Conditional Heteroscedasticity

In one exercise there was a question regarding Conditional Heteroscedasticity and whether it implies: 1. the independent, 2. depentent or 3. both correlated with the error term. I am aware that this type of Heteroscedasticity involves the independent varibale being correlated with the error. However, I thought that the dependent variable somehow always would be correlated with the error, because the regression equation cannot explain all variation. So I thought it was a trick question and that the answer should be both. Could someone tell me why the dependent variable is not correlated?

It’s not a violation if the Dependent variable correlates with the errors, because the Dependent is the output.

Conditional Heteroskedasticity is a phenomenon (size of X values affects size of errors) that violates the assumption.

Yes I know it is not a violation, but as I said. I thought the dependent always would correlate with the errors somehow, so that the answer should be both independent and dependent.

The dependent variable doesnt matter for CH