Dummy Variable as Dependent Variable in a Panel Regression

[Ha Bon]

Good Day Everyone!

I am doing a panel regression on Financial Distress Prediction and I have a problem in running panel models in Stata. I wanted to know which panel model should I use for my data by testing Fixed Effects, Random Effects, and Pooled OLS.

The first steps I did is to import the excel containing the data, [xtset COMPANY YEAR], [regress DummyVariable IndependentVariables], and vif (which resulted to 1.26).

Then the problem happens whenever I run xtreg DummyVariable IndependentVariables, fe or xtreg DummyVariable IndependentVariables, re. This command results to periods on the models and it omits all variables. I have to run the model specification test by testing which models (Fixed Effects, Random Effects, and Pooled OLS) is applicable for the data. What should I do from here?

[Carlo Lazzaro]

Ha Bon:
unless your aim is to perform a linear probability model. since your regressand is categorical, you should probably switch to -xtlogit-.
Be informed that, as far as the -fe- specification is concerned, -xtlogit- gives back conditional fixed effect to avoid incidental parameter bias (

http://www.econ.brown.edu/Faculty/Tony_Lancaster/papers/IncidentalParameters1948.pdf

).

[Ha Bon]

Thank you so much Mr. Carlo Lazzaro! This also answers my question (which I have not posted) whether I'll use logit command or xtlogit! Have a great week!