The KEEP statement specifies which variables in the data file are to be included in the OUT= data set. Only the time series and event variables can be specified in a KEEP statement. All the BY variables and the time ID variable DATE are always included in the OUT= data set; they cannot be referenced in a KEEP statement. If they are referenced, a warning message is given and the reference is ignored.

The variable list can contain variable names or name range specifications. See Variable Lists for details.

There is a default KEEP list for each file type. Usually, descriptor type variables, like footnotes, are not included in the default KEEP list. If you give a KEEP statement, the default list becomes undefined.

Only one KEEP or one DROP statement can be used. KEEP and DROP are mutually exclusive.

You can also use the KEEP= data set option to control which variables to include in the OUT= data set. However, the KEEP statement differs from the KEEP= data set option in several respects:

• The KEEP statement selection is applied before variables are read from the data file, while the KEEP= data set option selection is applied after variables are read and as they are written to the OUT= data set. Therefore, using the KEEP statement instead of the KEEP= data set option is much more efficient.

• If the KEEP statement causes no series variables to be selected, then no observations are output to the OUT= data set.

• The KEEP statement variable specifications are applied to each cross section independently. This behavior may produce variables different from those produced by the KEEP= data set option when order-range variable list specifications are used.