Logit regression with country, industry and year fixed effect

[Abdramane Cherif]

Hello, I have a problem for you to consider. I am conducting a study on access to credit. My sample consists of 13400 firms from several countries (27 in total). I am facing a design problem. Indeed, although the survey questionnaire is identical in all countries, the year of administration is not the same. For example, in Chad the survey was carried out in 2018 whereas in CAMEROON it was in 2016. My question is: given that my study period is 2010 to 2018 and that in each country the survey was conducted once, is this a panel? Or is it cross-sectional data? Since in each country the survey was carried out only once. So no repetition. If so, is it possible to estimate a simple logit (my dependent variable is binary) with fixed effect (industry, country and year)? can you know a synthax or stata command to estimate this model?
Thank you

[Joro Kolev]

I do not think that wondering whether your data is a panel or not leads to anything productive. Panel data is just a name, sometimes a useful name, sometimes not so useful name.

You can easily control for country and year fixed effects by factor variable notation, i.country and i.year.

Controlling for industry depends on how many industries you have and how many firms per industry.

You can either do i.industry if these are not too many, or you can set up your data as a panel

xtset industry

and then use -xtlogit-.

[Abdramane Cherif]

Thank you Joro Kolev