CMStatistics 2016: Start Registration
View Submission - CFE
A1403
Title: Semi-parametric estimation of noncausal vector autoregression Authors:  Christian Gourieroux - University of Toronto and CREST (Canada) [presenting]
Joann Jasiak - York University (Canada)
Abstract: Consistent semi-parametric estimation methods are introduced for mixed causal/noncausal multivariate non-Gaussian processes. We show that in the VAR model, the second-order identification is feasible to some extent, and it is possible to distinguish the mixed processes with different numbers of causal and noncausal components. For detecting the causal and noncausal components, we introduce a semi-parametric estimation method that does not require any distributional assumptions on the errors. This method is based on the generalized covariance estimator of the autoregressive coefficients. Although this estimator is not fully efficient, it provides the estimates in one single optimization, while the MLE would assume a parametric speciffication and require a number of optimizations equal to the number of all possible causal dimensions. The method is illustrated by a simulation study and applied to commodity prices.