CMStatistics 2018: Start Registration
View Submission - CFE
Title: Measuring asset market linkages: Nonlinear dependence and tail risk Authors:  Juan Carlos Escanciano - Universidad Carlos III de Madrid (Spain) [presenting]
Javier Hualde - Universidad Publica de Navarra (Spain)
Abstract: Traditional measures of dependence in time series are based on correlations or periodograms. These are adequate in many circumstances but, in others, especially when trying to assess market linkages (e.g., financial contagion), might be inappropriate. In particular, tail dependence measures based on correlations of single tail events have limited information on tail risk. We propose a new nonparametric measures of dependence and show how they characterize conditional dependence and persistence. We propose simple estimates for these measures and establish their limiting properties. We employ the proposed methods to analyze the persistence properties of some of the major international stock market indices during and after the 2007-2009 financial crisis. The results uncover a leading role of US and London in international diversification. Tail dependence, as quantified with the new measures, is more informative than the popular marginal expected shortfall for the US. We find a ubiquitous nonlinear persistence in conditional variance across all markets that is not explained by popular parametric models. Market crashes also show substantial persistence in likelihood and their systemic effects.