Happy New Year Everyone. The last half of the previous year was a bit quieter for this blog, mainly because I was busy on my grand project: the Financial Network Analytics (FNA) web service and enterprise software. The project is getting closer to an official release and FNA now has its own website at www.fna.fi. You can sign up there and try it online.

The second big change is that this blog has moved to www.fna.fi/blog/. It will maintain its original purpose of being a focal point for research and news on network analysis and agent-based models of financial systems. The complete three-year history of the blog will be available in the new address as well. Please update your bookmarks and sign up to the e-mail alerts if you have not done so yet. Alternatively, you can also follow the blog and other updates on Twitter and Facebook.

Thirdly, the Research Database has moved to www.fna.fi/library/. It now has a much better user interface and it is easier to browse and find articles. Perhaps importantly to some, you can also get bibtex references of each article. The library has some 110 articles and I am working on an update. Please let me know if yours is missing.

After a busy summer we are proud to release the first online version of the FNA (Financial Network Analytics) platform at www.fna.fi. The site is currently in ‘private alpha’ stage. This means that all main functionality is available but it may not be completely polished yet.

FNA is an application for exploring, analyzing and visualizing financial networks – in a way that much of the research featured in this blog has shown is useful and important. Below is a short “Getting started” video tutorial. In the video I create networks from link data, do centrality analysis of the network and visualize the results.

FNA is intended as a tool for financial regulators and overseers to analyze regulatory data with the tools of network analysis and simulations – and to explore the financial system through visualisations. In addition to the web version, FNA can be installed on company intranets and as a desktop software.

The tool is probably also interesting for researchers in the area. Our plan is to directly connect more data to the online platform so that the cumbersome steps of data cleaning and formatting can be bypassed. Currently a full time series of BIS statistics on banking sector claims by country is included (Table 9D). Better availability of public data should lower the entry barriers to this type of financial research. Also, when both research and policy analysis are available on the same platform, new research models can be made available for policy makers for empirical testing in a resource efficient way.

As for developers, FNA has a modular structure and allows new functionality and models to be easily added. Currently we have plugins for network analysis, basic charts, network layouts, data cleaning tools, a payment system simulator and an algorithm for identifying overnight and term loans from payment data. You can read more about these in earlier posts (1, 2). If you are a researcher or developer, you can have your models run from FNA via our API. You can use all the graphical user interface, data visualization and graph database functionality of FNA in your code. Please get in touch if you would like to know more.

Go to www.fna.fi

For the last couple of months we have been working hard on the new version of FNA. FNA is a platform for analyzing, exploring and visualizing financial networks (download FNA Info sheet).

Working with large (often messy) datasets is hard, so our focus has been to make it as easy and intuitive as possible. Thus FNA has features starting from cleaning and validating data, calculating network metrics to eventually visualizing and modeling the analyzed system.

We now have a ‘functioning prototype’ almost finished. We are now ironing out a few details and it will be available online in about a month for everyone to try it out. Meanwhile the following screenshots provide a preview.

FNA is based on web technologies and runs on your browser. The server where the commands execute can be installed on your local desktop, on the company intranet or you can use the web version running on our servers.

In addition to the command line interface (Picture 1) already available in the previous version, you can now also give commands with mouse by point-and-click (Picture 2). The command line interface itself is improved with an ‘auto-suggest’ function. As working with different files is often a slowing factor in analysis, the new FNA integrates file management into the user interface (panels on the right).

Picture 1 : Command line user interface

Picture 2: Point-and-click user interface

A major new feature set are visualizations. First of all you can several different network layouts. Currently we have a force-directed network layout (Picture 3), arc layout and a geographic overlay of networks available (Picture 4). As these are based on JavaScript templates (the ones below use Protovis) it will be easy for us to add new ones on the fly. If you have more than one network in memory, a similar chart is automatically created for all networks allowing you to browse the results quickly visually. The layouts are interactive allowing you to select, zoom and get more details on the network elements. You can also export them keeping interactivity.

Picture 3: Force-directed network layout

Picture 4: Geographic network layout

There are also many more familiar chart types (line, pie, bar, scatter plot, etc) available which can show statistics from any arc, vertex or network properties.

If you would like to learn more about FNA or participate in developing and testing it, please get in touch with us.

Here are a few papers that were recently added to the database.

A network analysis of global banking:1978-2009 (April 2011) by Camelia Minoiu and Javier A. Reyes carries out an empirical analysis on the structure and statistical properties of the global banking network. They use cross-border bank lending data for 184 countries over 1978–2009. They find that the network has several structural breaks in network indicators that relate to several waves of capital flows. Connectivity of the network tends to fall during and after systemic banking crises and sovereign debt crises. The 2008–09 global financial crisis stands out as an unusually large perturbation to the cross-border banking network.

Network view of cross-border banking in 2007 (Minoiu and Reyes 2011)

Measuring the systemic importance of interconnected banks (March 2011) by Mathias Drehmann and Nikola Tarashev develops a measure of systemic importance based on how banks propagate shocks across the banking system and are vulnerable to propagated shocks. They find substantial differences between alternative measures, which implies that prudential authorities should be careful in choosing the underlying approach.

Macro-prudential Regulation from the Perspective of the Financial Network: Empirical Analysis Based on the Data from Payment and Settlement System in China (April 2011) by Yandong Jia analyzes the properties of the Chinese banking network and proposes methods for macro-prudential supervision based on network analysis.

Complex stock trading network among investors (March 2010) by Zhi-Qiang Jiang and Wei-Xing Zhou analyze the statistical properties of stock trading networks based on order flow data of a highly liquid stock (Shenzhen Development Bank) listed on Shenzhen Stock Exchange during the whole year of 2003.

Financial Connections and Systemic Risk (July 2010) by Franklin Allen, Ana Babus and Elena Carletti develops a model where institutions form connections through swaps of projects in order to diversify their individual risk. They find that network structure matters when short-term financing is used. If expectations on payback are low, systemic consequences are created by banks decisions not to roll over the short term debt.

Quite a few conferences and workshops on financial networks are being held in June this year. Here are the ones with open participation.

First, there is NetSci2011 in Budapest on 6-10 June. It hosts two interesting “Satellite events”, a business network research conference Circuits of Profit on 6 June and Network of Networks which focuses on financial networks and systemic risk on 7 June.

On 10 June the Geneva Finance Research Institute (GFRI) hosts its first Financial Networks Conference. The agenda looks very promising.

On 20-24 ETH Zurich hosts the International Workshop on Coping with Crises in Complex Socio-Economic Systems. The conference brings together researchers interested in networks from complex systems, transportation, epidemiology or finance and economics perspectives.

The month ends with the University of Essex organized workshop on Markets and Networks on 30 June – 1 July.

Last but not least, the NET institute has published a Call for proposals on Summer Grants for scientific research projects in the areas of network industries.

Good luck for the applicants and looking forward to seeing you in the above workshop and conferences.

Its been three months since the last “Research update” and a number of interesting papers have been published meanwhile. Please let me know if you have one to add.

An agent-based model of payment systems by Marco Galbiati and Kimmo Soramäki (Journal of Economic Dynamics and Control) builds an agent based model where banks choose the amount of liquidity to settle a given flow of payments. The paper pays special attention to a realistic settlement process with complex dynamics and studies the equilibrium level of liquidity that is a result of the game between the banks. The paper investigates liquidity usage with various system sizes and volumes, and under different liquidity cost parameters.

The Brazilian Interbank Network Structure and Systemic Risk by  Edson Bastos e Santos and Rama Cont (Banco Central do Brasil Working Paper 219) uses unique data on bilateral exposures between financial institutions in Brazil. They explore the empirical properties of the network and relate them to capital structure of the banks. They find that the network is best modeled as a directed scale-free (weighted) graph with heavy tailed degree and weight distributions.

Network Structure and Systemic Risk in Banking Systems by Rama Cont, Amal Moussa and Edson Bastos e Santos studies the potential for default contagion using the same data set as above. They find that network-based measures of connectivity and concentration of exposures contribute to the systemic importance of an institution.  Imposing capital measures on the basis of the network based “Contagion index” allows a lower level of capital across the system for given level of protection than imposing them uniformly across all banks.

Norwegian overnight interbank interest rates by Akram, Q. Farooq and Casper Christophersen (Norges Bank Staff Memo) infers actual interest rates from the advances and returns settled in the Norwegian payment system and proposes a new overnight interest rate based upon them (NONIA). The paper investigates the properties and evaluates the reliability of the interest rate both under normal conditions and during the recent financial crisis.

Mapping Systemic Risk in the International Banking Network by Rod Garrett, Lavan Mahadeva and Katya Svirydenska (Bank of England Working Paper) partitions banking groups from 21 countries into modules that reflect the ease at which stress is transmitted between the banks within a cluster. They use BIS data on international banking group exposures and find substantial changes over the time period of 1985 to 2009. The image below shows the network in Q3 of 2009.

Monitoring the unsecured interbank money market using TARGET2 data by Ronald Heijmans, Richard Heuver and Daniëlle Walraven (DNB Working Papers, No 276) extends the algorithm by Furfine (1999) and identifies loans between Dutch banks up to a maturity of one year. Based on the new algorithm they develop a policy tool to monitor the interbank money market, both at the level of the market and at individual banks. (Note: the match -command in FNA provides a similar algorithm as the one used in their paper)

NetSci2011 will take place in Budapest on 6-10 June 2011 bringing together researchers, practitioners, and teachers in network science across disciplines. The conferece will host a number of shorter satellite events including the business network research conference Circuits of Profit and NetONets focusing on Systemic Risk and Infrastructure Interdependencies.

De Nederlandsche Bank and is organizing a conference on ‘Complex systems: Towards a better understanding of financial stability and crises‘ on 3-4 November 2011 in Amsterdam. The conference’s aim is to “bring together researchers and policymakers to deepen our understanding of macroeconomic and financial systems by using methods for complex systems”. The call for papers is open until 17 June.

We’ll try to be in both.

As many of you know, Financial Network Analyzer is the program that I have been developing for the past year and a half. Its currently mainly used by central banks for the analysis of financial microdata and for the simulation of payment systems.

After major rewrite of much of the application and thorough testing, FNA 1.3 is finally available. Its still in “beta” but most of it should work quite nicely already. Let me know if you try it and encounter any trouble. The program has three major components: financial network analysis, tools for working with large CSV based datasets, and a payment systems simulator.

The most important improvement, however, is the Manual with plenty of examples for each command that should enable users to quickly jump into using the program. The program now also has its own website and blog at financialnetworkanalyzer.com. The program’s blog will be the main source of updates on the program while this blog will continue to focus on financial network research.

A recent article in FNA blog looks into the visualization of financial networks. Visualisations will be the major update for the next version of FNA for which work has already started. The other major change is that version 2.0 will move FNA into a client-server architecture allowing a more scalable environment for large-scale modeling and simulations. You can browse the specifications for more details.

The Geneva Finance Research Institute (GFRI) is organising its first “Financial Networks” research conference on 10 June 2011 in Geneva. Their intention is to make it an annual event. The program is impressive with papers covering trading networks, networks of capital flows and network analysis on systemic risk and contagion. Registrations should take place before 27 May 2011.

I get this question asked every once in a while and here are some thoughts, personal experiences and little analysis of the Financial Networks Research Database.

Looking at the database (n=78), most articles were published as working papers or equivalent, a majority of which in central bank series. This is natural as central banks have access to detailed bilateral data required for any empirical analysis. As the surge in research on financial networks is rather new and the publication process time-consuming, some of them will likely end up in peer-reviewed journals. About a quarter of the papers in the database had already been published. The remaining ones appeared in Financial Stability Reviews or other reports or were still in the publication process (and were classified as mimeos). Of course, we need to keep in mind that the database consists likely a biased sample of all relevant research.

The most popular was the Journal of Economic Dynamics and Control with four papers including one by Marco Galbiati and me which took about two years to go through. Other finance/economics publications included various field journals such as Journal of Finance, Financial Markets, Institutions, and Instruments, Journal of Banking & Finance, and Journal of Money, Credit and Banking. For work related to central bank interests, the International Journal of Central Banking has become well appreciated even though it was launched only in 2008.

The study of networks is an interdisciplinary subject and quite a few papers were featured in physics journals such as Physica A, J. Statistical Mechanics or The European Physical Journal B. In practice the choice of publication venue is also related to what is appreciated by your peer group and to which groups you might be interested in reaching with your research. I myself have a few articles in Physica A. The journal has published a lot of theoretical work on the dynamics and statistical properties of complex networks and we thought they might be interested in our empirical work on network of payment flows between US banks.

Sociology and social network analysis – the oldest field in the study of networks – do not have any papers in the database. The new journal Social Network Analysis and Mining might fill this void. According to the Introduction to the first issue, network analysis in all disciplines is welcome and within the journal’s scope.

Happy to hear any comments.