The refined optimal instrumental variable method of time series analysis

The properties of estimators of noise model parameters are investigated. The estimator covariance matrix is taken as a measure of accuracy, and it is shown to be optimized by an appropriate selection of instrumental variable (IV). The refined-optimal IV method is then proposed. The analysis and Monte-Carlo simulation results indicate that the algorithm yields asymptotically efficient estimation results, even for low sample size and low signal/noise ratios.<>