含趋势项时间序列持久性变点监测

针对核加权方差比率统计量不是监测从非平稳向平稳变化持久性变点一致方法的问题，通过引进一个窗宽参数，提出了一种滑动核加权方差比率统计量来监测含趋势项时间序列从非平稳向平稳变化的持久性.在非平稳原假设下给出了监测统计量的极限分布和经验临界值表，在备择假设下证明了新方法的一致性.模拟结果表明新方法具有比原方法更高的势和更短的平均运行长度，最后通过分析人民币与美元汇率数据进一步说明了该方法的有效性.

Monitoring change in persistence in time series with deterministic trend

By introducing a bandwidth parameter, this paper proposes a moving kernel-weighted variance ratio statistic to monitor persistence change in time series with deterministic trend in terms that the existing kernel-weighted variance ratio statistic is not consistent when time series switch from nonstationary to stationary. Under the nonstationary null hypothesis we derive the limiting distribution of monitoring statistic, and tabulate some critical values. We also prove its consistency under the alternative hypothesis. Simulations indicate that the new procedure has higher empirical power and shorter average run length than the existing procedures in the literature. Finally, the validity of the new procedure is demonstrated by analyzing the foreign exchange rate dada of Chinese Yuan and U.S. dollars.