Financial Stress Tests and Scenario Analysis: An Empirical Study of Common Risk Factors on the Swedish Market
Independent thesis Advanced level (degree of Master (One Year)), 20 credits / 30 HE creditsStudent thesis
Global financial crises throughout the history have led to banking austerities and extensive regulations on capital allocation and risk management. Stress tests and scenario simulations with the aim to investigate a variety of developments in the financial market have become common approaches for supervisory authorities and regulators for testing banks’ and financial institutions’ resilience towards financially stressed situations. However, these approaches are often considered hypothetical and argued to be hard to assess and lack a statistical framework. In light of this, the purpose of this thesis is to investigate an alternative method that includes hypothetical scenarios and account for correlation and dependency structure when combining risk factor developments in a stressed scenario on the Swedish market. Throughout this thesis it is argued that statistical methods are possible to combine with hypothetical developments by estimating distributions and modeling dependencies via Copula functions. Our findings imply that it is in fact reasonable to assume simultaneous extreme developments for a selection of risk factors yet with some restrictions and with regards taken to their correlation structure. We argue that previous stress tests on the Swedish market have been aggressive mostly when considering the development of property prices and long-term interest rates. Stress tests form the basis for decisions regarding capital planning and sizes of capital buffers, thus it is vital for credit institutions to conduct stress tests with plausible scenarios for such capital buffers to be of appropriate size. Due to high exposure towards the property market, it is vital for the major Swedish banks to conduct stress test with reasonable levels of stress especially for real estate prices. Our suggested methodology and model generate risk factor developments close to levels often set by supervisory authorities such as the European Banking Authority. However, we distinguish us from classic stress testing methodology since we identify stressed scenarios by generating joint developments for all risk factors and not only rely on hypothetical assumptions. Furthermore, it can be concluded that previous stress tests have been aggressive but not completely unrealistic, a finding that contributes to the controversial topic on stress tests that implies that hypothetical tests are often unrealistic.
Place, publisher, year, edition, pages
2016. , 97 p.
Financial Stress Testing, Copulas, Kernel Density Estimation
Economics and Business
IdentifiersURN: urn:nbn:se:kth:diva-189440OAI: oai:DiVA.org:kth-189440DiVA: diva2:946035