tl;dr: Some people named Hróbjartsson and Gøtzsche did a meta analysis on clinical trials suggesting that the placebo effect is solely a reflection of bias or rigor in the given experiment, not of some physiological, "mind over matter" phenomenon.
More particularly, from the submitted article, "There is no measurable physiological benefit from placebo interventions for any objective outcome. There is a measured benefit for some subjective outcomes (mostly pain, nausea, asthma, and phobias), but the wide variation in effect size suggests this is due to trial design (and therefore bias) rather than a real effect." In my words, it might be expected that one would see a placebo effect if the clinical records are based on patient self-reports of, e.g., pain or nasal congestion, but a placebo effect would be quite astounding if they were statistically significant over multiple double-blind trials in detection of a chemical in a blood test.
and the link here draws a distinction that I think is important, the distinction between injury and pain. The paradox of pain, as this article says, is that patients with severe injury have little pain, while other patients with possibly no injury at all have severe pain. Granting that sham treatments (placebos) can reduce patient perception of pain, how do we establish that they produce more healing of injury, more consistently, than the passage of time as the body's evolutionarily selected healing mechanisms work?