Hi Stata Listers,
I am using the -nl- command to do nonlinear least squares. The equation I am estimating looks something like this " Y = T * X1 + (T * D) * X2 + (T * D^2) * X3 + (T * D^3) * X4 + a ton of fixed effects + time trends". (I have just over 4,000 observations - 40 countries over 104 weeks.) Thanks to an earlier Stata list answer from Kit Baum concerning -_robust- I was able to calculate a robust covariance matrix for the estimation of this model. My new question concerns the change in my standard error estimates.
My robust standard error estimates where much less than the original estimates. For instance the coefficient T was estimated as -.015 and the original standard error was .0541, while the robust standard error is .0034. This is more than a 10 fold decrease, while what I could find in the literature suggested that robust standard errors are usually larger than the original standard errors.
Although I am quite happy to have small standard errors I am worried that this could be an indication of a problem either in my model or in my estimation techniques. Does this change in standard errors ring any bells for anyone? I have searched several econometrics texts and haven't found anything.
Any help or direction you could provide would be appreciated.
Thanks,
Larry Chavis
PhD Candidate
Stanford Graduate School of Business
Stanford CA 94305
650-724-4909
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/