if you go -robust- you have also taken into account possible autocorrelation of the epsilon residual.
That said, the most substantive post estimation test should be aimed at checking for possible model (or better regressand functional form) misspecification.
Take a look at the following toy-example, when an ancillary and an augmented regressions are run with -fitted- and -sq_fitted- values as predictors:
复制代码
As expected, the -test- outcome shows that the regression model is misspecified (because -age- taken as the unique predictor cannot give a fair and true view of the data generating process under investigation. Moreover, -age- has a non-linear relatinship with the regressand, as it can be easily found out by replacing -age- with -c.age##c.age-).
[code]
xtreg ln_wage c.age##c.age, fe robust
Fixed-effects (within) regression Number of obs = 28,510
Group variable: idcode Number of groups = 4,710
R-sq: Obs per group:
within = 0.1087 min = 1
between = 0.1006 avg = 6.1
overall = 0.0865 max = 15