Today’s issue of Nature has an interesting article by Andrew G. Haldane and Robert M. May in which the “network lens” (as they call it) is applied to financial risks and instability in the world of banking. The network, in this case, consists of banks lending and borrowing from and to one another. The authors outline a variety of models that show how financial shocks can propagate through the network, leading to systemic failure such as the financial crisis of the past years. One of the more intriguing insights is that as banks grow more homogeneous in terms of their activities, they reduce their individual risk of failure but increase the instability of the system as a whole.
In practice, the networks in these models are all assumed to be Erdös-Rényi random networks, so one may wonder whether this approach deserves the name “network lens,” as this amounts to random, unstructured interaction. However, the authors do remark that empirical banking networks actually have fat tail-degree distributions and are dissortative, and that this makes things worse. All in all an interesting read, and of course, it’s nice to see this kind of work appearing in Nature. Also interesting is the fact that the second author appears to be a zoologist….