I have a difficult panel data set I am working with and wanted to be sure I was headed in the right direction as there are so many assumptions made for different models.
My data includes new business ventures with approximately 70 percent attrition due to failure or merger from 1994 to 2002 for my 1994 cohort (unbalanced).� My data would also be heteroskedastic and serially correlated.� I also would like to test higher order terms as shown by the following model:� �
Variables (Y) Performance (OIBD), (X) Cash slack, (Z) Dynamism
The equation would take the form:
Y = b1X + b2X^2 + b3Z + b4XZ + b5X^2Z + b0
I am currently using (controls not shown):
xtgls ..., panels(heteroskedastic) corr(psar1) force
My plan is to use FGLS (xtgls) and then perform a Heckman correction for the unbalanced panel issue.� Does this seem like a reasonable approach?
Thanks,
Steve
*
* 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/
