[Farhan Hasnat]

Andrew Musau Understood and thank you. Been a great help

[Mukesh Punia]

xtset id
xtivreg REM_PROXY_w POST_REG INBD ROA_w Size MTB_x_w LEV_w NOA_X_w (ABS_DA_w = C_PROD_w) i.Year, fe small

is equivalent to

ivregress 2sls REM_PROXY_w POST_REG INBD ROA_w Size MTB_x_w LEV_w NOA_X_w (ABS_DA_w = C_PROD_w) i.id i.Year

In such case, is it possible to to cluster standard error at "id" level in above example?


In a recent study using one period cross-section data which applied IV estimation approach, stated that: "District Fixed Effect" is applied and robust "Standard Error" is clustered at "District" level. Notably, Instrument was also "an average of a variable at district level"

As per my understanding this implies ,

ivreg2 y x₁ x₁ x_k i.District (endogenous_var=instrument_district_levelavg), robust cluster(District)

here i.District implied district fixed effect.
cluster(District) implies SE clustered at district level.

Is it possible to use a variable for both fixed effects and clustering Standard Error?

[Andrew Musau]

Is it possible to use a variable for both fixed effects and clustering Standard Error?

Yes. That is in fact the default if using an -xt- command with the -robust- option in Stata.