Suppose you want to estimate the model yt o 1x ut but you

Suppose you want to estimate the model: yt = o + 1x + ut but you find that in a regression of residuals ût on ût-1 that the coefficient on ût-1 is .52 and highly significant. How can you correct for the serial correlation and estimate the original equation under the Gauss-Markov assumptions? Assume exogenous regressors.