As a result of this, the usual t-statistics, F-statistic, R2, and χ2 of the affected OLS may no longer be valid. For instance, positive serial correlation in an OLS estimate will not affect the least squares slope estimates on the average (i.e. they are unbiased).