Optimal Lag Structure Selection in VEC-Models
Abstract
For the modelling of time series, multivariate linear and nonlinear systems of equations became a standard tool. These models are also applied for non-stationary processes. However, estimation results in finite samples might depend on the specification of the model dynamics. We propose a method for automatic identification of the dynamic part of VEC-models. Model selection is based on a modified information criterion. The resulting complex discrete optimization problem is tackled using a hybrid heuristic. We present the algorithm and results of a simulation study indicating the performance both with regard to the dynamic structure and the rank selection in the VEC-model.