Hello!
I'm using panel data including the following variables,
(y variable: abnormal return, cumulative abnormal return)
(x variable:GovScore log_BM log_Size CAR AR EnvirScore ESGScore Size Lev SocialScore)
These variable are all focused on a short event window for like a month, hence, especially for the score variable, it maintains the same, hence, that means it is time-invariant?
My goal was to carry out regressions and do the Hausman test to see if I should use fixed effect or random effect. And then hopefully to move on to find the relation between return and ESG scores.
But when I did the code:
This happened
I've read across in the thread of using a 'hybrid model' but few questions seems to encounter problems at the early stage of just doing the xtreg coding, hence, does it mean there is fundamentally something wrong with my dataset (my dataset is strongly balanced) and it should be better applied as an OLS model? But this would be very wired since I am carrying out an event study.
And when I did normal regressions, this was the result, does that mean the result is insignificant no matter if it was panel data or not?
This relates to a previous question posted on the server:
https://www.statalist.org/forums/for...edicted-return
Thank you very much,
Emma
I'm using panel data including the following variables,
(y variable: abnormal return, cumulative abnormal return)
(x variable:GovScore log_BM log_Size CAR AR EnvirScore ESGScore Size Lev SocialScore)
These variable are all focused on a short event window for like a month, hence, especially for the score variable, it maintains the same, hence, that means it is time-invariant?
My goal was to carry out regressions and do the Hausman test to see if I should use fixed effect or random effect. And then hopefully to move on to find the relation between return and ESG scores.
But when I did the code:
This happened
I've read across in the thread of using a 'hybrid model' but few questions seems to encounter problems at the early stage of just doing the xtreg coding, hence, does it mean there is fundamentally something wrong with my dataset (my dataset is strongly balanced) and it should be better applied as an OLS model? But this would be very wired since I am carrying out an event study.
And when I did normal regressions, this was the result, does that mean the result is insignificant no matter if it was panel data or not?
This relates to a previous question posted on the server:
https://www.statalist.org/forums/for...edicted-return
Thank you very much,
Emma
Comment