Hello,
I am using the XTLSDVC command to run a dynamic panel
data model with fixed effects. I have a reviewer who is insistent that I supply
and r-squared, although XTLSDVC does not supply one. As I understand it, the
LSDVC is correcting bias in the coefficients and their standard errors, but
overall model fit should be equivalent to the LSDV results (correct?). Thus, would
it be appropriate to run an XTREG with the lag of the DV and fixed effects and
use the R-squared from that regression? If so which R-squared should I report:
within, between or overall?
Thank you for any help or suggestions,
Jon O’Brien
UCD School of Business
Dublin, Ireland