A novel banking supervision method using the minimum dominating set

The financial crisis of 2008 demonstrated that effective and continuous supervision of economic systems is necessary. More specifically, close monitoring of banking institutions and their interrelations is of outmost importance in the effort to reduce and control systemic risk in the specific industry and the economy as a whole. This interconnectedness can be described and examined through Complex Network models. In this context, every node corresponds to one bank, every link describes the interconnections between two nodes and the  weight of every link shows the correlation level between the connected nodes. In this work, we propose the use of Complex Networks for the monitoring of the whole banking system through a smaller subset of it, the Minimum Dominating Set (MDS). Our approach has two steps: a) eliminate every link with a weight (correlation) below a threshold and b) indentify the MDS in the updated network. We show that the MDS nodes can be used as "gauges" for the monitoring of the whole network. In this paper we apply the MDS methodology on the 200 largest US banks, in terms of their total assets. We use interbank loans from the balance sheets and more specifically loans to depository institutions and acceptances of other banks,  to calculate the respective cross-correlation that reveals their interconnectedness. Using that method we can effectively monitor the whole 200 bank system through just the institutions of the MDS and the isolated ones. The proposed method can be proven useful to  Central Banks as a warning system that can swiftly identify incidents of bank distress and their paths of contagion for prompt intervention and regulation.