Regional Dummies, Fixed effects and first difference

[Harsha Konara]

I am new to STATA and considering a panel dataset with following variables.

dependent variable - Education expenditure as a percentage of GDP country-years
Independent variables - This include number of control variables + dummy to identify Pre and post GFC + Regional dummies (example, South Asia, Latin America, Africa, Europe)

Questions
01. Can I use Fixed Effects if I am using a regional dummy (collinearity problem)
02. When I am using first difference can I simply use xtreg d.(all variables)
03. When I am using regional dummies I need to use n-1 regions ?

Much appreciated if someone can help with STATA commands

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 (fixed spacing fonts help), and sample data using dataex.

You don't say what your panel will be. If you use country panels with fixed effects, then regional are unnecessary - they're taken up by the country effects.

While you can do it by first differencing variables or using areg, xtreg is built for your problem and so will make your like easier. Generally, you don't need to include first differences if you use xtreg. With the levels, the b is the change in predicted value (eg edu exp) for a one-unit change in the iv. But when you use changes for all your variables your parameter is the change in the change in edu exp for a one unit change in the change in the iv. Not so easily interpreted.

You should use factor variable notation (e.g., i.region) to do your dummies and interactions - this will automatically drop a region. It also makes post estimation analysis using margins much easier.