Abstract

This talk is concerned with analysis of high-dimensional time series. We consider two main issues in analysis of high-dimensional time series. The first issue is testing for serial correlations when the dimension is high and the second issue is variable selection. Limiting distributions of the statistics used in testing and variable selection are derived. Some finite sample corrections are also introduced. We use simulations to demonstrate the performance of the proposed statistics and to compare them with other existing methods, including LASSO. Real examples are also used.