Cointregration relationship in Panel Data

[Andres GonzaleE]

Hi Statalist users,

I am trying to look for a long-term price relationship between different mineral commodity prices, which is why I thought about making a Cointegration model using a single variable, LogPrice, of several mineral commodities. The data looks like this:

Mineral LogPrice Period ID
Copper 15 1 1
Copper 12 2 1
Copper 13 3 1
. . . .
. . . .
. . . .
Copper 16 99 1
Alumin 23 1 2
Alumin 25 2 2
Alumin 24 3 2
. . . .
. . . .
. . . .
Alumin 29 99 2

I conducted the Unit Root tests and all the variables are I(1). Problem is, I am simply not sure how to build a VECM with just one variable, LogPrice, using different commodities.
Furthermore, when I am attempting to determine the number of lags for a VECM, the varsoc function doesn't seem to work for Panel Data.

I know this is probably pretty basic but I've spent several hours on this and can't seem to get my head around it. Any help would be greatly appreciated.

Thanks,
Andrés

[Nicu Sprincean]

It seems that there is no panel VECM option in Stata, and neither in other softwares, as far as I know, although you cand find the user-written packages -pvar- and -pvar2- for Stata, but they do not incorporate a panel VECM too.
pvar: https://sites.google.com/a/hawaii.ed...love/home/pvar
pvar2: http://econweb.umd.edu/~decker/code.html
You may also want to take a look over these packages, which incorporate Error Correction Models: -xtpmg- and -xtdcce2-.