Converting Time Series Data to Different Periodicities
Statistical agencies may report time series data on a monthly
basis. This data can be converted to quarterly data or annual data by
using a suitable conversion method. For example, a monthly price index
can be converted to a quarterly price index by computing 3month
averages. Monthly data on the volume of retail sales
can be converted to quarterly or annual data by summation.
With SHAZAM, the SUM function on the GENR
command can be used to convert time series data to different
periodicities. The command format is:
The SUM function sums up n successive
observations on the variable var starting at observation 1.
Example
Suppose monthly data for variable VARM is available
for the sample period 1986M1 to 1996M12 (132 monthly observations).
The SHAZAM commands below convert the data to quarterly data by
using 3month averages.
SAMPLE 1 132
READ (file) DATEM VARM
* Set the sample to the number of quarterly observations
SAMPLE 1 44
* Convert to quarterly data by computing monthly averages
GENR VARQ=SUM(VARM,3)/3
* Generate a quarterly date variable starting in 1986Q1
TIME 1986 4 DATEQ
* Write the results to a data file
WRITE (newfile) DATEQ VARQ
STOP

The WRITE command is used to save the new data
to a data file.
[SHAZAM Guide home]
