Jordi Gali (aqui) faz uma análise dos métodos:
Central banks’ projections–i.e. forecasts conditional on a given interest rate path–are often criticized on the grounds that their underlying policy assumptions are inconsistent with the existence of a unique equilibrium in many forward-looking models. Here I describe three alternative approaches to constructing projections that are not subject to the above criticism, using two different versions of New Keynesian model as reference frameworks. Most importantly, I show how the three approaches generate different projections for inflation and output, even though they imply an identical path for the interest rate. The latter result calls into question the meaning and usefulness of such projections.
Ryan Avent da Economist (aqui) tem um bom comentário. O Fed não deu bola aos indicadores de mercado - todos indicando a expectativa de queda do dispêndio agregado. A crise foi muito pior por esta razão!