### mixed model: matrix of covariance of residuals

Posted:

**Fri Sep 13, 2013 2:54 pm**How can I evaluate the Matrix of covariance of residual that the routine produces? I have a panel of about 32000 observations (firm by year) and I am sure of the existence of a positive longitudinal correlation of Yit (a measure of productivity). So I indicate in the first window of the mixed model routine firms as subjects and year as repeated mesures. Therefor I indicate AR(1)-eterogeneous as covariance model of repeated meaures. Is the covariance matrix of residuals the starting Matrix used in the parameters estimation or is it the covariance structure which results after the model estimation, I mean the ending Matrix of covariance? many thanks