Assignment COMP0046 – Networks and systemic risk
Data
Please download the data in the folder “Assignment data” from moodle and unzip it. The data set contains a matrix of interbank exposures. The entry (i,j) represents the amount lent by j to i, a vector of external (non interbank) assets, and a vector of equities for a group of banks.
Tasks
1. You should present a statistical characterization of the system for what concern the distributional properties of balance sheets and interbank claims. You should generate the network corresponding to the pattern of interbank loans, characterize its properties (e.g. degree distribution, clustering, assortativity), discuss how it is different or similar to an Erdös-Rényi network, and how you do expect the network properties to affect the propagation of distress within the system, when this is measured with the Furfine algorithm.
2. You will then perform stress tests using the Furfine algorithm (you are also welcome to use contagion algorithms other than the Furfine if you prefer), analyze and discuss their results.
3. You will then consider that part of the external assets of each bank is invested in a common asset (you can consider a more complex structure of overlapping portfolios if you prefer), you will perform stress tests where shocks simultaneously propagate because of counterparty risk and overlapping portfolios, and you will compare these results with those of point 2.
You are free to explore scenarios of your choice for what concern the initial exogenous shock, and to make assumptions for what concern the recovery rate and the liquidity of external assets, as long as all assumptions and scenarios are clearly stated and justified in your written report.
Written report
A brief written report (indicatively around 2500 words plus figures and tables) containing the justification of the approach, the presentation of the results, the discussion of the results and conclusions should be submitted to moodle before the deadline
Marking
This assignment is worth 100% of the total exam mark. The marking will be based on the following criteria:
1) Clarity of the report (is the report clear and well structured, are figures informative, is the methodology explained well? Are modeling assumptions discussed?)
2) Results (are results sound?)
3) Critical discussion (are results correctly interpreted? Is there a discussion of
limitations and further challenges?)
The reports will be marked by academic staff.