questions for a fixed-effect regression

[Sally Lee]

Hi，everyone. I have a short panel data. N=32 T=18. The result of the hausman test is that ein fixed effect regression is perferred and I'm not sure if I should put i.year in the regression.

Another question ist that I found the problem of hetroscedasticity, autocorrelation and cross sectional dependency.I'm not sure which model I should use for the regression. I have seen in many posts that xtpcse or xtgls only for long panel data.
Can I use xtscc for a short panel data when there is hetroscedasticity, autocorrelation and cross sectional dependency? Or is xtreg, fe robust enough for hetroscedasticity, autocorrelation and cross sectional dependency?

looking forward for replies and I really appreciate it !

[Sally Lee]

And If I use xtreg i.year, fe robust then the result of testparm i.year is 0.0004

[Carlo Lazzaro]

Sally:
1) -xtreg,fe- requires -i.timevar- to be included in the right-hand side of the regression equation. Conversely, -xtreg- non-default standard errors do not take across panels correlation into account;
2) I would go -xtscc- (community-contributed module, as the FAQ recommend to call it), as it covers heteroskedasticit, autocorrelation and correlated between panels. Including -i.panelvar- is not necessary if you go -xtscc-. Please also note the -xtscc- does not support -fvvarlist- notation.