Hello everyone,

I have a question concerning autocorrelation in data affecting correlation coefficients. I have a time series of stock returns (e.g. 20 trading days) of different stocks (A, B, C and D) and I just want to calculate simple correlation coefficients (let it be Pearson's) for each pair of stock markets (AB, AC, ...). A paper I read therefore estimated returns via a VAR framework to control for serial correlation (and exogenous shocks) in the first place and computed afterwards the correlations.

Regarding to literature stock returns are often affected by autocorrelation and therefore should be controlled. As I'm not familiar with time series, I read a little bit about it and found, that autocorrelation causes problems for estimating OLS coefficient causing them to be inefficient.

My question is now, does it have any consequences if I just use my observations (without any adjustment for autocorrelation) and do the simple correlation calculations, because I won't estimate anything with it afterwards? (my literature just said, that it has only OLS effects...) Think that I just read too much for today and am now totally confused.

If I need do control for serial autocorrelation in the first place, is there any possibility to do it easier/otherwise than using a VAR framework?

Thank you!