I have referenced Summers before on this issue, because I think his point is both important, perhaps to the point of being dispositive, and widely overlooked by the studiously indifferent. But here is his latest blog post anyway.
Summers et al very usefully quantified this point back in 2014. I really wish they would update that work, because it is very helpful. I would wager that 90% of people opining on the importance of QE are not aware of point 4 above. Bond yields fell despite an acceleration of net default-free rates duration supply.
Update on April 21:
I repeat myself out of monomania and enduring amazement at stubborn consensus, but here are two fun facts:
If Summers is right about the primacy of signaling, as I believe he is, then it is quite important, because the Fed demonstrated ahead of — and upon — the taper that it was fully capable of separating the signal from portfolio balance effect. This reflects, I believe, that there is not actually any requirement to put their money where their mouth is. Words are enough. If QE were about backing up signaling with purchases, then the Twist operation would have made no sense. The Fed might signal at a five-year horizon, but never at a 30.
Ben Bernanke actually accepts this point, as applied to the taper tantrum, even though the tantrum remains — wrongly — exhibit A in the QE enthusiast’s case. (When the Fed threatened to taper, yields went up. Right?) For Bernanke the tantrum was much more about signaling than about portfolio balance. That Bernanke believes this is not evidence that it is so, obviously. Bernanke has said a lot of things in the past, including about QE, that I would say are flat wrong. Still, I find it interesting that Bernanke holds this view, despite having strong prior reasons not to, and that this point is so narrowly commented upon.