Mankiw's post on Paul Krugman's analysis of New Deal policies caught my eye:
Paul Krugman has posted a nice note analyzing New Deal wage policies using the model of aggregate supply and aggregate demand familiar to teachers of undergraduate macroeconomics. The essence of Paul's argument is that the economy of the Great Depression was in a liquidity trap, which implies that the AD curve is vertical, which in turn implies that policies that adversely shift the AS curve do not affect equilibrium output in the short run.
I am always heartened to see economists change the slope of their various curves to fit their arguments. I think we need more of this type of analysis to present to the public so that they can understand economics and economists better.
As a general matter, the state of aggregate demand depends on an amorphous variable called confidence. Anything that threatens to screw up AS in the long run most likely reduces confidence and AD in the short run. The textbook separation of AD and AS is useful for focusing discussion in the undergraduate classroom, but events in the real world are rarely so clean.