engleski

Inferring default cascades in financial systems

In finance, cascading failures of companies are an important issue, they impose danger on all market participants, and can lead to a systemic failure. Therefore, detecting cascades is of vital importance for the market itself. We propose test statistics for inferring such behaviour in a market. In addition to real data of company defaults, we combine two Poisson processes, to simulate the company failure process. The simulations are driven on Erdos-Renyi networks, and for each simulation, a randomized reference model is created. Using the simulated data, we aim to choose the best statistic for different ranges of network and process parameters. For quantifying the difference between statistics we used the Kolmogorov-Smirnov test.

inference on networks ; contact processes ; systemic risk and economic networks

nije evidentirano