Title: High dimensional influences and financial contagion
Authors: Massimiliano Caporin - University of Padova (Italy) [presenting]
Deniz Erdemlioglu - IESEG School of Management (France)
Stefano Nasini - IESEG School of Management (France)
Abstract: The availability of high frequency information over cross-sectional dimensions, along with the disposal of important indicators (such as volatility, jumps and trading activity) constitutes a precious source of complex information, mirroring intrinsically interconnected aspects of the financial system. We rely on a novel network-based statistical approach to capture financial contagion from these integrated and complex data structures. The model relies upon influences across a large panel of equities, accounting for three distinct forms of interdependence: (1) the lagged interdependence, (2) the interdependence among different variables measured from high frequency data, and (3) the interdependence between industry-based equity clusters. The flexibility of our approach allows us to recover the pairwise influence structure, based on a specialized estimation approach for high-dimensional parameter spaces. Given the estimated influence matrix from the market data, we study interdependence dynamics among companies and perform additional economic analyses.