EcoSta 2022: Start Registration
View Submission - EcoSta2022
A0582
Title: Weighted estimation procedures for time-varying heavy-tailed processes Authors:  Fumiya Akashi - University of Tokyo (Japan) [presenting]
Junichi Hirukawa - Niigata University (Japan)
Konstantinos Fokianos - University of Cyprus (Cyprus)
Abstract: A family of locally stationary processes is often useful when we model real data. Results for the finite variance case are extended to those of infinite variance cases. We consider a parameter estimation problem of autoregressive models with time-varying coefficients and propose a self-weighted local least absolute deviation regression estimator. The model is possibly finite or infinite variance and heteroscedastic one, and under mild conditions for the moment of the model, we show asymptotic normality of the proposed estimator. Some simulation experiments illustrate the finite sample performance of the proposed estimator.