Dummy as dependent variable reghdfe

[Marina Skogen]

Dear all,

I am currently writing my thesis where I am looking at the relationship between mutual fund flows and advertising. One of the things I want to look at is whether mutual funds tend to advertise after good performance. Thus, I wanted to use a dummy equal to 1 if advertising was undertaken in the subsequent month, and zero otherwise. I am working with an unbalanced panel dataset.

So far in my thesis I have used reghdfe to do my analysis, using fund and time as FE. Until now no dependent variable have been a dummy variable, so I was therefore wondering if it is possible to use reghdfe when the dependent variable is a dummy? I have seen many places that xtprobit or xtlogit have been recommended, but I was still wondering whether it is possible to use reghdfe.
Also, if one do use reghdfe, what would the exact interpretation be in terms of the explanatory variables? So for example:

Adv_t = alpha + beta*return_t-1

Will it also be a probability as for xtlogit?

Many many thanks for your help.

Regards,

Marina

[Carlo Lazzaro]

Marina:
welcome to this forum.
I would say that the main question is that -xtlogit, fe- allows conditional fixed effects, which are different from the ones calculated in linear models (like Sergio Correia's wonderful -reghdfe-)..
-xtprobit- is actually out of debate, as it does not allow fixed effect at all.
I would recommend you to discuss these issues with your supervisor.