On uniform asymptotic normality of sequential estimates of the parameters in unstable autoregression
The paper proposes new sequential least squares estimates for the parameters in autoregressive processes of order đť‘ť. The construction of the procedure, in contrast to those known in the literature, makes use of only one least squares estimate (LSE) for the vector of unknown parameter for any...
| Published in: | МеждŃнародная наŃчная конференция "РобаŃтная ŃтатиŃтика и финанŃовая математика - 2018" (09-11 июля 2018 Đł.) : Ńборник Ńтатей С. 38-47 |
|---|---|
| Main Author: | |
| Other Authors: | |
| Format: | Book Chapter |
| Language: | English |
| Subjects: | |
| Online Access: | http://vital.lib.tsu.ru/vital/access/manager/Repository/vtls:000649668 Перейти в каталог НБ ТГУ |
| Summary: | The paper proposes new sequential least squares estimates for the parameters in autoregressive processes of order đť‘ť. The construction of the procedure, in contrast to those known in the literature, makes use of only one least squares estimate (LSE) for the vector of unknown parameter for any order đť‘ť. The main point is that the sample Fisher information matrix in the LSE is properly modified by introducing special stopping rules for collecting the data. It is shown that in the i.i.d. case with unspecified error distributions, the new estimates have the property of uniform asymptotic normality for unstable autoregressive processes under some general condition on the parameters. The cases of AR(1), AR(2) and AR(3) processes are considered in detail. The results of numerical simulations are given. |
|---|---|
| Bibliography: | Библиогр.: 15 назв. |
| ISBN: | 9785946217507 |