Multi-feature evaluation of financial contagion

Area of Science:

• Quantitative Finance
• Econometrics
• Financial Modeling

Background:

• Financial contagion, the spread of market turmoil, is typically assessed via correlation matrices.
• Existing methods often fail to capture the multi-dimensional impact of crises on return distributions.
• Higher moments (skewness, kurtosis) and their dynamic interactions are crucial for understanding contagion.

Purpose of the Study:

• To develop a computationally inexpensive methodology for evaluating financial contagion beyond correlation matrices.
• To extend contagion analysis to the evolution of multiple higher mixed moments of financial returns.
• To provide a more comprehensive understanding of intermarket relationships during crises.

Main Methods:

• Modeling the joint distribution of normalized index returns using polynomials with time-evolving coefficients.
• Estimating coefficients via exponential moving averages for computational efficiency.
• Applying Principal Component Analysis (PCA) for dimensionality reduction of evolving moments.

Main Results:

• A novel approach to capture multi-dimensional contagion dynamics across financial markets.
• Identification of dominant, relatively independent features describing intermarket relations.
• Features provide insights into the current crisis stage and contagion intensity.

Conclusions:

• The proposed method offers a simple and efficient way to enhance financial contagion analysis.
• It captures complex, multi-dimensional interactions in market returns, improving crisis assessment.
• The methodology provides valuable information for understanding and potentially mitigating contagion effects.