Dynamic Networks in Large Financial and Economic Systems

Abstract

This paper identifies frequency-dependent network structures that evolve over time. To measure such dynamic networks, we propose a computationally efficient framework that is widely applicable to many economic and financial datasets, and readily available for high dimensional models. We provide Monte Carlo evidence that our measures are able to reliably recover true network connections from a battery of DGPs and also develop a testing procedure for statistical differences among frequency-dependent network connections. Our empirical application on firm-level realized volatilities documents substantial heterogeneities in dynamic network structures that may be useful as an online monitoring tool to help guide macro-prudential policy.

Keywords: Dynamic networks, Network connections, Time-varying parameter VAR, Realized Volatility, Connectedness, Spillovers.

JEL Classification: C10, C40, C55, C58, G00