Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: specifying j in Xijt


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: specifying j in Xijt
Date   Fri, 21 Apr 2006 00:17:26 +0100

I don't claim divine inspiration. I just found
the question unclear. 

This may be called panel data or set up formally 
as panel data, but that seems irrelevant to the 
model posed, which despite various complications
seems firmly cross-sectional. Thus the basic
structure would seem to be that an observation 
is an import from one country into another at 
a time. 

However, that may be wrong, as I don't see an 
explanation of a_it, a_jt, a_ij. 

On the terminology "gravity model": that term 
was once popular in some fields for models attempting
to predict interactions (here trade) from characteristics
of the places concerned, but there was typically
a distance (travel time, travel cost) variable as
well by loose analogy with an inverse square distance
term in Newtonian mechanics, hence the name. I see 
no distance term here. 

Nick 
n.j.cox@durham.ac.uk 

Clive Nicholas
 
> Kremena Platikanova replied:
> 
> > Thank you for your response.
> >
> > Here is a little more information about my panel data. The 
> indeces in Xijt
> > stand for i=importer country, j=exporter country, t=year.
> >
> > My goal is to estimate (for multiple industries) the 
> gravity equation:
> >
> > Y_ijt=a+a_it+a_jt+a_ij+b1*X1_it+b2*X2_jt+b3*R_it+e_ijt
> >
> > where Y=Imports, X_1it=importer characteristics, X2_jt=exporter
> > characteristics, R_it=reform dummy (=1 for the year of and the years
> > following
> > the strengthening of intellectual property rights in the importer
> > country).
> >
> > But again my question is how to specify the variable 
> corresponding to the
> > index j?
> 
> All you've done is restate the question, so I don't have that 
> much more to
> say to this.
> 
> One comment I will make is on your nesting of 'importer' nations into
> 'exporter' ones. Now, pardon me for being a part-time 
> political scientist,
> but that looks a bit suss to me. Conceptually, to me at any rate, they
> exist on the same level. If, and only if, I'm correct on that, my
> suggestion is the same: i.e., run -tsset- twice - once for 
> the 'i-nations'
> and once again for the 'e-nations' - and see whether any 
> differences occur
> in fitting the model you state above.
> 
> I rather think that there are more important estimation 
> issues to consider
> here, anyway (e.g., which panel model to use, fixed effects or
> first-differences, are lags to be used, etc.), which will 
> have probably
> have greater impact on the findings of your model than how 
> you should use
> -tsset-.
> 
> The fact that nobody else has yet posted a response to your 
> query would
> appear to suggest that this really is the only solution 
> anybody can think
> of, barring divine inspiration.
 

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index