Reverse stress tests can be used to identify the scenarios that exhaust the capital buffer of a bank and cause the default of the bank. This report presents a framework to apply a quantitative reverse stress test on the portfolio of a bank. The framework includes the selection of appropriate risk factors, the identification of the distribution of these risk factors, including dependency, estimation of the influence of these risk factors on the occurrence of a default and a grid search to identify the set of stress scenarios. Applying this framework to a standardized portfolio leads to a set of stress scenarios, that can be used as early warning signs or as input for a traditional stress test. Furthermore, to decrease the computation time needed in the grid search used in this framework, the application of scenario reduction algorithms is analyzed. However, due to computational limits, it is not possible yet to reduce the scenario set in an efficient manner. Furthermore, the incorporation of higher frequency data for the risk factors is evaluated, using a Mixed Data Sampling (MIDAS) extension. As a result of assumptions made to prevent the scenarios set to increase even further, the more extreme scenarios are smoothed out, so no stress scenarios were identified using this extension.

, , , ,
Dijk, van . D.
hdl.handle.net/2105/43410
Econometrie
Erasmus School of Economics

Oers, B. van. (2018, September 19). Reverse stress testing: A framework to identify stress scenarios. Econometrie. Retrieved from http://hdl.handle.net/2105/43410