Karsten,
Sample size can not be "too large". However, when you investigate a
microeconometric panel then the subjects (individuals, firms, ...) are
likely to be cross-sectionally correlated. And this source of
cross-sectional correlation can lead to severely biased standard error
estimates - even if the cross-sectional dependence is rather small.
Therefore, you should use a panel estimator which accounts for
cross-sectional (or "spatial") dependence in the residuals of a panel
regression. Daniel Hoechle implemented the nonparametric Driscoll and
Kraay (1998, Review of Economics and Statistics) covariance matrix
estimator in Stata. You can download his -xtscc- procedure from within
Stata (type -net search xtscc-). The -xtscc- command comes along with
an informative paper on the properties of several commonly applied
covariance matrix estimators when cross-sectional dependence is
present.
I hope this helps.
Best,
Ingo
*
* 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/
