Can autocorrelation mask/conceal (at least in part) collinearity?
Can autocorrelation mask/conceal (at least in part) collinearity?
Because as I understand it, collinearity increases standard error in the regression coefficients and autocorrelation decreases it. So, these two bad things are pulling in opposite directions.
Comments
Autocorrelation decreases standard error only when left unaccounted for. If you properly account for the autocorrelation, the standard error increases (because effectively you have fewer observations). They may pull in opposite directions but probably in a bad way :-)
EJ