Priority of the research project: Current stress tests only consider a handful of scenarios and do not quantify the plausibility of scenarios. Furthermore, second round effects arising from the reaction of financial institutions to shocks are rarely considered. This current approach may lead to an illusion of safety, to premature reaction on alarming stress test results, and to the underestimation of risk.
 
In this project we aim to develop a second generation stress testing framework, which is both systematic and systemic. It systematically analyses a multitude of sufficiently plausible scenarios in order to identify the worst case for the financial system. It models systemic risk arising from fire sales and contractual networks, where financial institutions are affected by the reactions of others to the stress. Building on results in various strands of the academic literature, we propose to develop a robust model relying on available data and to implement a prototype software for systematic and systemic stress tests.

Content of the research project

We propose to develop the core elements of a modern stress testing framework:

  • The automated evaluation of a large number of scenarios reduces the risk of missing out dangerous but plausible scenarios. All current stress tests use only a handful of stress scenarios. There are no procedures in place to check whether other, equally plausible scenarios might be more dangerous.
  • The quantification of the plausibility of scenarios prevents consideration of highly implausible scenarios. Since all stress tests used by authorities at the moment avoid the quantification of the probability that particular scenarios might occur, they are prone to trigger premature reactions to harmful stress test results in implausible scenarios.
  • The use of distribution scenarios allows for a unified treatment of trading book and banking book.
  • Modelling fire sales allows for a realistic quantification of the consequences of bank reactions to stresses. Despite the known fact that the most important mechanism by which stress manifests itself in a crisis are deleveraging processes, stress testing frameworks in place fail to take account of this fact.
  • Modelling of network contagion effects allows for a realistic quantification of consequences of bank defaults. Current stress testing frameworks ignore contractual interconnectedness. This impairs the credibility of the stress tests.

Applied Research Methods

We briefly describe the methods to develop the five core elements mentioned above:

  • The automated evaluation of a large number of scenarios is an exercise in modelling, data acquisition, and data handling. We try to develop models which are simple enough to allow for fast, automated scenario evaluation. This requirement must be balanced with the amount of detail and accuracy required in scenario evaluation.
  • The plausibility of point scenarios (realisations from the joint risk factor distribution) will be quantified by their Mahalanobis distance, as proposed first by Studer and developed by Breuer and Krenn. The plausibility of distribution scenarios will be quantified by their relative entropy or some other f-diverdence, as proposed by Breuer and Csiszar.
  • The determination of the worst case distribution scenario will be implemented along the lines sketched in Breuer and Csiszar. The determination of worst case scenarios for the combined trading and banking books is still to be solved.
  • Fire sales, their price effects, and the consequences for financial institutions will be modelled along the lines of Cont and Schaaning (2016). This will require data first about the balance sheets of all major financial institutions in the system, and second about market depth of the major asset markets in which these financial institutions are active.
  • Network contagion effects will be modelled along the lines of Noe and Eisenberg, and implemented following Elsinger and Summer.

Facts

Project nameSystematic and Systemic Stress Tests
ProgrammeJubilaeumsfonds der Oesterreichischen Nationalbank
TopicStress Tests
Project index number17671
Project durationSeptember 2018 – July 2021
Project budget141.000,— EUR

Advisory Board

Dr. Martin Gächter
Finanzmarktaufsicht Liechtenstein

Dr Martin Summer
Oesterreichische Nationalbank

Mag. Claus Puhr
Oesterrreichische Nationalbank

Prof. Dr. Branko Urosevic
Universität Belgrad

Dr Eric Schaanning
European Central Bank

Contact

Prof. (FH) Dr. habil. Dipl.-Phys. (ETH) Thomas Breuer
Lecturer

+43 5572 792 7100
thomas.breuer@fhv.at

Details

Vineeth Naik, MSc, B.Tech
Research Assistant, Josef Ressel Centre for Applied Scientific Computing


Details

Mag. Dr. Martin Jandacka
Mathematical finance

+43 5572 792 7101
martin.jandacka@fhv.at

Details