Seasonal adjustment of time series observed at mixed frequencies using singular value decomposition with wavelet thresholding
报告人： Wei Lin，The University of International Business and Economics
地点：Room 217, Guanghua Building 2
In this paper, we propose a flexible seasonal adjustment method that accommodates the seasonal time series observed at mixed frequencies and possessing abrupt changes of seasonality under the innocuous assumption that the nonseasonal component is stationary or difference stationary. Through a generalized difference, we remove the stochastic trend of the mixed frequencies time series. At the same time, we express the seasonal component into the matrix that has a low rank SVD structure. The right singular vectors is deemed as the seasonal patterns and the corresponding left singular vectors capture the smoothness feature of seasonal patterns. To estimate the SVD structure of seasonality and thus recover the seasonal component, we propose an effective algorithm that applies the wavelet smoothing technique to the left singular vectors. Our proposed method not only encourages persistence of seasonality, but also allows for the possible existence of abrupt changes in seasonality. Using simulated data, we find that (i) when the seasonality is moderate or strong our proposed method performs well by delivering small absolute and relative losses of estimated seasonal seasonality and correctly detecting the underlying seasonality structure; and (ii) the performance of our proposed method improves further as the seasonality becomes more salient, the proportion of high frequency observations in the sample becomes larger, and the number of observations increases.
About the Speaker:
林蔚，现任对外经济贸易大学国际经济贸易学院副教授。于2013年获得美国加州大学河滨分校经济学博士。他的研究领域是应用计量经济学，研究兴趣包括非参数方法，时间序列分析和季节调整等方面。他的研究成果已发表于Journal of Business and Economic Statistics, Journal of Econometrics, Journal of Applied Econometrics等学术刊物上。
Your participation is warmly welcomed!