项目名称: 一类半参数时间序列模型的统计推断
项目编号: No.11271095
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 数理科学和化学
项目作者: 李元
作者单位: 广州大学
项目金额: 68万元
中文摘要: 半参数时间序列模型为近年来发展起来一类新的时间序列模型,它既具有参数模型的优点,同时又融合了非参数模型的优点。因此它成为时间序列研究的一个热点问题。本项目致力于研究半参数 GARCH 模型,它包括:半参数 GRACH-M 模型,单指数 GARCH 模型,部分线性 GARCH 模型,变系数 GARCH 模型等。我们将通过连续状态的马氏链方法来研究该类模型中半参数自回归模型的平稳遍历性和矩存在条件。通过截面似然方法和样条方法来建立半参数GARCH模型的参数部分和非参数部分的估计,在观测变量平稳和非平稳的条件下来讨论估计的渐近性质。然后借助于经验似然方法来讨论半参数GARCH模型参数部分和非参数函数部分的统计推断问题,建立其置信区间。最后进行数值模拟和实证分析研究。该项目的研究将大大丰富时间序列的半参数和非参数方法。
中文关键词: 半参数模型;时间序列;变系数;非平稳;估计
英文摘要: Semiparametric time series models have become one class of new models developed in recent years. These models have advantages of both parametric models and nonparametric models, which makes it greatly concerned in time series.We deal with semiparametric GARCH models in this project, which inculdes semiparametric GARCH-M models, single index GARCH models, partial linear GARCH models, varying coefficients GARCH models, etc. Conitions for stationarity, ergodicity and moments of autoregressive models will be studied by Markov chain methods of continuous states of time series. Estimates of parameters and nonparametric functions will be given by the profile likelyhood method and spline method. Asympotic properties of estimates will be discussed in both stationary and nonstationary conditions of explantory variables. Then inference on parameters and nonparametric functions in models will be made by empirical likelyhood methods and their confidence intervals will be given.Finally simulations and empirical studies will be done. This project will enrich theories of semiparametric and nonparametric time series.
英文关键词: semiparametric model;time series;varying coefficient;nonstationary;estimator