fixed effect logit regr. error and panel data problem

[romayer]

Hi Everybody,

I've started working with eviews and have encountered some problems in my analysis.(working with eviews 7)

Q1 The first error has to do with a year and firm fixed effect logit regression I attempt to do. I have created year and firm dummies (based on variables year and cusip9) in eviews. When I run the logit regression with all dummies (year and firm leaving out 1 for firm and 1 for year) I get a "quasi complete separation error" on my firm dummies. I have read about this error in the manual as well as on the forum and the only response I have seen so far is removing the firm from the sample. However I have over 500 cross sections and so far have not been able to solve it. The logit regression can be run with year fixed effects only. Any recommendations? The file is in the attachement "eviews acq q1" the group is prepared under group01acq and 1 firm and year dummy are already left out.

Q2 My second problem concerns a panel data set. I have an unbalanced panel dataset on 606 firms (cross id "cusip9") and 1399 (period id "madate") exact dates. And in the time lays my problem. Im a ols regression in which i am regressing a range of independent variables on car_22, see the Q2 file, "group 1" and equation "eq1" in the workfile.
I want to control for year fixed effects as well as firm fixed effects.
I cannot structure my data set on cusip9 (firm identifier) and year, since there are several firms with multiple observations in one year, and therefore year is not a unique identifier. My supervisor however stresses me to include year fixed effects, although I have explained that it is not possible in my eviews set.
This is where I am lost.
- Controlling for time and cross section fixed effects (in the regression options) based on cusip9 and madate gives me a near singular matrix error.
- Only controlling for cross section fixed effects (in the regression options) and adding manually 8-1 dummies for the 8 years gives me a regression output, but i don't know if this is a valid method of achieving what I want to achieve.

Any help would be greatly appreciated. I hope it is clear what I want to achieve.

Sorry beforehand if I missed any answers for my questions, I've searched the forum thoroughly but have not found a solution yet.

thanks in advance

Best Rod