Barkley Rosser wrote:
> ----------------- HES POSTING -----------------
> Pat Gunning,
>
> Sims's approach is fully atheoretical.  It is one way to go,
> but hardly the only one.  His approach is more influential
> than it was initially, but it is hardly proof that AD is a
> useless anachronism that should not be taught to
> students.  Should we spend our time in principles
> courses avoiding supply and demand and simply
> teaching vector-autoregressive, time-series modeling?

Sim's approach (the vector autoregression or VAR) in Macroeconomics and 
Reality was indeed atheoretical.  But by as early as 1982 and completely 
by 1986, Sims had moved over to the structural vector autoregression or 
SVAR, which restores much of the structure to the large macro models 
that he was criticising.  He did this because critics convinced him that 
without some level of identification no causal or policy interpretations 
could be made of the estimates.  The SVAR has become one of the most 
ubiquitous tools of empirical macroeconomics.

It is also worth saying that Sim's criticism was not of any theory 
(Keynesian, monetarist, AD, AS, or whatever) but of the econometric 
practices that were used to map from theoretical to empirical 
time-series models.  The SVAR can be used to investigate Keynesian AD or 
James Ahiakpor's classical quantity theory.  It is a neutral tool in 
that respect.  Naturally, there are methodological issues as to whether 
it is a good tool or not, but it does not itself take sides in any 
theoretical debate.

I have written both on the history and methodology of the VAR and SVAR 
in a variety of places, including in my /New Classical Macroeconomics:  
A Sceptical Inquiry /and in /Causality in Macroeconomics/, as well as in 
papers that are referenced on my website:  http://www.econ.duke.edu/~kdh9

Kevin Hoover