there are some previous posts dealing with heteroskedasticity and autocorrelation in panel data, nevertheless I have not found any post discussing heteroskedasticity and cross-sectional dependence at the same time.
I am analysing some panel data, which suffer from both aforementioned issues. To deal with a cross-sectional dependence I use a Driscoll And Kraay (1998) Robust Covariance Matrix Estimator (function vcovSCC in R). To deal with heteroskedasticity I use a classic Robust Covariance Matrix (function vcovHC, specifically method arellano).
Obviously, both methods result in different SE. Is there a way how to account for both heteroskedasticity and cross-sectional dependence at the same time? If not, which of these robust cov. matrices should I use?