Measuring Interconnectedness Between Financial Institutions with Bayesian Time-Varying Vector Autoregressions

Abstract

We propose a market-based framework that exploits time-varying parameter vector autoregressions to estimate the dynamic network of financial spillover effects. We apply it to financials in the Standard & Poor's 500 index and estimate interconnectedness at the sector and institution level. At the sector level, we uncover two main events in terms of interconnectedness: the Long Term Capital Management crisis and the 2008 financial crisis. After these crisis events, we find a gradual decrease in interconnectedness, not observable using the classical rolling window approach. At the institution level, our framework delivers more stable interconnectedness rankings over time than other market-based measures of systemic risk. Keywords

Refined Measures of Dynamic Connectedness based on Time-Varying Parameter Vector Autoregressions

“…The Diebold and Yılmaz (2009, 2012 VAR-based connectedness approach has already attracted significant attention by the existing economic and financial literature, and has been applied to issues relating to the stock market interdependencies, volatility spillovers, business cycle spillovers, as well as bond yield spillovers (see, inter alia, Alter and Beyer 2014; Antonakakis 2012; Awartani and Maghyereh 2013;Bekaert et al 2014;Bubák et al 2011;Diebold and Yilmaz 2015;McMillan and Speight 2010). At the same time, alternative measures of connectedness have been provided by Baruník et al (2016Baruník et al ( , 2017, Baruník and Křehlík (2018), and Geraci and Gnabo (2018).…”

Fragmentation, integration and macroprudential surveillance of the US financial industry: Insights from network science

“…To represent the interdependencies at stake in the financial system, we follow [ 6 ] who propose a framework based on time-varying parameter vector autoregressions, as in [ 7 , 10 ] to recover a network of financial spillovers—or causality-based network—that is entirely dynamic. In their framework, financial institutions represent nodes in a directed network.…”

“…Finally, the errors, u t , are assumed to be normally distributed, with mean zero and variance-covariance matrix Σ (see Section A in S1 Appendix of the Supporting Information). The original model of [ 6 ] allows for heteroskedasticity and fat-tailed errors. Here, however, we adopt a simpler approach and standardize (so to have unit variance) the returns, r 1 t , …, r Nt , in a previous step by using a GARCH(1, 1) model to estimate the time-varying volatility.…”

“…Following [ 6 ], the linkages between financial institutions are retrieved from stock market prices by means of Bayesian Time-Varying VAR framework in the vein of the one developed by [ 7 ] for macroeconomic data. One of the main advantages of this methodology is that it allows to generate temporal networks in a widely spaced period of 20 years, where connections are evolving gradually through time.…”

“…One of the main caveats of this generation of studies lies in the way the temporal nature of financial networks was dealt with as they did rely on successive snapshots of the industry at different time periods by means of rolling windows regressions. Recently, there have been a few studies among which [ 6 , 8 ] some have developed methodological approaches specifically designed to better tackle the time varying dimension of the financial system and infer fully dynamic networks. However, these contributions do not exploit in full the set of tools offered by network science to document salient patterns, relying on degree centrality measures.…”