Multiscale Tail Risk Connectedness of Global Stock Markets: A LASSO-Based Network Topology Approach

Yuting Du,Xian Yang,Xu Zhang,Zhijing Ding,Giacomo Fiumara

doi:10.1155/2022/7635144

Yuting Du, Xian Yang + Show 3 more

Open Access

https://doi.org/10.1155/2022/7635144

Copy DOI

Abstract

Due to the advent of deglobalization and regional integration, this article aims to adopt LASSO-based network connectedness to estimate the multiscale tail risk spillover effects of global stock markets. The results show that tail risk varies across frequencies and shocks. In static analysis, the risk is centered mostly on the developed European and North American markets at a low frequency (long term), and regionalization is imposed on the moderate frequency (midterm). Moreover, emerging markets could be sources of risk spillover, especially at the highest frequency (short term) where there is no absolute risk center. In dynamic analysis, we use rolling window estimation and find that different frequencies identify distinct episodes of shocks, which provides us with the reason for the diverse risk centers at different time scales in static analysis. Our findings provide heterogeneous financial practitioners, regulators, and investors with diverse characteristics of stock markets under multiple time horizons and help them operate their own trading strategies.

Highlights

Due to the heterogeneity and complexity of the financial markets [31], and there may be opposite observations when it comes to the relationship between stock returns and macroeconomic variables like inflation [32], we seek to discover the global network connectedness among stock markets at multiple frequencies in this paper
We examine the risk spillover effect of global stock markets including 85 indexes and 62 sovereign countries at multiple frequencies from June 2006 to March 2021
We use the network connectedness estimation introduced by Diebold and Yilmaz [3] to overcome the welldocumented weakness of alternative approaches, and we use wavelet packet decomposition to reconstruct the daily trading returns and least absolute shrinkage and selection operator” (LASSO) to conduct high-dimensional network estimation

Summary

Introduction

Due to the heterogeneity and complexity of the financial markets [31], and there may be opposite observations when it comes to the relationship between stock returns and macroeconomic variables like inflation [32], we seek to discover the global network connectedness among stock markets at multiple frequencies in this paper. We employ a wavelet packet to examine the multifrequency (multiscale) risk spillover connectedness among stock markets across the globe, that is, the characteristics of the risk contagion effect between different countries from the long term to the short term. Even emerging markets can be an unstable source of risk at the highest frequency (short term), which improves the effectiveness of combining macroprudential and monetary policy. Such an observation can be used by central banks. This article broadens our horizons (numerous markets in both developed and emerging countries are included) and captures frequency properties in a more profound way (the use of WPT) when observing the channel of stock market risk contagion.

Objectives

Methods

Conclusion