csdid with one cross-section

[Mikael Pasternak]

Hello!
I only have data with one cross section (not repeated) at the individual level, but the treatment I investigate has a staggered adoption by district and cohort. I feel that I can use the csdid or did_multiplegt commands to assess the effect of my treatment, by comparing treated individuals to individuals in the same district in never treated cohorts, or individuals from the same cohorts in never treated districts. How should I implement that in the data? My outcome is constant over time (education).

The only way I see with stata commands is to create a "fake" panel by adding observations for each year. Is there another (better) way?

[Jared Greathouse]

I don't understand. Can you explain what you mean by your outcome is constant over time?

[Mikael Pasternak]

My outcome is education. Once individuals receive education, it does not evolve over life. I can thus observe it only once for each individual. I do not have a pre-treatment value in a sense.

[Jeff Wooldridge]

If there's no before period, I don't see that there's much that can be done. If the intervention is assumed to have affected education level, how can that happen if your measure of education is the same in all time periods for a given individual? By definition, the intervention will have no effect. If you did create a "fake" panel and you used two-way fixed effect, the unit fixed effect will entirely explain education.

[Mikael Pasternak]

Unit fixed effects yes but what about district and year of birth fixed effects?