Lagged variable and panel structure

[Roman Neuenschwander]

Dear Statalists

At the moment I am dealing with a apparently simple problem. The following data serves as an example: I would like to create a dummy variable that takes the value 1 if the observation of the same company (gvkey) from the previous year fulfils a certain condition. My question: How is it possible that the first observation has the value 1? Why is the panel structure not taken into account? I am aware that I can simply delete the first observation. However for a better understanding, I would like to know why Stata sets the value 1 for the first observations with this command. Interestingly, the panel structure is taken into account when I generate a lag variable (first observation is always missing which is correct).

Many thanks for your help.
Roman

xtset gvkey year
       panel variable:  gvkey (strongly balanced)
        time variable:  year, 2008 to 2010
                delta:  1 unit

. gen lag = L.intcov
(6 missing values generated)

. gen dummy = 1 if L.intcov > 1
(8 missing values generated)

. list

     +-------------------------------------+
     | gvkey   year   intcov   lag   dummy |
     |-------------------------------------|
  1. |     1   2008       -1     .       1 |
  2. |     1   2009        1    -1       . |
  3. |     1   2010        2     1       . |
  4. |     2   2008       -2     .       1 |
  5. |     2   2009        3    -2       . |
     |-------------------------------------|
  6. |     2   2010        4     3       1 |
  7. |     3   2008        5     .       1 |
  8. |     3   2009        4     5       1 |
  9. |     3   2010        1     4       1 |
 10. |     4   2008       -1     .       1 |
     |-------------------------------------|
 11. |     4   2009        1    -1       . |
 12. |     4   2010        1     1       . |
 13. |     5   2008       -2     .       1 |
 14. |     5   2009       -2    -2       . |
 15. |     5   2010        2    -2       . |
     |-------------------------------------|
 16. |     6   2008        5     .       1 |
 17. |     6   2009       -5     5       1 |
 18. |     6   2010       -1    -5       . |
     +-------------------------------------+

[Andrew Musau]

Missing is evaluated greater than one.

assert .< 1
assert .> 1

Result.:

. assert .<1
assertion is false
r(9);

. assert .>1

.

[Roman Neuenschwander]

Thank you Andrew, it works perfectly!