A0774
Title: Parametric estimation for weak fractional time series
Authors: Tetsuya Takabatake - The University of Osaka (Japan) [presenting]
Abstract: Parametric estimation is considered for fractional time series models with general memory parameters and an innovation process satisfying the weak white noise condition. Estimation is performed via a conditional-sum-of-squares estimator, based on truncating the infinite-order autoregressive representation of the observed fractional time series, covering both stationary and non-stationary ranges of the memory parameter. Consistency and asymptotic normality are established under minimal conditions on the innovation process. Moreover, time permitting, the estimator's performance is also demonstrated through simulations.
