A Matrix-Variate t Model for Networks

Monica Billio; Roberto Casarin; Michele Costola; Matteo Iacopini

doi:10.3389/frai.2021.674166

A Matrix-Variate t Model for Networks

Monica Billio^*, Roberto Casarin, Michele Costola, Matteo Iacopini

^*Corresponding author for this work

Research output: Contribution to Journal › Article › Academic › peer-review

Abstract

Networks represent a useful tool to describe relationships among financial firms and network analysis has been extensively used in recent years to study financial connectedness. An aspect, which is often neglected, is that network observations come with errors from different sources, such as estimation and measurement errors, thus a proper statistical treatment of the data is needed before network analysis can be performed. We show that node centrality measures can be heavily affected by random errors and propose a flexible model based on the matrix-variate t distribution and a Bayesian inference procedure to de-noise the data. We provide an application to a network among European financial institutions.

Original language	English
Article number	674166
Pages (from-to)	1-7
Number of pages	7
Journal	Frontiers in Artificial Intelligence
Volume	4
Issue number	May
DOIs	https://doi.org/10.3389/frai.2021.674166
Publication status	Published - May 2021

Bibliographical note

Funding Information:
We thank Maurizio La Mastra for excellent research assistance. This research used the SCSCF and the HPC multiprocessor cluster systems provided by the Venice Centre for Risk Analytics (VERA) at the University Ca’ Foscari of Venice. MB, RC, and MC acknowledge financial support from the Italian Ministry MIUR under the PRIN project Hi-Di NET – Econometric Analysis of High Dimensional Models with Network Structures in Macroeconomics and Finance (grant agreement no. 2017TA7TYC). MI acknowledges financial support from the Marie Skłodowska-Curie Actions, European

Funding Information:
We thank Maurizio La Mastra for excellent research assistance. This research used the SCSCF and the HPC multiprocessor cluster systems provided by the Venice Centre for Risk Analytics (VERA) at the University Ca' Foscari of Venice. MB, RC, and MC acknowledge financial support from the Italian Ministry MIUR under the PRIN project Hi-Di NET ? Econometric Analysis of High Dimensional Models with Network Structures in Macroeconomics and Finance (grant agreement no. 2017TA7TYC). MI acknowledges financial support from the Marie Sk?odowska-Curie Actions, European Union, Seventh Framework Program HORIZON 2020 under REA grant agreement no. 887220.

Publisher Copyright:
© Copyright © 2021 Billio, Casarin, Costola and Iacopini.

Keywords

62F15
62M10
65C05
Bayesian
C11
C32
C58
financial markets
matrix-variate distributions
networks
t distribution

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.3389/frai.2021.674166

Handle.net

Persistent link

Cite this

@article{ae7ae048a1d24c0baabef0c6cba0b93d,

title = "A Matrix-Variate t Model for Networks",

abstract = "Networks represent a useful tool to describe relationships among financial firms and network analysis has been extensively used in recent years to study financial connectedness. An aspect, which is often neglected, is that network observations come with errors from different sources, such as estimation and measurement errors, thus a proper statistical treatment of the data is needed before network analysis can be performed. We show that node centrality measures can be heavily affected by random errors and propose a flexible model based on the matrix-variate t distribution and a Bayesian inference procedure to de-noise the data. We provide an application to a network among European financial institutions.",

keywords = "62F15, 62M10, 65C05, Bayesian, C11, C32, C58, financial markets, matrix-variate distributions, networks, t distribution",

author = "Monica Billio and Roberto Casarin and Michele Costola and Matteo Iacopini",

note = "Funding Information: We thank Maurizio La Mastra for excellent research assistance. This research used the SCSCF and the HPC multiprocessor cluster systems provided by the Venice Centre for Risk Analytics (VERA) at the University Ca{\textquoteright} Foscari of Venice. MB, RC, and MC acknowledge financial support from the Italian Ministry MIUR under the PRIN project Hi-Di NET – Econometric Analysis of High Dimensional Models with Network Structures in Macroeconomics and Finance (grant agreement no. 2017TA7TYC). MI acknowledges financial support from the Marie Sk{\l}odowska-Curie Actions, European Funding Information: We thank Maurizio La Mastra for excellent research assistance. This research used the SCSCF and the HPC multiprocessor cluster systems provided by the Venice Centre for Risk Analytics (VERA) at the University Ca' Foscari of Venice. MB, RC, and MC acknowledge financial support from the Italian Ministry MIUR under the PRIN project Hi-Di NET ? Econometric Analysis of High Dimensional Models with Network Structures in Macroeconomics and Finance (grant agreement no. 2017TA7TYC). MI acknowledges financial support from the Marie Sk?odowska-Curie Actions, European Union, Seventh Framework Program HORIZON 2020 under REA grant agreement no. 887220. Publisher Copyright: {\textcopyright} Copyright {\textcopyright} 2021 Billio, Casarin, Costola and Iacopini.",

year = "2021",

month = may,

doi = "10.3389/frai.2021.674166",

language = "English",

volume = "4",

pages = "1--7",

journal = "Frontiers in Artificial Intelligence",

issn = "2624-8212",

publisher = "Frontiers Media",

number = "May",

}

TY - JOUR

T1 - A Matrix-Variate t Model for Networks

AU - Billio, Monica

AU - Casarin, Roberto

AU - Costola, Michele

AU - Iacopini, Matteo

N1 - Funding Information: We thank Maurizio La Mastra for excellent research assistance. This research used the SCSCF and the HPC multiprocessor cluster systems provided by the Venice Centre for Risk Analytics (VERA) at the University Ca’ Foscari of Venice. MB, RC, and MC acknowledge financial support from the Italian Ministry MIUR under the PRIN project Hi-Di NET – Econometric Analysis of High Dimensional Models with Network Structures in Macroeconomics and Finance (grant agreement no. 2017TA7TYC). MI acknowledges financial support from the Marie Skłodowska-Curie Actions, European Funding Information: We thank Maurizio La Mastra for excellent research assistance. This research used the SCSCF and the HPC multiprocessor cluster systems provided by the Venice Centre for Risk Analytics (VERA) at the University Ca' Foscari of Venice. MB, RC, and MC acknowledge financial support from the Italian Ministry MIUR under the PRIN project Hi-Di NET ? Econometric Analysis of High Dimensional Models with Network Structures in Macroeconomics and Finance (grant agreement no. 2017TA7TYC). MI acknowledges financial support from the Marie Sk?odowska-Curie Actions, European Union, Seventh Framework Program HORIZON 2020 under REA grant agreement no. 887220. Publisher Copyright: © Copyright © 2021 Billio, Casarin, Costola and Iacopini.

PY - 2021/5

Y1 - 2021/5

N2 - Networks represent a useful tool to describe relationships among financial firms and network analysis has been extensively used in recent years to study financial connectedness. An aspect, which is often neglected, is that network observations come with errors from different sources, such as estimation and measurement errors, thus a proper statistical treatment of the data is needed before network analysis can be performed. We show that node centrality measures can be heavily affected by random errors and propose a flexible model based on the matrix-variate t distribution and a Bayesian inference procedure to de-noise the data. We provide an application to a network among European financial institutions.

AB - Networks represent a useful tool to describe relationships among financial firms and network analysis has been extensively used in recent years to study financial connectedness. An aspect, which is often neglected, is that network observations come with errors from different sources, such as estimation and measurement errors, thus a proper statistical treatment of the data is needed before network analysis can be performed. We show that node centrality measures can be heavily affected by random errors and propose a flexible model based on the matrix-variate t distribution and a Bayesian inference procedure to de-noise the data. We provide an application to a network among European financial institutions.

KW - 62F15

KW - 62M10

KW - 65C05

KW - Bayesian

KW - C11

KW - C32

KW - C58

KW - financial markets

KW - matrix-variate distributions

KW - networks

KW - t distribution

UR - http://www.scopus.com/inward/record.url?scp=85117919835&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85117919835&partnerID=8YFLogxK

U2 - 10.3389/frai.2021.674166

DO - 10.3389/frai.2021.674166

M3 - Article

AN - SCOPUS:85117919835

SN - 2624-8212

VL - 4

SP - 1

EP - 7

JO - Frontiers in Artificial Intelligence

JF - Frontiers in Artificial Intelligence

IS - May

M1 - 674166

ER -

A Matrix-Variate t Model for Networks

Abstract

Bibliographical note

Keywords

UN SDGs

Access to Document

Handle.net

Other files and links

Fingerprint

Cite this