A0812
Title: Causality tests between two autoregressive time series under weak dependency
Authors: Chafik Bouhaddioui - UAE University (United Arab Emirates) [presenting]
Abstract: A generalized test for non-causality is proposed at various horizons for infinite-order vector autoregressive. Multiple-horizon infinite-order VAR is first introduced, which can be approximated by multiple-horizon finite-order vector autoregressions. The order is assumed to increase with the sample size. Under some weak conditions on the error terms, the estimation of parameters obtained from the approximation of the infinite-order autoregression by a finite-order autoregression is studied. The test statistics can be implemented through linear regression methods where the errors are weakly dependent. The asymptotic distribution of the new test statistic is derived under the hypothesis of non-causality at various horizons. An asymptotic power of the test is studied. A parametric bootstrap procedure is also considered. The method is applied to a VAR model from Canadian economic data.
