M.Sc. C.S. Thesis Defense: Elijah Jesus T. Puaben (Minimal Density Reconstruction and Exposure Limits for Financial Networks)
June 24, 2021
Meeting ID: 810 1374 8664
Meeting Password: EPDefense
Adrian Roy L. Valdez, Ph.D. (Adviser)
Jaymar B Soriano, D.Sc. (Reader)
Elvira P. De Lara-Tuprio, Ph.D. (Chairman)
This work aims to address the underestimation of systemic risk, as measured through a method following the work of Huang et.al., by the minimum density reconstruction as shown by the work of Ramadiah et.al. Two approaches were considered to reach this objective. One considers the use of the 90th and 99th percentiles of the observed systemic risk values across the networks in an ensemble. The other considers imposing exposure limits onto the minimum density reconstruction method. Data from the 2011 stress tests of the European Banking Authority was used to derive a reference network. The first approach had shown to be capable of decreasing the number of cases for which the systemic risk of the reference network is underestimated. However, this result comes with increase in the number of cases where overestimation is observed and the magnitudes of these increase as the considered percentile is increased. The results from the second approach help verify a limitation of the minimum density reconstruction which limits the channels for risk to propagate due to small number of existing edges. The observations found in this work show that minimum density ensembles yield low systemic risk for cases with small market impact and if couple with large initial price shock, can result to consistent underestimation of the risk of the actual network.