经济与金融系学术讨论会第40期

题 目：Market Interdependence, Systemic Risk and Diagnostic Tests in Large Mixed Panels

主讲人： Cindy Shin Huei Wang（台湾清华大学，副教授）

时 间：2018年11月27日（周二）12:00-13:00

地 点：同济大厦A楼505室

摘要：

This paper investigates an AR(Autoregressive)-filtered version of several conventional diagnostic tests for cross sectional dependence in large mixed panels, including the adjusted LM test, the CD test by Pesaran (2008, 2015) and the Schott test by Schott (2005). We show that the revised tests asymptotically follow the standard normal distribution. The distinctive feature of these new tests is its simplicity to implement even though the exact time series property of each component of a panel is unknown or unobservable or the panel is large with mixed time series properties. Simulations show that the AR-filtered version of the CD test perform the best over other testing procedures in term of the finite sample performance and the computation time, especially for those cases with large cross-sectional dimension N (100 and 500). We also provide a new perspective to the linkages between asset returns of banking and non-banking firms and the volatility interrelation in global equity markets before and after economics activities using our novel AR-filtered CD statistics as it could behave like an early warning indicator of systemic risk and financial crises.