Panel Data Dummy Variable Query: Binary Outcome Related

[Shahnawaz Mohammad]

I shall make it simple, suppose, my data is balanced, my data structure is as follows,

ID || Year || Country || X ||


Now, say I create a dummy variable in this database (without recoding any existing variables) and want to run a Panel data model related to Binary outcome, say, logistic,

xtlogit X, fe/Re

my query : How reliable is this model estimation in econometrics basics-wise? Just curious, saw some earlier posts on dummies while browsing.

Thanks!

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

I'm not sure what || means or what the bold || means. If you have individual or firms within countries by year and individuals or firms don't change country, then if you xtset your data by firm(or person)-year, I think you can legitimately do xtlogit. The individual/firm effect will incorporate non-changing features of individuals/firms which includes country. If you use a continuous dv, xtreg is probably most appropriate. It is seldom advisable to make a continuous variable into a dummy unless there are extremely strong theoretical reasons to do so. You throw away information by making a dummy.