Wooldridge (2013) says that the Breusch-Godfrey test "requires the homoscedasticity assumption".
I tested for heteroscedasticity in my model by applying the Breusch-Pagan test. The test indicates that my residuals are not homoscedastic. As a next step I am trying to decide wether the residuals are autocorrelated or not. I applied the Breusch-Godfrey test, which indicates that the residuals are autocorrelated.
Is the result of the Breusch-Godfrey test valid in this case? If not - how do I test for autocorrelation in the presence of heteroscedasticity?