The easy way might be to use an 18F and LONGs.

The harder way is to break the 16bit number down at some
point, do some calcs and if the final result fits in the 16 bit size
bring it all back together.

Do you want to display the result only or save it to a variable?