Set dataset as a panel

[ruth matko]

Hello,

I have the problem to set my dataset as a panel because I have repeated time values within my panel.

My dataset looks something like the table below, the problem is that I have each observation seperatly and not all observations for each year and quarter like 2007/Q1, 2007/Q2 and so on

That's why I get the error "repeated time values within my panel". Is it possible to merge all 2007 Q1 observations together (to sum it up) or is there another way to use may data as a panel?

 
age
countrydummy
year
quartal
country
educ
nat

42
1
2007
Q1
AT
2
nat

5
1
2007
Q1
AT
1
EU

66
1
2007
Q1
AT
3
nat

33
1
2007
Q1
AT
1
EU

2
1
2007
Q2
AT
2
nat

Thank you,
best regards
Ruth

[Carlo Lazzaro]

Ruth:
if you do not plan to use time-series commands, you can simply -xtset- your dataset with -panelid- only:

xtset panelid

[ruth matko]

Thank you Carlo,

but actually I want to run a gmm regression with the data, does it work withou setting my data as a panel?

Best regards,
Ruth

[Carlo Lazzaro]

Ruth:
sorry, I'm not familiar with -gmm-.
My previous reply referred to -xtset- in general.
Anyway, you may try and see what happens.

[Phil Bromiley]

gmm is not a specific estimator - it is a general method that has been implemented in many different estimators for different situations. For example, xtivreg2 offers a gmm estimator - it would require a panel setup. But ivreg2 also offers a gmm estimator but does not use a panel set up. xtabond and related estimators (given xt prefix) require panel set ups as far as I know. But I suspect they require setting both panel and date since they rely on lags as instruments.

It is possible to use collapse to change to one observation per country-year if that makes sense in your context.

[Amin Sofla]

In order to set your data as a panel, you need a panel variable (and a time variable). You should first decide what your panel variable is. Next, depending on your context, you can also use the combination of the year and the quarter as the time variable for your panel (if time variable is necessary).

[ruth matko]

Thank you.
My panel variable is country or countrydummy and my time variable should be year, but the problem ist that I have multiple observations on the same year within country. I don't know how to solve this probleme. I want to use xtabond for my gmm regression. I will try now the command collapse.

[Karel Novak]

I had a similar problem to yours. The command I used was same for as above, but instead of set date yearly, I set date daily:

xtset country_variable date_variable, daily

[ruth matko]

Thank you but I tried this code before and then I got the error "repeated time values within panel "

[Amin Sofla]

Then, you need to investigate the duplicates.

Code: 
duplicates tag [varlist] [if] [in] , generate(newvar)