Publication at Faculty of Mathematics and Physics |

2022

Graph theoretical techniques are utilized to examine the centrality structure of overnight index swap (OIS) networks. Correlation based graphs are constructed to encode pairwise relationships between distinct OIS rates.

Multiple notions of graph centrality are considered, and the time evolution of these measures is studied. A principal component analysis based centrality measure is constructed to examine comovements between full OIS curves.

Numerical examples demonstrating these ideas are provided.