It's not unethical to give a placebo as long as you inform the patient that it is a placebo. (I probably wouldn't intentionally give an *active* placebo, although I'm starting to think of a lot of the meds I do prescribe this way.)
I don't have time to dig up the study now, but apparently even if you tell the patient "This is a sugar pill but I think it will help you," most of them will agree to take it and the placebo effect is quite strong.
I would totally give out sugar pills if I could find them. I'm considering telling people to take some kind of benign OTC vitamin pill instead, before I give them an SSRI.