how to drop an observation if it does not appear in both years

[Yun Cheng]

Hello!

I have some data with firm and year and some other variables, I want to make sure each firm has data in both 2019 and 2020, if there is only 1 year of data and the other year is missing, I need to delete it. How can I do that? Thanks!

For illustration, here are some data

* Example generated by -dataex-. To install: ssc install dataex
clear
input double firm float year
 1 2019
 1 2020
 2 2019
 2 2020
 4 2020
 5 2019
 5 2020
 6 2019
 6 2020
 7 2020
 8 2019
 8 2020
 9 2020
10 2019
11 2019
11 2020
12 2019
12 2020
14 2020
16 2019
17 2019
17 2020
19 2019
19 2020
20 2019
20 2020
format %ty year

[Carlo Lazzaro]

Yun:
you may want to consider:

. bysort id (year): drop if _N<2


. list

     +-----------+
     | id   year |
     |-----------|
  1. |  1   2019 |
  2. |  1   2020 |
  3. |  2   2019 |
  4. |  2   2020 |
  5. |  5   2019 |
     |-----------|
  6. |  5   2020 |
  7. |  6   2019 |
  8. |  6   2020 |
  9. |  8   2019 |
 10. |  8   2020 |
     |-----------|
 11. | 11   2019 |
 12. | 11   2020 |
 13. | 12   2019 |
 14. | 12   2020 |
 15. | 17   2019 |
     |-----------|
 16. | 17   2020 |
 17. | 19   2019 |
 18. | 19   2020 |
 19. | 20   2019 |
 20. | 20   2020 |
     +-----------+

.

A cautionary tale is called for, though: this way you're making up your dataset, without delving into the reason(s) why some panels have one wave of data only.