CFE 2020: Start Registration
View Submission - CMStatistics
B1189
Title: Nonlinear causality for CHARN models Authors:  Xiaoling Dou - Waseda University (Japan) [presenting]
Abstract: The CHARN model was proposed in financial data analysis. Because of its non-normality, non-linearity and the blindingly general form, it has come into use in various fields of time series. We consider a nonlinear causality for the CHARN models. We show that the causality of the CHARN models can be evaluated by a Portmanteau test, based on a constrained maximum likelihood estimator of the parameters, and the test statistic has an approximate asymptotic Chi-square distribution. We describe the Chi-square Asymptotics of the Portmanteau test for a CHARN model, provide calculations of the test statistic and investigate the performance of the Portmanteau test by simulation.