Man, Arnold Kling is so cranky! While it's sometimes on point, and often entertaining, it's also sometimes just plain wrong.
Case is point is his recent anti-VAR rant.
Here's a snippet but do read the whole thing:
"The VAR crowd cheerfully ignores all the details in macro data. The economist with a computer program that will churn out VARs is like a 25-year-old with a new immersion blender. He does not want to spend time cooking carefully-selected ingredients. He just wants to throw whatever is in the pantry into the blender to make a smoothie or soup. (Note that I am being unfair to people with immersion blenders. I am not being unfair to people who use VARs.)"
It did take macro folks quite a while to come to the following realization, but pretty much everyone is on the same page now and agrees that,
In order to do policy analysis with a VAR you have to solve the same identification problem that a old fashioned structural model has to solve. It's not a free lunch.
Further more, pretty much everyone agrees (except for a few special cases) that
achieving identification by making the system recursive is not a very smart strategy.
So "the VAR crowd" is working on improved identification strategies (long run restrictions, sign restrictions, identification via covariances) and has made a lot of progress. Look, identification issues are not unique to VARs. They plague virtually all non-experimental studies.
Another area that, contra Kling, the crowd has made huge progress on is structural change.
VARS with common factors whose loading are time variant. VARs with regime switching both in the means and in the conditional variances. VARs with time-varying coefficients.
One area where VARs have gone var beyond old fashioned structural models is modeling time varying conditional variances (and covariances) using either GARCH or stochastic volatility.
A lot of this work is being done using Bayesian computational tools, but it's very mainstream. Primiceri's 2005 RESTUD piece is a good place to start.