Xtset for Same "Stage" but Different Time

[Alexis CHEN]

Dear all,

I have a company dataset. Each company has two-stage (e.g., stage = 0 or stage =1). However, the times of the two stages are different across companies.
Are my data still considered to be panel data if I want to set "xtset id stage"? Can I still use panel models and control year-month dummies in my regression?

Many thanks,
Alexis

[Carlo Lazzaro]

Alexis:
welcome to this forum.
The panel recipe prescribes to have the same sample (more or less, as attrition may play a role as time elapses) measured on the same set of variables at equally spaced time interval (say; patient baseline visit at t0; then four follow-up visits, one per each quarter if the study lasts one year) .
While -2/+2 months departures from the scheduled measurement date are often allowed in longitudinal stata protocols, most depend what you mean by different in your case.
More substantively, what do the two stages you mention imply? Do they measure the very same set of variables or not?

[Alexis CHEN]

Hi Carlo,

Thanks a lot for your reply! The stage refers to the firm's funding stage (e.g., seed, series A/B/C, etc). However, the time might vary over 10 years.

I am struggling with choosing the right model to analyze this data. Do you think the longitudinal model is preferred in my case? And do you think it is better to include stage in xtset, or simply control the stage in my model?

Many thanks,
Alexis

[Carlo Lazzaro]

Alexis:
provided that my last experience with finance dates back to >30 years ago, for your description I would go pooled OLS, including -stage- and -time- in the right-hand side of your regression equation and imposing -vce(cluster firm)- standard errors.
That said, take a look at the literature in your research field for better and updated advice.

[Alexis CHEN]

Hi Carlo,

Appreciate your time and help!
I will try pooled OLS and searching for more references under finance context.

BR,
Alexis