What you want here is not a lag. It is a lead (look ahead). Due to the construction of the data step processing, it does not support a lead function. However, you can do a workaround like this
data have;
input var1 $ var2:mmddyy10.;
format var2 mmddyy10.;
datalines;
A 1/1/14
A 1/2/14
A 1/3/14
A 1/4/14
A 1/5/14
A 1/6/14
A 1/7/14
A 1/8/14
A 1/9/14
A 1/10/14
A 1/11/14
A 1/12/14
;
data want;
merge have have(firstobs=2 keep=var2 rename=(var2=var3));
run;