Vector Autoregressive Order Selection in Practice

Vector time series analysis takes the same model order and model type for the different signals involved. Selection criteria have been developed to select the best order to simultaneously predict the different components of the vector. The prediction of single channels might require a different order or type for the best accuracy of each separate signal. That can become a problem in multichannel analysis if the individual signals have completely different model orders. Therefore, univariate and multichannel spectra are not similar. Furthermore, the selected order may vary in practice with the number of signals that are included in a vector. A turbulence example shows the results of order selection and the consequences in estimating the coherency between the same two components from vector signals with dimensions two and five.