optimal lag lengths in newey

[Paris Rira]

Dear Profs and Colleagues,

I have panel data.

xtset NPC_FIC year

Panel variable: NPC_FIC (unbalanced)
 Time variable: year, 2010 to 2020, but with gaps
         Delta: 1 year

after regression, I realized that there exist heteroskedasticity and autocorrelation

reg mig_jump immi_sh firm_age foreign_aff i.sector i.region

in order to correct them simultaneously, I use neway

newey mig_jump immi_sh firm_age foreign_aff i.sector i.region, lag(4) force

The question arises that whats the optimal lag lengths? How can I be assured that lag(4) or lag (3) or ... is the proper one?

Another question is that can I use newey with random effect ? because Hausman Test says to run random effect.

Any help appreciated.

Cheers,
Paris

[Andrew Musau]

See #4 https://www.statalist.org/forums/for...lation-command. For newey standard errors with panel data, see

ssc install newey2
help newey2

The linked selection criteria assume that you have time-series data. Why use newey standard errors with panel data? If you have \(N> T\), clustering on the panel identifier will yield standard errors that are robust to heteroscedasticity and arbitrary forms of serial correlation. So you would just go xtreg, fe with the -robust- option.

[Paris Rira]

Thank you Andrew. The first question has been solved. About the second one, still I am confused, how can I deal with random effect .