Autocorrelation, the correlation of a signal with a delayed copy of itself, can significantly impact the efficiency of an Ordinary Least Squares (OLS) estimator. When autocorrelation is present in the residuals of an OLS model, it violates one of the key Gauss-Markov assumptions – that errors are uncorrelated. This violation leads to inefficient parameter estimates because the standard errors tend to be underestimated, making statistical inference unreliable. The OLS estimator remains unbiased and consistent but loses its Best Linear Unbiased Estimator (BLUE) property, meaning there could be other estimators providing more accurate results. Therefore, detecting and correcting for autocorrelation is crucial when using OLS estimation to ensure efficient and reliable outcomes.
