The data set needs to be expanded so that each person has
an id number to be used in the repeated statement. (You could
also just use a "freq count;" statement in genmod instead of the
data step "do i=1 to count;", but GEE then says there are only 14
clusters when there really are 1306.)  Anyway, the following code
works on my machine.

data one;
options nodate nonumber ps=2000;
input race $ y count;
do i=1 to count;
output;
end;
cards;
b 0 119
w 0 1070
b 1 16
w 1 60
b 2 12
w 2 14
b 3 7
w 3 4
b 4 3
w 4 0
b 5 2
w 5 0
b 6 0
w 6 1
;
data two;set one;
id=_n_ ;
proc genmod;
class race id;
model y=race / dist=poisson link=log type3;
repeated subject=id / type=ind;
run;