Quantile regression (QR) for fixed effect panel data : which command is more approriated ?

[sedki zn]

Dear all,

i'm working on may PhD thesis ( the impact of risk taking on firm performence) using a date of 781 observations ( T=11).

i would like to apply the QR but i do not know which command i have to use : qreg2 , ivqreg2 or xtqreg !

(for my modeste knowledge i think that i cannot use xtqreg because T is too short)

i would be gratful if you can halep me with it

kind regards
sedki

[sedki zn]

any help please !!!

[Jeff Wooldridge]

Sedki: In the second edition of my MIT Press book — or, you can find my lecture notes with Imbens online — I proposed as an approximation applying pooled quantile regression using the Mundlak device of including time averages along with the time-varying variable (and time constant variables and time dummies). Then cluster your standard errors. If you did pooled OLS you’d get the fixed effects estimator. That’s the main motivation as an approximation in the quantile case. I don’t think the quality of the approximation has been studied. But it’s very easy and can be used with any T >= 2.