How to generate lag variable by groups?

[Jason Liang]

To whom it was attached attention by:
I have a simple dataset like below. I want to generate lag.2 var of "ToverOs" so that I can calculate:
ToverOs_t-ToverOs_t-2

input long stkcd str12 Trddt double(ToverOs day)
1 "2008-01-02" 1.141407 1
1 "2008-01-03" 1.569018 2
1 "2008-01-04" 1.131045 3
2 "2008-01-02" 1.595663 1
2 "2008-01-03" 1.653669 2
2 "2008-01-04" 1.854244 3
4 "2008-01-02" 7.767716 1
4 "2008-01-03" 6.816699 2
4 "2008-01-04" 5.304759 3

the solution of mine is to generate a lag.2 var by stkcd, but tsset doesn't work. It said:

repeated time values in sample

any other solutions to this issue? Thanks!

[Carlo Lazzaro]

Jason:
you may want to consider:

. generate double new_Trddt = date( Trddt , "YMD")

. format new_Trddt %td

. tsset stkcd new_Trddt

Panel variable: stkcd (strongly balanced)
 Time variable: new_Trddt, 02jan2008 to 04jan2008
         Delta: 1 day

. bysort stkcd ( new_Trddt ): g wanted=L2.ToverOs

. list

     +------------------------------------------------------------+
     | stkcd        Trddt    ToverOs   day   new_Trddt     wanted |
     |------------------------------------------------------------|
  1. |     1   2008-01-02   1.141407     1   02jan2008          . |
  2. |     1   2008-01-03   1.569018     2   03jan2008          . |
  3. |     1   2008-01-04   1.131045     3   04jan2008   1.141407 |
  4. |     2   2008-01-02   1.595663     1   02jan2008          . |
  5. |     2   2008-01-03   1.653669     2   03jan2008          . |
     |------------------------------------------------------------|
  6. |     2   2008-01-04   1.854244     3   04jan2008   1.595663 |
  7. |     4   2008-01-02   7.767716     1   02jan2008          . |
  8. |     4   2008-01-03   6.816699     2   03jan2008          . |
  9. |     4   2008-01-04   5.304759     3   04jan2008   7.767716 |
     +------------------------------------------------------------+

.

[Jason Liang]

That's exactly what i wanted to do. Thank you Sir.