Estimation of time-invariant regressors in panel models: How to implement 'xtfef' command in Stata?

[Emna Trabelsi]

Dear Statalist users,

Does someone know how to implement the procedure suggested in Pesaran and Zhou (2017) and Law and Zhou (2017) to estimate time-invariant variables in in a fixed panel model? The 'xtfef' command used in Law and Zhou (2017) is not available in Stata.

Kind regards,

Emna

References

1- Pesaran, M. H. and Q. Zhou (2017). Estimation of time-invariant effects in static panel data models. Econometric Reviews, 1137-1171.
2- Law, Y and Q. Zhou (2017). Estimation of time-invariant effects in both balanced and unbalanced static panel models: A discussion of the “xtfef” command for FEF and FEF-IV estimators. Working Paper.

[Justin Niakamal]

Looks like it can be found in #10 here

https://qiankunzhou.weebly.com/research.html

[Sebastian Kripfganz]

The FEF method of Pesaran and Zhou is a special case of the two-stage procedure discussed in the following paper:

• Kripfganz, S., and C. Schwarz (2019). Estimation of linear dynamic panel data models with time-invariant regressors. Journal of Applied Econometrics 34 (4), 526-546.

The two-stage procedure is implemented in my xtseqreg Stata package.

[Emna Trabelsi]

The FEF method of Pesaran and Zhou is a special case of the two-stage procedure discussed in the following paper:

• Kripfganz, S., and C. Schwarz (2019). Estimation of linear dynamic panel data models with time-invariant regressors. Journal of Applied Econometrics 34 (4), 526-546.

The two-stage procedure is implemented in my xtseqreg Stata package.

Dear Sebastian,
If I understand well, 'xtseqreg' handles time invariant variables in both static and dynamic panel models. Doesn't it ?

Best,
Emna

[Sebastian Kripfganz]

That is correct. You can use xtseqreg for static or dynamic models.

[Emna Trabelsi]

OK. Thank you so much.

[Azharuddin Akhtar]

Hi Sebastian. I have count data as an outcome (0-3, 0-4) and the treatment indicator is an endogenous in two period panel data. The problem is the instrument we have taken is only provided in single period under the assumption it will continue to be same in the second period. The procedure provided in the discussion apparently for same issue. My question is does the command appropriate for count data, if not then do we have to normalize it? Please let me know I got stuck in the estimate process really bad.

[Sebastian Kripfganz]

You can apply the xtseqreg command as long as you are estimating a linear regression model, no matter what type of data.

[Eric Atsu Avorkpo]

Hi Sebastian. Thanks for the information you posted "You can apply the xtseqreg command as long as you are estimating a linear regression model, no matter what type of data." Is this applicable to dynamic relationship? where we introduce the lag of the dependent variable as an independent variable?
Can you please suggest any other model that can serve as althernative to the xtseqreg model?