Definition of homoskedasticity (OLS assumption)

[sleepybird]

Hi David,
Does homoskedasticity simply means the variance of the residual is constant across all observation in the sample? But does it also assume that it does not depend on the independent variables?

On some parts of the notes/readings, it simply refers to constant variance but other times it also say it does not depend on independent variables (i.e., conditional/unconditional heteroskedasticity).

Thanks.
Sleepybird

 

[Aleksander Hansen]

Yes, constant variance, provided the variance is defined and/or finite.
One implies the other: if homoskedastic, the residuals do not depend on the regressor(s).