New nonparametric measures for instantaneous and granger-causality tail co-dependence.

Area of Science:

• Econometrics
• Financial Risk Management
• Time Series Analysis

Background:

• Traditional risk spillover measures often struggle with nonlinear dependence structures.
• Assessing financial contagion requires methods that can capture causality and tail co-movement.
• Existing parametric approaches may be susceptible to market irregularities.

Purpose of the Study:

• To develop a novel nonparametric methodology for assessing risk spillovers in a time-series framework.
• To introduce a new measure, nonlinear CoVaR (NCoVaR), for cross-sectional conditional tail co-movement.
• To adapt NCoVaR for contagion analysis, including Granger causality, and develop formal nonparametric tests.

Main Methods:

• Explicit nonparametric measure of cross-sectional conditional tail co-movement (NCoVaR).
• Adaptation of NCoVaR for Granger causality analysis.
• Construction of formal nonparametric tests for independence and Granger non-causality based on U-statistics.
• Empirical illustration using euro area sovereign and banking sector data.

Main Results:

• The proposed NCoVaR captures highly nonlinear dependence structures.
• Nonparametric tests demonstrate superior size and power properties compared to parametric counterparts in simulations.
• The methodology effectively assesses risk transmissions between sovereign and banking sectors.
• New measures are less susceptible to market irregularities than parametric analogues.

Conclusions:

• The developed nonparametric methodology offers a robust framework for analyzing risk spillovers and contagion.
• NCoVaR and associated tests provide a clearer understanding of sovereign-bank risk feedback loops, particularly in volatile markets.
• This approach enhances financial risk management by offering more reliable insights into systemic risk.