In fairness I've thought it was defended in a confusing way in the past.
Bob scrutinizes Krugman's posts more than anyone I know, so I was a little surprised to see him miss the whole point in this post. Here he suggests that Christina Romer is parting from Krugman on the confidence fairy.
Look there are a couple things investors could have confidence in:
1. Balanced budgets (ceteris paribus, that's a nice thing to see)
2. Maintenance of demand (ceteris paribus, that's a nice thing to see)
3. Maintenance of future demand (ceteris paribus, that's a nice thing to see)
Mocking the confidence fairy - as I have always read it - is mocking the idea that #1 is going to turn the economy around when the major problem is a lack of demand, not a lack of loanable funds. Call me crazy, but that's just what Krugman has referenced every single time. And there were times - like during the Bush or the Clinton years - when Krugman noted the benefits of balancing budgets. Context matters, people.
But don't take my word for it. Read this Krugman post where he spells it out to others that apparently don't get the point:
"Expectations and the Confidence Fairy
Some readers have asked whether there isn’t an inconsistency between my view that the Fed can promote economic recovery by changing expectations about future policy, and my ridicule of austerity proponents who invoke “confidence” as a reason to believe that austerity will actually be expansionary. But there isn’t really any inconsistency; it’s an orders of magnitude thing.
What the expansionary austerity types are claiming is that the indirect effect of austerity on confidence will outweigh the large direct depressing effect of cutting government spending now. That’s a very tall order. Consider a very simple New Keynesian model, like the one I used in my old Japan paper (pdf). This model assumes infinitely lived consumers with free access to capital markets, assumptions that would seem to be very favorable to the notion that changes in expected future policy matter. Yet even there, a perceived permanent fall in government spending will at best have zero effect on output; if there’s any notion that the cuts are temporary, they’ll be contractionary. Add more realism, and the odds of expansionary austerity get even worse.
By contrast, expectations-based monetary policy has no direct effect on the economy today, so any positives from expectations make it favorable over all. You don’t have to believe that the effects are really big to believe that they might be there.
Now, there is room for skepticism over the effectiveness of “credibly promising to be irresponsible” — which is why from the beginning of this crisis I’ve always favored using fiscal policy as the main answer, with unconventional monetary policy as a supplement. But the Fed should be doing what it can — and finally, it seems to be moving in that direction."
Now can anyone tell me with a straight face that Krugman and Romer aren't essentially on the same page? Just like Romer wrote: "Recent research suggests that New Deal programs may actually have had their primary impact on the economy by influencing consumer and business expectations of future growth and inflation."