The FLTDIV keyword indicates that all divide operations within expressions are computed in floating point and return a value of type float. If not specified or specified with *NO, then divide operations are performed in packed-decimal format (unless one of the two operands is already in float format).
If FLTDIV is specified, optionally with *YES, then all divide operations are performed in float format (guaranteeing that the result always has 15 digits of precision).