A1311
Title: Normal mixture quasi-maximum likelihood estimation of double autoregressive models
Authors: Christina Dan Wang - NYU Shanghai (China) [presenting]
Abstract: The estimation of a double autoregressive model (DAR) with skewed and heavy-tailed innovation is investigated. A new estimation method, the normal mixture quasi-maximum likelihood estimation (NM-QMLE), is proposed to estimate the DAR model with the non-Gaussian behavior. Under regularity conditions, consistency and asymptotic normality are established for NMQMLE. The numerical simulation for the DAR model with heavy-tailed and skewed innovation indicates that the NMQMLE outperforms several commonly adopted QMLE's. Finally, An empirical example on the S\&P 500 index illustrates the application of the new estimation method.