So, to answer your question directly, what you propose would be possible, but the larger question is whether or not it's worth the effort. The calculations I offered show you roughly what the magnitude of savings could be. We're likely talking dollars saved per year, not hundreds, so if you're doing a purely economic analysis, the savings seem hardly worth it. Most of the power taken from the wall really does make it to the computer; only a small fraction is wasted by the power supply.
If you're thinking more along the lines of "think globally, act locally; every little bit helps," then the "ROI" analysis involves other considerations.