Bookmark and Share

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

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

st: Re: bilateral panel data

From   "Joseph Coveney" <[email protected]>
To   <[email protected]>
Subject   st: Re: bilateral panel data
Date   Mon, 7 Apr 2014 09:43:55 +0900

I'm not sure what your problem is, but it would probably be better for you to show your actual code and output.  I recommend what Marcello Pagano just gently admonished, namely, to switch over to the new forum ( where you can more readily highlight your code and output, or even attach them to your post as text files, along with an excerpt of the Stata dataset.

Also, rather than grouping countries as if each participated in international trade in the context of only one pair, wouldn't it resemble reality more if you set the two trading partners up in a cross-classified random effects model?  As you have it now (if I understand it correctly), when you have group 1 = Canada & U.S.A. and group 2 = Mexico & U.S.A., you deny your model the information that groups 1 and 2 share a common trading partner.

Joseph Coveney

-----Original Message-----
From: [email protected] [mailto:[email protected]] On Behalf Of Andreas Dimopoulos
Sent: Monday, April 7, 2014 07:20
To: [email protected]
Subject: st: bilateral panel data

Hi there,

I am dealing with bilateral data for a specific number of years.
Particularly, I have pairs of countries and the trade between them as well
as some country-specific characteristics such as the GDP of each country.
Some of the variables remain constant (e.g. distance between countries)
while others change. My data also consists of the reporter and the partner.
Therefore, as I have read from previous posts I generated a group by
variable using the command: "egen group=group(reporter partner), label".
Then, I did "xtset group year" and tried to run the xtreg between the
dependant variable (lntrade) and the relevant GDP values (lnGDPi, lnGDPj).
However, stata gives me the outcome 0 (omitted variable). I then checked
for the correlation between the explanatory variables and found that there
is no perfect collinearity. Hence, I do not know why stata keeps on giving
this omitted variable outcome.

Any ideas will be more than useful.

*   For searches and help try:

*   For searches and help try:

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