preduct residuals for all my panel units

[Olivia Johns]

Hi i have a panel data that contains 30 units. I would like to run the following for all my panel units, and have the result saved as a variable.

regress y x1 x2 dv dv2 (for all my panelid)
predict residual, res (for all my panelid)
dfuller residual (for all my panelid)

I am guessing i need to create a loop?

[Carlo Lazzaro]

Olivia
the following toy-example stops one step behind -dfuller- (something I'm not familiar with):

use "http://www.stata-press.com/data/r14/nlswork.dta", clear
forval i = 1/3 {
reg ln_wage c.age i.grade if idcode==`i'
predict res`i' if idcode==`i', res
}

[Olivia Johns]

what if my panelid isn't ordered incrementally? it's just random country codes

[Nick Cox]

http://www.stata.com/support/faqs/da...ach/index.html Method 1

[Jesse Wursten]

Why exactly do you want to do this? If you want to get panel-id specific coefficients, I recommend you use -xtmg- (available on SSC), which produces the panel specific betas for you, as well as the average. If you want to test for the presence of a unit root, you should use a panel unit root test, such as Pesaran's CADF (-pescadf-, -xtcips-) or any of the options in -xtunitroot-. They have considerably more power than any combination of panel-unit-specific tests.