报告题目：On the least squares estimation of 2-threshold-variable autoregressive models
报 告 人：李东 清华大学 副教授
报告时间：2020年6月18日 上午 10:40-11:40
报告地点：腾讯会议
点击链接入会，或添加至会议列表：
https://meeting.tencent.com/s/45Q0l8jvECuM
会议 ID：214 844 911
会议密码：200618
校内联系人：赵世舜 zhaoss@jlu.edu.cn
报告摘要：
Most threshold models contain a single threshold variable. However, in many empirical applications, models with two or perhaps more threshold variables may be needed. This paper develops the least squares estimation (LSE) of the 2-threshold-variable autoregressive (2-TAR) model and studies its asymptotic properties. We show that the LSE is strongly consistent and the estimated thresholds are n-consistent and asymptotically independent, each converging weakly to the smallest minimizer of a one-dimensional two-sided compound Poisson process. The slope parameters are √n-consistent and asymptotically normal, being asymptotically independent of the estimated thresholds. To construct confidence intervals of threshold parameters, a local logistic method is employed to simulate the limiting distribution of the estimated threshold. Simulation studies are conducted to assess the finite-sample performance of the LSE. Finally, we present two real examples, one on the U.S. GNP and the other stock returns, to illustrate the efficacy of our modelling. Our results can be extended easily to the k-threshold-variable case,k>2 , although for most practical situations k=2 seems sufficient.
报告人简介：
李东，清华大学统计学研究中心长聘副教授，2010年毕业于香港科技大学，2013年加入清华大学。主要研究兴趣：非线性时间序列分析，金融计量学，网络数据分析与大数据。目前担任全国工业统计学教学研究会常务理事，中国青年统计学家协会常务理事，北京大数据协会常务理事，中国概率统计学会副秘书长，中国现场统计研究会计算统计分会理事，北京应用统计学会理事。主持国家自然科学基金委面上项目2项；结题青年基金项目1项。
