Every working day some $150 billion flows through Australia’s Interbank system.
Postgraduate student Andrey Sokolov from the University of Melbourne, together with colleagues from Melbourne and Swinburne universities, is analysing the flow of that money to study the dynamics of the overnight loan flows and the stability of the network.
The team is developing dynamic models that test whether the daily flow of funds between Australia’s banks is as robust as it seems – and what might cause its collapse.
Following this money trail as it grows and evolves should help financial regulators to protect our banks better.
Further information:
Dynamics of the Australian Interbank Loan Flows
Andrey Sokolov 1 , Rachel Webster 1, Andrew Melatos 1, and Tien Kieu 1,2
1School of Physics, University of Melbourne, Parkville, VIC 3010, Australia
2Centre for Atom Optics and Ultrafast Spectroscopy, Swinburne University of Technology, Hawthorn, VIC 3122, Australia
Abstract summary:
We analyse the network properties of interbank transactions on the Australian real-time gross settlement system and study the dynamics of the overnight loan flows and the network stability.
Abstract:
I. INTRODUCTION
The crucial effect of the network structure and topology on the emergence of contagion and systemic risk has been recognised and a number of recent studies have been devoted to analysing properties of financial networks in a number of countries (for example, see [1]).
We analyse the network of transactions on the Australian real-time gross settlement system administered by the Reserve Bank of Australia (RBA) for settling large-value transactions between Australian banks. The data comprises all transactions recorded by the Reserve Bank Information and Transfer System (RITS), including those originating from SWIFT, Austraclear and RBA intra-day repos, covering five consecutive days of the week of 19 February 2007.
The distinct feature of the data set is that it includes the source and destination of every transaction (54 banks and RBA). Tracking the counterparties for each transaction yields a rich network structure for the settled transactions. The value of each transaction is also supplied in the data, so that not only the network structure but the cash flows on the network can be examined.
II. DATA ANALYSIS
In this study we apply standard network analysis techniques detailed in [2,3,4] to uncover the properties of the cash flow network. We combine the data for consecutive days and use a matching procedure [5] to discover transactions corresponding to the overnight loans.
The imbalance in a bank’s exchange settlement funds resulting from the daily flow of transactions is to some extent counteracted by the flow of overnight loans on the network. We examine the relation between the cash and loan flows by comparing the cash and loan networks over several days. The evolution of the network properties (degree distribution, clustering, etc) over the five days gives useful constraints on the flow dynamics.
III. MODELING
We use the observed network properties and structure to set up a dynamic model of loan flows on a range of generic financial networks. The aim of the dynamic modeling is to explore the long-term stability of the system that is driven by random imbalances due to trading, on the one hand, and the interbank loans that mitigate the degree of the imbalances, on the other.
We employ a variety of methods including simple analytical models and multi-agent simulations. The latter is especially easy to implement and represents a powerful approach for modeling complex economic and financial systems, since it only requires specification of the microscopic rules that govern agent’s behaviour without imposing global constraints on the dynamics.
The ultimate goal of the project is to develop dynamical models of network evolution and stability that can be used for formulating policy recommendations for financial regulation.
REFERENCES
[1] Iori, G., De Masi, G., Precup, O.V., Gabbi, G., Caldarelli, G, A network analysis of the Italian overnight money market, Journal of Economic Dynamics & Control, 2008, Volume 32, 259–278 [2] Kolaczyk, E.D., Statistical analysis of network data: methods and models, Springer Verlag, 2009 [3] Jackson, M.O., Social and economic networks, Princeton University Press, 2008 [4] Caldarelli, G., Scale-free networks: Complex webs in nature and technology, Oxford University Press, 2007 [5] Ashcraft, A.B. and Duffie, D., Systemic Illiquidity in the Federal Funds Market, American Economic Review, Papers and Proceedings, 2007, Volume 97, 221-225Contact:
Dr Andrew Melatos at amelatos@unimelb.edu.au