Conditioning on value of first year for each country in unbalanced panel data

[Alexandre Kaisin]

Hi all!

I have country/year panel data. I would like to drop some observations according to specific values of some variables but I would like to do so according to the value for the first year in which I have data for the country.
The problem I am facing is that I have an unbalanced panel so I cannot use a command of the sort:

gen temp=1 if value<x & year==2000

How can I make sure I base my condition on the value for the first year for each country in my dataset.

My dataset is of the following form:

Country	Continent	Year	agprod	drought
ABW	LAC	1972	13413513	0
ABW	LAC	1973	15213523	1
ABW	LAC	1974	13513513	1
ADO	WEOFF	1972	353413	0
ADO	WEOFF	1973	234233	0

Many thanks!

[Nick Cox]

The first year for each country is given by subscript [1] after sorting.

bysort country (year) : gen drought1 = drought[1]

See https://www.stata-journal.com/articl...article=dm0055 for many more tricks in this territoryl

[Carlo Lazzaro]

Alexandre:
provided that I'm not completely clear with what you're after, let's assume, in the following toy-example, that you're interested in -age- at the first year this value is available for the different -panelid- the dataset is composed of:

. use "https://www.stata-press.com/data/r16/nlswork.dta"
(National Longitudinal Survey.  Young Women 14-26 years of age in 1968)

. bysort idcode (year): gen wanted=age if _n==1
(23,825 missing values generated)

. list idcode year age wanted if id==1

       +------------------------------+
       | idcode   year   age   wanted |
       |------------------------------|
    1. |      1     70    18       18 |
    2. |      1     71    19        . |
    3. |      1     72    20        . |
    4. |      1     73    21        . |
    5. |      1     75    23        . |
       |------------------------------|
    6. |      1     77    25        . |
    7. |      1     78    26        . |
    8. |      1     80    28        . |
    9. |      1     83    31        . |
   10. |      1     85    33        . |
       |------------------------------|
   11. |      1     87    35        . |
   12. |      1     88    37        . |
       +------------------------------+

.

[Alexandre Kaisin]

Thank you both so much for your swift answers!

As I said, I have panel observations for several years for each country. I would like to select (and eventually drop) countries for which, for example, GDP is under a certain threshold (GDP < x) for the first year in which data is available for that country (so first occurrence in the dataset).

I think I can come up with something that works using your inputs.

[Nick Cox]

 
 bysort country (year) : drop if GNP[1] < 42

[Carlo Lazzaro]

Aleksandre:
it's a good (and rewarding) habit to save a copy of your original dataset before -drop- anything.
As a famous song goes

The future's not ours to see

and you may need your original dataset for further analyses.