CODEWITHSUNDEEP
panelstata
Yan Song
Yan Song

Reputation: 2405

Is there a way to generate lags in a panel without collapsing the data?

I have a dataset that looks like this

State     Year      Policy      other_variables
 a        2000        0               18
 a        2000        0               19
                  .
                  .
                  .
 a        2001        1               86
 a        2001        1               23

The poicy value is constant within each state and year. But it changes for different state and different year. The other_variables are different for each observation.

I want to generate lags of the policy value for each state. However, I cannot use xtset state year and then use the L operator. There are repeated values within each state year combination. I know that collapsing the dataset, generate lag variables and then merge back to the dataset would work. My question is is there an easy way to do this operation?

Upvotes: 1

Views: 726

Answers (1)

Nick Cox
Nick Cox

Reputation: 37208

This may help: 

clear 
input str1 State     Year      Policy    
 a        2000        0               
 a        2000        0               
 a        2001        1               
 a        2001        1               
 end 

 bysort State (Year) : gen diff = Policy - Policy[_n-1] if Year == Year[_n-1] + 1 
 by State Year: replace diff = diff[_n-1] if missing(diff) 

 list, sepby(State Year) 

     +------------------------------+
     | State   Year   Policy   diff |
     |------------------------------|
  1. |     a   2000        0      . |
  2. |     a   2000        0      . |
     |------------------------------|
  3. |     a   2001        1      1 |
  4. |     a   2001        1      1 |
     +------------------------------+

Upvotes: 1

Related Questions