Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Asymetrical dimensions in the RHS and in the LHS


From   Fabien Bertho <[email protected]>
To   [email protected]
Subject   Re: st: Asymetrical dimensions in the RHS and in the LHS
Date   Tue, 29 Mar 2011 17:23:00 +0200 (CEST)

Thank you for your answer.

I think that I found the literature on this field [Moulton (1986, 1990) and Cameron, Gelbach, and Miller (2006)]

To deal with this issue I have to use clustered standard errors.

My model is the following:

Y_odk=X1_od+X2_od+X3_od+FE_k+FE_o+FE_d

With o, the origin country; d the destination country; k, the product and FE are fixed-effects.

My variable of interest varies by od. Hence, the solution would consist in clustering at the higher level of aggregation -- i.e. by od. 

As I have country fixed effects, is it necessary to use multiway clustering? Or is it possible to include cluster by origin only?

Thank you.

> ----------------------------------------
> From: Nick Cox <[email protected]>
> Sent: Mon Mar 28 10:58:57 CEST 2011
> To: <[email protected]>
> Subject: Re: st: Asymetrical dimensions in the RHS and in the LHS
> 
> 
> I think your problem is like many others. You can
> 
> regress LHS RHS
> 
> or whatever else you want
> 
> and Stata will be indifferent to whether some of your variables
> include blocks of constants except insofar as some variables are
> entirely constant. The bigger problem is quite what error structure
> would be better than that crude pooled estimation, as clustering by
> origin, by destination, by origin-destination pair, by product would
> all a priori (your words) seem to be plausible.
> 
> But isn't there literature that discusses how to model such data?
> 
> Nick
> 
> On Mon, Mar 28, 2011 at 9:18 AM, Fabien Bertho
> <[email protected]> wrote:
> > Thank you very much for your answer.
> >
> > Actually, I cannot perform cross-tabulations because my dataset is too large. However, I thing that there is variation in all dimensions.
> >
> > If I understood well, it is something which is not impossible to do a priori?
> >
> >> From: Charles Koss <[email protected]>
> 
> >> I think, you also need to take care of the interpretations from such
> >> estimates. Especially if some of your dimensions show no variation.
> >> Have you perform cross-tabulations among dependent and independent
> >> variables?
> >>
> >> On Fri, Mar 25, 2011 at 8:30 AM, Fabien Bertho
> >> <[email protected]> wrote:
> >> > I would like to run various types of estimations, on the one hand OLS and TSLS IV regressions; on the other hand tobit and ivtobit regressions.
> 
> >> >> From: Oliver Jones <[email protected]>
> 
> >> >> I'm no expert, but I'm sure it depends on how you estimate it...
> 
> >> >> Am 25.03.2011 11:29, schrieb Fabien Bertho:
> 
> >> >> > I would have a basic question. I am estimating a very basic equation.
> >> >> >
> >> >> > In this equation, the dependent variable varies by three dimensions (odk) -- i.e. origin country, destination country and product. On the right-hand side, some explicative variables vary by (od), others vary by (k) but none by (odk)
> >> >> >
> >> >> > Is it correct to estimate such an equation?
> >> >> >
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/


--------------------------------------------------------------------------
Tous les courriers électroniques émis depuis la messagerie
de Sciences Po doivent respecter des conditions d'usages.
Pour les consulter rendez-vous sur
http://www.ressources-numeriques.sciences-po.fr/confidentialite_courriel.htm

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index