Panel data set, including several fixed effects

[Natalie Peltzer]

Hello,

I have a panel data set with the following main regression

Panel variable(1-357): Bank_Number; Time Variable (1-3): ST_Projection


regress y x1 x2 x3 x4


How do I include year fixed effects (variable: ST_Year) and country fixed effects (Variable: Country)?


Thanks in advance for the answer.

[Andrew Musau]

How do I include year fixed effects (variable: ST_Year) and country fixed effects (Variable: Country)?

Why not bank fixed effects? Can you guarantee that there is no bank-level heterogeneity? To include country and year effects, you need both the country identifier and time variable to be numeric. If the former is a string variable, see

help encode

Then, the regression is simply

xtset Country
xtreg y x1 x2 x3 x4 i.ST_Year, fe cluster(Bank_Number)

or

regress y x1 x2 x3 x4 i.Country i.ST_Year, cluster(Bank_Number)

[Natalie Peltzer]

Why not bank fixed effects? Can you guarantee that there is no bank-level heterogeneity? To include country and year effects, you need both the country identifier and time variable to be numeric. If the former is a string variable, see

help encode

Then, the regression is simply

xtset Country
xtreg y x1 x2 x3 x4 i.ST_Year, fe cluster(Bank_Number)

or

regress y x1 x2 x3 x4 i.Country i.ST_Year, cluster(Bank_Number)

Thanks Andrew!

The exercise from my professor said not to include bank fixed-effects.

If I only want to include year fixed effects the Code would be?

regress y x1 x2 x3 x4 i.ST_Year, cluster(Bank_Number)

[Andrew Musau]

Is this homework? If so, see #4 https://www.statalist.org/forums/help#adviceextras

[Natalie Peltzer]

No, its not homework that will be graded, just some extra exercises for us to practice

[Andrew Musau]

If I only want to include year fixed effects the Code would be?

regress y x1 x2 x3 x4 i.ST_Year, cluster(Bank_Number)

Correct. In linear fixed effects models, you can include fixed effects using indicators for each level of the FE variable (omitting one indicator).