st: re: assistance with time series data

[Kit Baum <baum@bc.edu>]

< >
Justin said

I am working with person-year data and trying to estimate the time it  
takes someone to move after they have finished their education. The  
data tell me for each year of a person's life, whether they moved  
that year and whether (and what level) of school they were in. I have  
a variable that tells me the age someone made a move (could have  
multiple moves per lifetime) and the age that someone finished school  
(could only finish school once in lifetime). To get what I want, the  
interval between finishing school and move, I subtract the persons  
age when they moved from the time they finished school. The works out  
fine unless someone made a move before finishing school, which is  
where I'm stuck. For instance, someone moved at age 15 and at age 22  
and they finished school at age 21. I'm only interested in the move  
made after finishing school but with the coding, it's only accepting  
the first move. So the time between finish school and moving is -6  
rather than +1. I've tried several  things but nothing is working.


. use schoolmove, clear

. list, sepby(person)

     +---------------------------------+
     | person   age   inschool   moved |
     |---------------------------------|
  1. |      1    15          1       0 |
  2. |      1    16          1       0 |
  3. |      1    17          1       0 |
  4. |      1    18          0       0 |
  5. |      1    19          0       0 |
  6. |      1    20          0       1 |
  7. |      1    21          0       0 |
     |---------------------------------|
  8. |      2    18          1       0 |
  9. |      2    19          1       1 |
 10. |      2    20          1       0 |
 11. |      2    21          1       0 |
 12. |      2    22          0       0 |
 13. |      2    23          0       1 |
 14. |      2    24          0       0 |
     +---------------------------------+

. g movedafter = !inschool & moved

. egen leftschool = max(age*inschool), by(person)

. g sinceschool = age*movedafter - leftschool if movedafter
(12 missing values generated)

. list, sepby(person)

      
+------------------------------------------------------------------+
     | person   age   inschool   moved   moveda~r   leftsc~l    
sinces~l |
      
|------------------------------------------------------------------|
  1. |      1    15          1       0          0          
17          . |
  2. |      1    16          1       0          0          
17          . |
  3. |      1    17          1       0          0          
17          . |
  4. |      1    18          0       0          0          
17          . |
  5. |      1    19          0       0          0          
17          . |
  6. |      1    20          0       1          1          
17          3 |
  7. |      1    21          0       0          0          
17          . |
      
|------------------------------------------------------------------|
  8. |      2    18          1       0          0          
21          . |
  9. |      2    19          1       1          0          
21          . |
 10. |      2    20          1       0          0          
21          . |
 11. |      2    21          1       0          0          
21          . |
 12. |      2    22          0       0          0          
21          . |
 13. |      2    23          0       1          1          
21          2 |
 14. |      2    24          0       0          0          
21          . |
      
+------------------------------------------------------------------+


Notice that Ms. 2 moved during school, but that move is not considered.


Kit Baum, Boston College Economics and DIW Berlin
http://ideas.repec.org/e/pba1.html
An Introduction to Modern Econometrics Using Stata:
http://www.stata-press.com/books/imeus.html


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/