FinNet: Financial network analysis in R got straight

Author: Fabio Ashtar Telarico, University of Ljubljana, FDV

Introduction

The numerous packages helping users to deal with network data in R (most notably network and igraph) do not offer tools specific for financial networks.

For years now, authors and analysts have worked on financial data using ad-hoc tools or programming languages other than R. So, the package FinNet was born to provide all R users with the ability to study financial networks with a set of tool especially designed to this purpose. Specifically, FinNet offers both brand new tools and an interface to the almost limitless capabilities of igraph and network.

The first release of FinNet introduces the backbone of this system, which will be expanded with new releases. A basic introduction is available in this readme. For more information, visit: FinNet's webiste

Installation

From CRAN (stable version)

install.packages('FinNet')

From GitHub (development version)

if(!require('remotes'))install.packages('remotes')
remotes::install_github('faTelarico/FinNet')

Motivation

Before FinNet there was not R package especially conceived for financial-network analysis on the CRAN. True, similar results could be achieved combining functions from several packages. However, FinNet allows users (mainly analysts and researchers) to focus on what matters, their analyses, and let the package take care of tedious operations and conversion.

In a nutshell, FinNet is designed to put ease of use first.¹ Moreover, it allows for a very lean installation since the only required packages are included with base R: grDevices and methods. Yet, it is extremely flexible insofar as it allows users to leverage the capabilities of other packages available on CRAN:

• igraph or network to interface FinNet with these packages;
• knitr or pander to print better summaries to the console;
• Matrix to optimise the storage of relations between agents;
• SPB to get better progress bars; and
• yahoofinancer to automatically retrieve data from Yahoo! Finance

Features

At this stage of development, FinNet provides the following:

• Specialised S4 classes for financial agents (firm), their relations (financial_matrix), and interfaces to other packages (network_financial and igraph_financial);
• Functions to register information about financial agents, including retrieving them from Yahoo! Finance;
• Function to encode the relations between these agents and their owners or managers into incidence matrices;
• Function to encode the relations between these agents (based on common ownership, board interlocks, or both) into adjacency matrices;
• Function to encode the relations between these agents (based on common ownership, board interlocks, or both) into extended network or igraph objects.

Comparison with potentially similar R packages

A simple look at the list of packages available on CRAN shows that there are no packages for financial-network analysis.

The most up-to-date book on the subject, Introduction to R for Quantitative Finance (Daróczi et al. 2023, 125-138) uses igraph do analyse the network, but offers no indications as regards data managing and network construction.

Advanced features

With implementation in both C++ and R:

• Functions to identify systematically important agents in the network (see Komárková, Hausenblas, and Frait 2012; Lai and Hu 2021);
• Different measures of centrality (see Nakamoto, Chakraborty and Ikeda 2019);
• Estimation of systemic risk and simulating financial contagion (see Kali and Reyes 2013; and Summer 2013).

Next steps

The upcoming features of the package are:

• Functions to estimate influence and influence-maximisation algorhitms (see Mizruchi and Bunting [1981] (https://doi.org/10.2307/2392519); Khalife, Read and Vazirgiannis 2021);
• Improve support for network/sna;
• Introduce a graphical interface to execute most operation.

Footnotes

1. See full documentation: here. ↩