Dear Sir/Madam
I have a dateset that suffers from heteroskedasticity and serial correlation. To address the serial correlation lag my variables of the fixed effect regression. I retrieve the residual and I regress it on its lag. I find that there is not serial correlation anymore. Hence, I run my fixed effec regression with vce(robust) standard error in stata to address the heteroskedasticity issue as well. Nevertheless, the residual of my fixed effect regression with vce(robust)standard error shows serial correlation again when I regress it on its lag. Does it mean that I have serial correlation? My dataset is unbalance so the only test in STATA that I can use to check serial correlation is xtserial which can not be used again when I use lag variables. What is the best approach to address the issue? Is there also a test to test for endogeneity?
Dear Ale Zappala,
I suggest you to see links and attached files on topic.
https://www.stata.com/manuals14/xt.pdf
https://www.surveydesign.com.au/docs/manuals/stata16/xt.pdf
Best regards
- Badges
- Science topic