Comment of the Day:
MOAR Musings on Whether We Consciously Know More or Less than What Is in Our Models...: "Ah, almost the usual...:
...This time I'm not typing that risk isn't a scalar quantity, I'm typing that confidence isn't a scalar quantity. Krugman was definitely talking about confidence that the national treasury is solvent even if it doesn't fiscally dominate the monetary authority and monetize the deficit.
Summers and Blanchard are talking about confidence in the private sector--confidence that banks are solvent or confidence that there are profitable investment opportunities or confidence that house prices will go up forever.
In particular, the sort of loss of confidence which he discusses is called "bond outflows" by Olivier Blanchard, Jonathan D Ostry, Atish R Ghosh, & Marcos Chamon. To me, Blanchard et al. strongly support Krugman. It is only by eliding the difference between confidence in different entities (which is exactly the distinction stressed by Blanchard et al.) that their paper can be used to counter Krugman.
Krugman's claim was not that confidence doesn't matter -- he noticed 2008. He is saying that the sort of loss of confidence caused by large public debt causes higher not lower GDP in an economy in a liquidity trap.
I think one basis for his critique of the confidence fairy is the conflation of confidence in different aspects of different entities. The other is the idea that we can forecast shifts in confidence (or will be able to in the future after complete failure in all of history) and can manage confidence.