Optimization of the Generalized Covariance Estimator in Noncausal Processes

This paper investigates the performance of the Generalized Covariance estimator (GCov) in estimating mixed causal and noncausal Vector Autoregressive (VAR) models. The GCov estimator is a semi-parametric method that minimizes an objective function without making any assumptions about the error distribution and is based on nonlinear autocovariances to identify the causal and noncausal orders of the mixed VAR. When the number and type of nonlinear autocovariances included in the objective function of a GCov estimator is insufficient/inadequate, or the error density is too close to the Gaussian, identification issues can arise, resulting in local minima in the objective function of the estimator at parameter values associated with incorrect causal and noncausal orders. Then, depending on the starting point, the optimization algorithm may converge to a local minimum, leading to inaccurate estimates. To circumvent this issue, the paper proposes the use of the Simulated Annealing (SA) optimization algorithm as an alternative to conventional numerical optimization methods. The results demonstrate that the SA optimization algorithm performs effectively when applied to multivariate mixed VAR models, successfully eliminating the effects of local minima. The approach is illustrated by simulations and an empirical application of a bivariate mixed VAR model with commodity price series.