I am starting to get my head around the state of modern macroeconomics. In this connection there is a good post today on FT Alphaville.

My impressions of the field are these:

1. DSGE modelling has taken over academic macroeconomics.
2. People who are paid to make forecasts use empirical macro models.
3. DSGE would be a reasonable approach if it could capture the full complexity of the economy, if it treated events that are not random external shocks as endogenous, and if its inputs are stable. At present it fails on the first and second counts; I do not know about the third.
4. Given 3, it seems odd to use DSGE to make policy prescriptions.
5. Empirical macro models are the only available substitute.