three dimensional panel analysis

[litosbrito]

I also am doing three dimensional panel analysis- country, industry,year.

i=country, 10
j=industry, 10
t=year, 12

How can a treat this data as panel data?

When i use tsset comand i have this mensage:

. tsset id period
repeated time values within panel
r(451);

end of do-file

r(451);

[Richard Williams]

Can you do

egen idcode = group(country industry)
xtset idcode period

The egen command will create a unique code for each observed combination of country and industry.

[litosbrito]

Thank you Richard Williams!

I will try it!!

[litosbrito]

Already tried and I got! thank you so much Richard Williams

[Richard Williams]

Situations like yours are fairly common, and it has always seemed odd to me that you can't specify more than one variable for the panel id. Depending on how your vars are coded, you might also be able to do something like

gen long idcode = country * 100 + indcode

but you would need to make sure it actually worked right. The last two digits would identify the industry and the other digits would id the country. Adapt the code if the industry codes are not two digits, e.g. if industry codes are three digits then multiply country by 1000.

[litosbrito]

Yes it will be useful if STATA allow that.
I will try this suggestion to see what i can get!!

[litosbrito]

I try and it works too!!

Many thanks

[Richard Williams]

I kind of like my latter suggestion because the value of the idcode is not dependent on what countries and firms actually happen to be in the data set. But again you have to make sure that you code it right and that it works the way you want.

[Ralph Klijn]

Hello,

I have the same issue and what able to solve it using

 
 egen idcode = group(country industry) xtset idcode period

However, now I want to calculate the lagged variable of a variable that is unique to country and industry. about 25% of the output in missing values even tough I have no missing values in my non-lagged data. Does anybody know a solution? Best regards

[Nick Cox]

Ralph: Please don't post in two or more threads simultaneously unless they are about completely different questions.

If you have further questions on this, follow your other thread https://www.statalist.org/forums/for...nt-and-company