Library Header Image
LSE Theses Online London School of Economics web site

Fire sales and policy interventions in financial networks

Pang, Raymond Ka-Kay (2023) Fire sales and policy interventions in financial networks. PhD thesis, London School of Economics and Political Science.

[img] Text - Submitted Version
Download (3MB)
Identification Number: 10.21953/lse.00004538


In this thesis, we model the impact of fire sales and the mitigation of systemic losses from policy interventions using tools from the theory of financial networks and complex systems. A fire sales event occurs when banks sell a large quantity of assets at discounted prices. There is a mark-to-market adjustment on assets sold, leading to a depreciation in the asset value and fuelling further fire sales. This channel of systemic risk can be a large contributor to losses, as observed in systemic events like the Great Financial Crisis. The impact of these events spurred a range of policy interventions to enhance financial stability. We use network models to analyse fire sales and policy interventions in interconnected systems. Our first main result shows that under a partial information setting, policy interventions for mitigating fire sales losses using matrix reconstruction methods outperform policy interventions that do not account for institutions overlapping portfolios. We focus on optimising policy interventions when only the partial information is known, and how this compares with the fully observed data. Using matrix reconstruction methods, we find policy interventions under partial information can be similar to policy interventions under fully observed data. The similarity in performance under partial information highly depends on the chosen matrix reconstruction method. The second main result is developing a new reverse stress test approach for a multi-stage fire sales event. Under this new approach, we find losses under these derived scenarios are larger than benchmark scenarios used in other stress tests. A reverse stress testing approach is taken, as the scenario reflects the largest losses of banks from the input data and given fire sales mechanism. We find the losses from these derived scenarios are large and have not been observed in previous studies. Our third main result develops a clearing model for banks that post collateral as part of their financial obligations, where we account for two distinctive channels of fire sales. In this clearing situation, we consider the counterparty losses between banks, fire sales losses from assets used as collateral and fire sales losses from externally held assets. In this new collateral model, the inclusion of external asset holdings towards fire sales losses in a clearing situation can result in larger losses, compared to banks holding no external assets. We find the total losses depend on the overlap between both fire sale channels, and the network topology of the interbank network.

Item Type: Thesis (PhD)
Additional Information: © 2023 Raymond Ka-Kay Pang
Library of Congress subject classification: H Social Sciences > HG Finance
Q Science > QA Mathematics
Sets: Departments > Mathematics
Supervisor: Veraart, Luitgard A. M.

Actions (login required)

Record administration - authorised staff only Record administration - authorised staff only


Downloads per month over past year

View more statistics