Robust Regression or Robust Errors

[Alex King]

Hi,

I'm fairly new to stata and am trying to estimate the effect of drug crimes on house prices. I am using a fixed effects model but it suffers from heteroskedasticity and outliers. Therefore, I am not sure wether to use rreg with i.Year and i.WardNumber as time and entity fixed effects or to run a xtreg, fe robust regression. (Note that in the rreg regression I cropped the ward number coefficients but I have controlled for the ward fixed effects). Is there not a way to control for outliers and heteroskedasticity at the same time? I prefer the rreg results as the coefficients are significant but is it wrong to not control for heteresokedasticity?


Any advice would be appreciated!

Alex

[Richard Williams]

I know that rreg is generally looked down upon. For alternatives, see

http://www.stata-journal.com/article...article=st0173