Use the univariate with kernel option as below,

data tmp;
  do i=1 to 1000;
     x1=rannor(123);x2=3+2*rannor(123);
         z=ranuni(123)<0.3;
         x=z*x1+(1-z)*x2;
         output;
  end;
run;


proc univariate data=tmp noprint;
   histogram x / kernel(c = 0.50
                        l = 1 );
  inset n mean std min max skewness kurtosis/pos=ne;
run;

use kernel estimate in matlab

谢谢！正在学习SAS，还不会用MATLAB