A1488
Title: The likelihood ratio test for changes in high-dimensional idiosyncratic network
Authors: Xiaohan Xue - University of Bath (United Kingdom) [presenting]
Abstract: A novel framework is introduced, employing the likelihood ratio test to detect change points in high-dimensional networks. The framework introduces a sophisticated factor model for handling common variations in stock returns, employing tools like Graphical Lasso for estimating sparse precision matrices essential in high-dimensional settings. The simulations demonstrate the framework's effectiveness in controlling type I error rates and its power to detect true positives under various scenarios. Applying the methodology to the daily returns of S\&P 500 constituents illustrates practical implications. The results identify multiple structural breaks coinciding with major economic events, underscoring the test's potential in real-world scenarios where detecting such breaks can significantly impact investment strategies and risk assessment.
