Hi!
I run an unbalanced panel with a total of 475,000 firm-years spread over up to 12 years per firm. Firstly, I am not able to turn more than app. 10 % of the sample in xtabond2 due to size limits, but even with xtabond, I have problems. Whichever way I specify my rather simple regression of debt-ratio, I cannot avoid that the Sargan test rejects the null of no over-identification. It is frequently discussed that this test has finite sample biases, but does anyone know whether it is reliable for really large samples? - and would there be any relevant alternatives? If not, what are the potential solutions?
Thanks for any advice!
Best regards
Aksel Mjos
Norwegian School of Economics and Business Administration, Bergen
*
* 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/
