Dynamic structural systems under indirect observation: Identifiability and estimation aspects from a system theoretic perspective (Q1090052)

In this expository paper, the author presents similarities between linear stochastic state space models and structural dynamic models with latent variables (the last one is used in Social Sciences). The classical concepts, in systems theory, of observability and controllability, and theorems for local parameter identifiability are discussed. Problems of latent vectors estimation - by Kalman-Bucy filters - and of consistent estimation of unknown model parameters - by non-linear M. L. estimation procedures - are also discussed. Finally, it is shown that Jöreskog's LISREL procedure is a special case of this general method.