[Gretl-users] About Durbin-Watson reported on panel data (FE) estimation

Hello everyone, Which version of the Durbin-Watson test for autocorrelation is the one reported on the output of the Fixed Effects estimation using panel data? I read that the modification of the original test for balanced panel data is by Bhargava et al. (1982) and a version for unbalanced panel data is from Baltagi and Wu (1999). Which one is being reported on Gretl? Thanks a lot! IP.