Errors-in-variables identification using covariance matching and structural equation modeling

Two approaches for errors-in-variables identification are compared. Covariance matching (CM) is known to be a computationally efficient method with good performance. Structural equation modeling (SEM) has been used for many years for static problems, particularly for social science applications. It is shown here how the SEM approach can be applied also for dynamic (time-series) problems, and that the resulting method is closely related to the CM approach.