Proc FREQ Remove Missing Values as Category

[sjj_]

How do I build a frequency table without including missing values as a category in the TABLES statement? I thought SAS doesn't include missing values as a category by default, but when I run the code below, that doesn't seem to be the case (missing values in 'ethnicity' are interpreted as a category in my output data set 'eth'). 

proc freq data=dt noprint ;
by firm year;
tables ethnicity / out=eth ;
run;

Sample output:

firm year ethnicity      percent

1    1990  .                 50

1    1990  caucasian  50

[PaigeMiller · Accepted solution]

When ethnicity='.' this is not a missing value for a character variable. It has a value of one character, which is a period.

If you want a character variable to have a missing value, it must be a single space, such as in this code:

ethnicity=' ';