the BLS use to get rid of the seasonal component? (Detailed information available on-line.)
  10. What time series components can you identify in these series? Discuss the factors that affect retail gas prices.
  11. Calculate the seasonal difference in the series in each month.
  12. Describe the seasonal adjustment based on this difference.
  13. How else could you seasonally adjust the data?
  14. What relationship exists between the secular trend and the seasonal effect in time series data?