Director of CSU’s Centre for Research in Complex Systems (CRiCS), Professor Terry Bossomaier, believes information flow is an important characteristic of complex systems such as the environment, organisations, the economy, and climate.
A team that included Professor Bossomaier and researchers from the University of Sussex in the UK developed a method of estimating the accuracy of the predictions of these flows, which is found in a paper in the latest issue of the prestigious international journal, Physical Review Letters.
"In social systems, information flows might be specific and easy to chart and explain. In other systems it is subject to more subtle influences," Professor Bossomaier said.
"The study of these flows gained great attention over the years. Clive Granger won the 2003 Nobel Prize for Economics for his investigation into the Granger Causality, a method of forecasting how one macro-economic quantity influences another. In using Granger Causality, we don’t make assumptions about what information is being transferred, but just look at how the behaviour is statistically linked over time."
Professor Bossomaier notes, however, that Granger Causality has one big problem.
"It only works for a simple model of behaviour called a linear system," he said.
However in the past 10 years, a new measure has been developed to apply to more realistic non-linear systems, called transfer entropy, which has been very difficult to calculate.
Dr Lionel Barnett from the University of Sussex and Professor Bossomaier have now developed a new approach to address this problem.
"We have developed a method of measuring transfer entropy which is both more accurate and provides an immediate measure of its accuracy," Professor Bossomaier said.
"Armed with this new tool, we are now studying information flows in business cycles and stock markets, as we want to investigate how big changes in economic systems such as financial crashes come about."