[Eric de Souza]

There are two solutions. Either you manage to detect the source of heteroscedasticity and autocorrelation (if you have time series data) and find regressors to eliminate both; or you robustify the standard errors for any residual heteroscedasticity and/or autocorrelation in the residuals

[Mathias Saeten]

I test different calendar effects on different indices. Some have heteroscedasticity and some have not. Some have autocorrelation and some others have not. Would it be a good idea to apply Newey-West standard errors on all regressions just to be consistent? I know Newey-west standard errors produces consistent std.erros in case of hetroscedasticity AND autocorrelation. But what if I only have autocorrelation for example? Or only hetroscedasticity? Would Newey-West bias my standard errors?

[Mathias Saeten]

Please help

[Eric de Souza]

Prof Wooldridge has given you the answer:
" Personally, I would report both the OLS estimates and the GARCH(1,1) estimates. But in the first case, make the standard errors robust to heteroskedasticity, just to be consistent. In the GARCH(1,1) case, compute the
Bollserslev-Wooldridge standard errors that are robust to nonnormality using the vce(robust) option. "