5
$\begingroup$

Background:

I have estimated a model using panel data with the Arellano Bond estimator (see e.g., http://www.fordham.edu/economics/mcleod/Elitz-usingArellano%E2%80%93BondGMMEstimators.pdf) and n=300. As it turns out, the distribution of my residuals is significantly different from the normal wrt. skewness and kurtosis.

Question:

Is it possible/how can I infer whether the calculated p-values should be higher or lower given the information available (dist, sd., kurtosis, skewness, etc.)?

I hope my question is sufficiently clear.

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

1 – non-normal residuals might indicate poor model fit. Panel-GMM models are quite tricky, something might be wrong with you specification.

2 – Have you tried the robust standard error calculation methods?

Improve this answer
$\endgroup$
1
  • $\begingroup$ 1 - Yes, that's true. Problem is that I have a lack of relevant data to correct for it, and need some tip of whether or not my inference now is likely to be "less valid" when correcting. 2 - Yes. $\endgroup$
    – hoyem
    Jan 31, 2014 at 10:11

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.