xtlsdvc and xtbcfe

[Prateek Bedi]

Hello,

I am trying to fit a dynamic model determinants of corporate cash holdings for ~1700 firms across 16 years. I have a long list of explanatory variables (10 control variables and 1 variable of interest). In order to do so, I am using LSDV Corrected (xtlsdvc) and Bootstrapped Bias Corrected estimators (xtbcfe) in Stata 14. I am doing these computations on Dell Inspiron 5558 laptop with 4 GB RAM and i5 second generation processor. I have the following queries:

1. Both these estimators (xtlsdvc and xtbcfe) are taking too much time to return the output. Specifically, xtbcfe did not show the results even after two hours and I had to abandon the estimation. Similarly, xtlsdvc took almost an hour to return the output. In such a scenario, what can I do to obtain quicker results?

2. Which estimator should be preferred between xtlsdvc and xtbcfe, and why?

3. Apart from the original papers which introduce these estimators and their respective Stata help files, is there any other source which explains the operational nuances of these estimators. I basically would like to read about the reasoning of choosing a particular option in the syntax of these estimators.

Thanks!!

[Phil Bromiley]

You didn't get a quick answer. You've posted enough times to know how to post. Please follow the FAQ.

These are user written and not among the very widely used procedures. What you see is what you get. As for processing time, let it run overnight.

There are what appear to be similar Stata procedures in the panel estimation section. A Stata version is likely to be better optimized for speed.

[Prateek Bedi]

Thanks for your reply, Sir!