Repeated time values in sample r(451)

[Trang Dong]

Dear Statalist!

I have panel data. I want to select number of lags.
I type:

tsset quocgia year
-> tsset quocgia year
panel variable: quocgia (strongly balanced)
time variable: year, 1998 to 2013
delta: 1 year

varsoc lngdp lnyt lngd
-> repeated time values in sample
r(451);

What should I do next?
Thank you!

[Nick Cox]

Many posts here on varsoc and panel data: e.g. http://www.statalist.org/forums/foru...ith-panel-data

[Trang Dong]

Tks Nick! But I can't find any suitable answer for my case.
Sorry for my bad English. :'(

[Nick Cox]

Not my field, but the accepted answer appears to be that you need to loop over panels to apply varsoc to each individually. But your problem probably needs a different approach. Sorry, but I can't advise.

[Eric de Souza]

-varsoc- is meant for time series data, not for panel data

[Nick Cox]

Eric: sure, but that dataset is not panel data. (Or, if you like, it's a single panel.)

The problem is with panel data: does applying varsoc then make sense? how best to do it? As #2 implies, this is frequently asked here.

[Eric de Souza]

Nick: #1 indicates that there are several panels. The error message he gets is what you get when you apply -varsoc- to panel data:

webuse grunfeld
varsoc invest mvalue kstock

The dataset "grunfeld.dta" contains 10 panels (companies), and, therefore, the same year is repeated ten times.
Trang Dong doesn't have a long enough time series for each panel to apply a -var- (Vector AutoRegression) to each panel separately.

[Nick Cox]

Eric: Quite so. I understood the first point exactly.

Your point about series length does seem crucial.

[Shubham Sharma]

How to decide the lag length for panel data? kindly help