Here is code that fills out all missing months in the years in the data (data must be sorted as in the BY statement): data want; set temp; by col1 col2 col3 year; _prev_month=lag(month); _value=value; value=0; if first.year then do month=1 to month-1; output; end; else do month=_prev_month+1 to month-1; output; end; value=_value; output; if last.year; value=0; do month=month+1 to 12; output; end; drop _:; run; This is maybe not the most elegant solution, but it may execute faster than the other solutions suggested. ... View more

Alternatively   data tempp; length first last col_a col_b $10; input first last col_a col_b; datalines; jone smith aa bb jone smith cc bb jone smith dd cc jone smith ee aa peter paul zz dd peter paul pp dd peter paul aa cc peter paul bb aa ; run; data want; if _N_ = 1 then do; dcl hash h(dataset : "tempp", multidata : 'Y'); h.definekey('first', 'last', 'col_b'); h.definedone(); end; set tempp; by first last; if h.check(key : first, key : last, key : col_a) = 0; run;   Result   first last col_a col_b jone smith aa bb jone smith cc bb peter paul aa cc ... View more