Identification of simultaneous equation models with measurement errors based on time series structure (Q1118309)

The paper establishes simple conditions leading to the global identifiability of multivariate ARMAX errors-in-variables models in the form of a simultaneous equation system with errors in the exogenous variables. The identification is based on the conventional rank condition and on time series structure inherent to the model. The conditions particularly identify models with non-autocorrelated exogenous variables in the absence of over-identifying zero restrictions. The model allows autocorrelated and cross-correlated errors.