Skip to content

GMM

4 messages · Pankaj K Agarwal, Eric Zivot, Mark Leeds

Original
Mark Leeds
# 
Hi : Yes but GMM is used more for either A) when the OLS assumptions are
not true ( i.e autocorrelation or heteroscedasticity. ) or B) you have a
function that is not
necessarily linear like it is in the case of OLS.

Also, Achim could definitely say more on this but there are various
techniques for the construction of HAC estimators so using sandwich may not
necessarily give the same results as GMM even in the OLS case.











On Mon, Jun 27, 2016 at 2:17 PM, Pankaj K Agarwal via R-SIG-Finance <
r-sig-finance at r-project.org> wrote:

            

      
      
      
    

        

      
      
    

        

  
  


  


      

    

    

      
      

        


  

        

      Eric Zivot
    

    

      
      #
    

  


  

      
No
OLS is a special case of GMM where the number of moment conditions is the same as the number of parameters. In this case the efficient weight matrix does not matter for estimation but does matter for the calculation of an estimate of the asymptotic variance matrix of the OLS parameters. This is what HAC standard errors do in the sandwich function vcovHAC()

-----Original Message-----
From: R-SIG-Finance [mailto:r-sig-finance-bounces at r-project.org] On Behalf Of Pankaj K Agarwal via R-SIG-Finance
Sent: Monday, June 27, 2016 11:17 AM
To: R-sig-finance <r-sig-finance at r-project.org>
Cc: H.K Pradhan <pradhan at xlri.ac.in>
Subject: [R-SIG-Finance] GMM

Apologies if this question is irrelevant for this group.
Does using HAC standard errors (Newey and West: package sandwich) in OLS regression make using GMM (package: GMM) redundant?
 Regards,Pankaj K Agarwal
+91-98397-11444http://in.linkedin.com/in/pankajkagarwal/

_______________________________________________
R-SIG-Finance at r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-sig-finance
-- Subscriber-posting only. If you want to post, subscribe first.
-- Also note that this is not the r-help list where general R questions should go.

  
  


  


      

    

    

      
      

        


  

        

      Mark Leeds
    

    

      
      #
    

  


  

      
Thanks Eric for your explanation. But I'm a little confused. I understand
that in
the case of OLS, the system is perfectly identified. But are you saying that

A) if he used lm and sandwich, then he won't get a correct answer

B) that if he uses GMM, he won't get a correct answer  because it's not
designed
for perfectly identfied systems. ( i.e: there is no weighting matrix ).

C) if he uses both, he'll get different answers.

Thanks.

And to the person who asked the question originally, below is a short but
sweet intro to GMM.

http://lipas.uwasa.fi/~sjp/Teaching/gmm/lectures/gmmc3.pdf

        
On Tue, Jun 28, 2016 at 12:55 PM, Eric Zivot <ezivot at uw.edu> wrote: