Dear all,
I am doing quantile analysis on a large bilateral panel dataset of Mergers and Acquisitions investments (M&A), with a gravity approach. I have a large number of zero flows, therefore standard quantile regression is not advisable. I already computed non-IV regression using the mcmccqreg command written by Matthew J. Baker. However, it is taking a very long time to converge because of the large sample size, the large number of zeroes, and the huge amount of fixed effects. I found this paper
where an interesting approach based on a 3 stages censored quantile regression is developed, to take into account the large amount of fixed effects in panel data. This methodology seems suitable for quantile regression with high dimensional FE, and it has been applied also in this paper
where they also implemented an IV analysis based on such 3-steps estimator.
I already found in a previous thread the "STATA translation" of a similar approach to the one I would like to apply (it can be found here: https://www.statalist.org/forums/for...for-panel-data), and I was wandering whether somebody could give me some clue to implement the approach by Figueiredo et al (2014, 2016).
Best,
Filippo Santi
I am doing quantile analysis on a large bilateral panel dataset of Mergers and Acquisitions investments (M&A), with a gravity approach. I have a large number of zero flows, therefore standard quantile regression is not advisable. I already computed non-IV regression using the mcmccqreg command written by Matthew J. Baker. However, it is taking a very long time to converge because of the large sample size, the large number of zeroes, and the huge amount of fixed effects. I found this paper
When elephants mate do ants get crushed? The WTO impact on trade dispersion. E Figueiredo, LR Lima, G Schaur Working Paper, University of Tennessee at Knoxville, 2014.
Figueiredo, E., Lima, L. R., & Orefice, G. (2016). Migration and regional trade agreements: A (new) gravity estimation. Review of international economics, 24(1), 99-125.
I already found in a previous thread the "STATA translation" of a similar approach to the one I would like to apply (it can be found here: https://www.statalist.org/forums/for...for-panel-data), and I was wandering whether somebody could give me some clue to implement the approach by Figueiredo et al (2014, 2016).
Best,
Filippo Santi
