Speaker: Haeran Cho (Bristol)

Title: Detecting (multiple) change-point in panel data

Abstract: In this talk, we propose a method for detecting multiple change-points in the mean of high-dimensional panel data. CUSUM statistics have been widely adopted for change-point detection in both univariate and multivariate data. For the latter, it is of particular interest to exploit the cross-sectional change-point structure and achieve simultaneous change-point detection across the panel, by searching for change-points from the aggregation of multiple series of CUSUM statistics, each of which is computed on a single series of the panel data.

The proposed Double CUSUM statistic achieves consistency in change-point detection in terms of both the total number and the locations of detected change-points. Besides, it attains the high-dimensional efficiency (Aston and Kirch 2015) comparable to that of the oracle change-point statistic under the presence of moderate cross-correlations, without requiring any prior knowledge on the size of changes, their directions or the underlying error structure. The Double CUSUM statistic is compared to other change-point statistic proposed for a single change-point detection in panel data, both theoretically and empirically.