estimating panel data regression
Posted: Sun Mar 21, 2010 8:04 am
by ec07ssn
i have cross sec id, date id, 1 dependent variable and 2 explanatory variables in a panel data. now how do i run the regression for that... bcos when i use the normal estimate equation i get the t stat too high.
Re: estimating panel data regression
Posted: Thu Apr 22, 2010 2:17 pm
by Olufemi
The High T statistic could be due to both between and within or what we call inter and intra class Heteroscedasticity and thus you may consider using the Feasible Weighted least squares or Generalized least squares to overcome the problem of inter class heteroscedasticity but you will still need the Newey west to generate within class robust standard errors and thus your T statistic will be fine.