A0543
Title: Generalized covariance estimator under misspecification and constraints
Authors: Aryan Manafi Neyazi - York University (Canada) [presenting]
Abstract: The asymptotic properties of the generalized covariance (GCov) estimator are investigated under misspecification. We show that GCov is consistent and has an asymptotically normal distribution under misspecification. Then, GCov-based Wald-type and score-type tests are constructed, all of which follow a chi-square distribution. Furthermore, the indirect GCov(IGCov) and the constrained GCov (CGCov) estimators are proposed. The IGCov estimator is useful for estimating models indirectly, based on simulations, such as non-invertible moving average models. Consequently, an IGCov specification test is developed. The CGCov estimator extends the use of the GCov estimator to a broader range of models with constraints on their parameters. The asymptotic distribution of the CGCov estimator is investigated when the true parameters are far from the boundary and on the boundary of the parameter space. The finite sample performance of proposed estimators and tests is validated in the context of noncausal-noninvertible and DAR models. Finally, two empirical applications are provided by applying the noncausal model to the final energy demand commodity index and also the DAR model to the US 3-month treasury bill.
