(Unbalanced) Panel data sample for an event study analysis

[Laura Hoeybye]

Hi everyone

I'm currently working on a project, where my analysis will consist of an event study.

I have a dataset where individuals are observed over af 18 years period, but every individual is not necessarily observed in each year.

Essentially, I want to construct a sample where each individual is observed five years prior to the event (eventtime = 0) and 10 years after. But in order to avoid that the sample gets too small, I want to keep individuals who are observed at least 8 times over the 15-years period, and at least 1 time before the event and 1 time after.

For now, I have an event-time variable, which have values from -17 to +17 where the value "eventtime=0" is the year the event of interest happens.

Is there anyone who has a suggestion for a code, that can create the sample with the above-mentioned criteria?

Thank you in advanced.

[Lance Erickson]

Hi Laura,

I’m sure that there are some on the list that will have some ideas for you but most who frequent here prefer to offer specific solutions. So, without you posting a snippet of your data (using dataex), you’re less likely to get the kind of help that will be helpful.

[Laura Hoeybye]

Hi Laura,

I’m sure that there are some on the list that will have some ideas for you but most who frequent here prefer to offer specific solutions. So, without you posting a snippet of your data (using dataex), you’re less likely to get the kind of help that will be helpful.

Hi Lance
Thank you so much for your reply - I'll try posting an example of my data

[Laura Hoeybye]

Hi everyone

I'm currently working on a project, where my analysis will consist of an event study.

I have a dataset where individuals are observed over af 18 years period, but every individual is not necessarily observed in each year.

Essentially, I want to construct a sample where each individual is observed five years prior to the event (eventtime = 0) and 10 years after. But in order to avoid that the sample gets too small, I want to keep individuals who are observed at least 8 times over the 15-years period, and at least 1 time before the event and 1 time after.

For now, I have an event-time variable, which have values from -17 to +17 where the value "eventtime=0" is the year the event of interest happens.

Is there anyone who has a suggestion for a code, that can create the sample with the above-mentioned criteria?

Thank you in advanced.

Here is a an example of my data.

Each individual has and id, and the outcome variable is savings - so I would like to see how savings is effected by the event, at "eventtime=0"

input long id float(savings eventtime)
6872 -7 -10
6872 -9 -9
6872 -7 -8
6872 1560 -7
6872 1560 -6
6872 -9 -5
6872 -7 -4
6872 -7 -3
6872 -7 -2
6872 -7 -1
6872 -7 0
5804 -7 -8
5804 -8 -7
5804 -8 -6
5804 -7 -5
5804 -8 -4
5804 0 -3
5804 1500 -2
5804 1500 -1
5804 1501.1548 0
5804 -8 2
5804 -8 4
5804 -7 7
0366 -8 -2
0366 0 -1
0366 0 0
0366 0 1
0366 0 2
0366 0 3
0366 0 4
0366 0 5
0366 0 7
0366 0 8
0366 0 9
0366 0 10
0366 -8 11
0404 -8 -4
0404 0 -3
0404 600.4619 -2
0404 0 0
0404 0 1
0404 0 2
0404 833.3333 3
0404 0 4
8243 908.7532 -8
8243 909.6998 -7
8243 948.7298 -6
8243 1200.9238 -5
8243 1280.9854 -4
8243 1351.0393 -3
8243 1383.3334 -2
8243 1501.1548 -1
8243 1000.7698 1
8243 1273.7327 2
8243 -7 3
8243 -7 4
8243 1701.3087 5
8243 -7 6
8243 -7 7
8243 -7 8
8243 -7 9
8278 260 -13
8278 0 -12
8278 0 -11
8278 866.6667 -10
8278 1036.4968 -9
8278 -7 -8
8278 1333.3334 -7
8278 1301.0007 -6
8278 -7 1
8278 -7 2
8257 -8 -3
8257 -8 -2
8257 -8 -1
8257 -8 0
8257 -8 1
8257 -8 2
8257 -8 3
8257 -8 4
8257 -8 5
8257 -8 6
8257 -8 7
8257 -8 8
8257 -8 9
8257 -8 10
8257 -8 11
8257 -8 12
8257 -8 13
8257 -8 14
1606 1400 -1
1606 0 0
1606 0 1
1606 0 2
1606 0 3
1606 0 4
1606 0 5
1606 0 6
1606 -8 7
1649 1872.4404 -1
1649 2058.5835 0