Identifying Interbank Loans from Payments Data

The interbank overnight cash market is central to the implementation of monetary policy in Australia. The Reserve Bank of Australia (RBA) has historically gathered information about this market via a survey that provided a summary of participants' daily lending and borrowing. These data were highly aggregated and, until May 2016, were the RBA's only source of quantitative information on this market. This paper develops an innovative algorithm that identifies overnight interbank loans from the millions of payments settled through Australia's high-value payments system. Using this algorithm, we are able to construct a historical loan-level database. By comparing aggregates from our database to the survey data collected by the RBA, we conclude that our algorithm successfully identifies these loans. Between 2005 and 2015, the daily correlations between the summary data and the algorithm output are greater than 90 per cent. The novel features of our algorithm are important; applying existing algorithms to Australia produces correlations below 40 per cent. So these novel features may also be useful for identifying overnight interbank loans in other countries. Using the loan-level database, this paper explores features of the market that were previously unobservable. But the main advantage of this new database is the ability to conduct previously infeasible analyses of this market, such as forthcoming analysis of how the market evolved during the global financial crisis.