How did we go from a strong housing market and the low interest rates to $200 billion total loss in financial institutions? Why is it that “bubbles” repeatedly form in the economy and inevitably burst?

Image taken from CagleCartoons.com 

Whether the system you are looking at is a financial network of banks or an ecosystem of organisms, there are similarities within network structure that can be used to predict a large disruption which moves the system from one state to another. In the recently published article “Ecology for Bankers” (directions for accessing full text below) researchers analogize financial systems to ecosystems. Because of certain feedback mechanisms and underlying linkages within a system, changes arise from the activity of a select few. This can even be seen even in such issues as global climate change, electrical systems or the Internet. The authors point out that there is a relatively small amount of study on systemic risk (risk of stock market crash) versus conventional risk (risk of day to day trades) even though the consequences of systemic risk overwhelmingly outweigh any conventional risk. Looking at the network structure of ecosystems, the authors found ‘large’ nodes that had a disproportionately large number of connections with ‘small nodes’. The small nodes similarly had few connections with large nodes. Network robustness has often been attributed to the degree in which a system can be broken down into discrete systems – or modularity. This modularity was seen in the ecosystem, though sometimes it is a tradeoff between local and systemic risk. In a network in which there are strong connections between all nodes, it is near impossible to stop a disruption such as an epidemic or a forest fire. In a network which there is too much separation, especially in the financial system, it could lead to fragmentation and also increase systemic risk.

Figure 1. Graph of the $1.2 trillion dollars in day to day transactions between thousands of banks.

The authors analyzed the topology of interbank payments using a settlement system run by the Federal Reserve System. Gathering data from 9500 banks and 700,000 transfers, the edges between the nodes were classified as strong if there were many transfers. Surprisingly, 75% of all the payment flows go through only 0.1% of the nodes and 0.3% of the edges. Again, large banks were connected disproportionately to the small banks. This would signal that there is large stability in the system because it shows modularity. The researchers pointed out that there is also the factor of ‘contagion dynamics’ in which people’s perceptions affect the system such as the overvaluation of internet stocks and the onset of panic behavior. ‘Contagion dynamics’ of public perceptions and asset valuation are an overlooked part in researching systemic risk and may be tied into the underlying risk in the network structure.

Figure 2. The core of the Figure 1 in which 75% of the value of transfers happens between 66 banks with 25 banks being completely connected. Both miniscule percentages when compared to the total number of institutions.

This research relates directly to the concepts of network structure that we covered earlier in the class. However it brings a new meaning to Granovetter’s strength of weak ties. While weak ties are not only useful in getting jobs, they are also important in the stabilization networks from catastrophic changes. Strong ties, while they may alleviate local risk, often are the cause of widespread systemic risk. The weak ties that form are often the local bridges between smaller modular networks within a larger one. Again, it is this modularity that builds a robust system. It is interesting to consider the large effect of public perception which has yet to be studied in depth. The paper was not clear in how this ‘contagion dynamic’ can be extrapolated onto other applications such as ecosystems. Author Malcolm Gladwell describes in The Tipping Point drastic changes in a different light focusing on both positive and negative tipping points. He attributes social epidemics as facilitated by the Law of the Few and the Stickiness Factor (which together create several tightly connected large nodes), and the Power of Context (which ties in unpredictable environmental effects). I believe more of this topic will be covered in class once we go into Information Cascades, Network Effects and the Diffusion of Innovations. More research definitely needs to be done with systemic risk range because of the drastic consequences from the environment to the stock market. Whether it is the next housing market credit crunch, collapse of aquatic ecosystems due to overfishing, or California sinking due to global warming, we often hold a false sense of security when we view stability. In the future, we can use network structure to identify the characteristics and predict early instability and correct for it before we reach a tipping point.

Sources

Complex Systems: Ecology for Bankers

Robert M. May, Simon A. Levin and George Sugihara

Nature Vol 451|21 February 2008

http://www.nature.com/nature/journal/v451/n7181/full/451893a.html (abstract)

To access the Full Text as a Cornell affiliated student or faculty

1. http://erms.library.cornell.edu/search/tnature/tnature/1%2C30%2C32%2CB/frameset&FF=tnature&1%2C1%2C

2. Click “Nature Journals Online”

3. Log in using NetID and Password

4. Search for “Ecology for Bankers”