time-invariant variables in fixed effect - industry classification

[sultan al shekaili]

hello

i am working on analysing the significance of some firm specific factors on the leverage level. the firms are from 18 industries. i am using panel data fixed effect regressions. i want to run the fe regression on the full sample and the subsamples of industries to see if the same factors explain the leverage levels similarly in all industries or not.

i generated dummy variables for each industry but Stat omit them and does not provide estimations. probably because these dummies are time invariant.

is there anyway to achieve my goal without having to subsample and regress 18 times.


thanks
sultan

[Carlo Lazzaro]

Sultan:
if you're interested in estimating coefficients for time-invariant predictors, -fe- machinery will not support that task.

[sultan al shekaili]

thank carlo,

can i use kind of " if" option to instruct Stat to only regress firms belonging to the same industry:

something like:

" xtreg y xi x2 , fe if ind=energy "

regards
sultan

[sultan al shekaili]

thank carlo,

can i use kind of " if" option to instruct Stat to only regress firms belonging to the same industry:

something like:

" xtreg y xi x2 , fe if ind=energy "

regards
sultan

forgot to put a question mark here.

if any can help, please do

thanks

[Carlo Lazzaro]

Sultan:
yes, you can.
But the -if- conditional statement should be moved before the comma.