EcoSta 2021: Start Registration
View Submission - EcoSta2021
A0432
Title: Bayesian modeling of time-varying extremal dependence in international stock markets Authors:  Junho Lee - University of Edinburgh (United Kingdom) [presenting]
Miguel de Carvalho - FCiencias.ID - Associacao para a Investigacao e Desenvolvimento de Ciencias (Portugal)
Antonio Rua - Banco de Portugal (Portugal)
Abstract: A Bayesian time-varying model is proposed to capture the dynamics of extreme joint losses in international stock markets over the last thirty years. The model relies on dual nonparametric time-varying extremal dependence measures, which can be used to assess the strength of dependence of extreme joint losses over time under the settings of asymptotic dependence and asymptotic independence. These measures are approximate by generalized additive models with a large threshold. The dynamics underlying the extremal dependence structure are tracked using Bayesian smoothing methods based on penalized B-splines. A simulation study is presented to assess the performance of the proposed methods. We analyse five international stock market indices and reveal complex extremal dependence behaviours among the indices, suggesting evidence for smooth transitions between regimes of asymptotic dependence and asymptotic independence.