Institutional Repository of Vilnius University
Kredito rizikos modelių kiekybinis rizikos reikšmingumo vertinimas /
Abstract
dc:descriptionModel risk quantification models are a new and underdeveloped field, even though in institutions dealing directly with money, models are used to decide wheter a credit should be granted or not. As the use of such models becomes more widespread, it is important for financial institutions to weigh up the potential cost of errors in the models they use. Therefore, this thesis proposes a model risk quantification model for credit risk models based on the loss distribution approach, widely used in operational risk. The thesis addresses the lack of data problem on model risk events and their losses. In the search for the most appropriate distribution to model loss severity, a Gaussian mixture model is proposed as an alternative to the models traditionally used in operational risk models. It is observed that, of the five models investigated, the application of the Gaussian mixture model in distribution modelling is more accurate than traditional theoretical distributions for four of these models. The thesis concludes by presenting the overall results of the loss distribution approach and the annual expected losses, 99.9 quantile value-at-risk, expected shortfall and economic capital are presented, as well as comparisons between the individual models. A sensitivity test of the model shows that only the expected loss is proportional to the number of risk events in the model, while all other indicators change less than the number of loss events. Economic capital proved to be the most stable.
Degree
thesis:*- Grantor dc:publisher
- Institutional Repository of Vilnius University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Auga, Aurimas,
Rights
dc:rights- Statement dc:rights
-
- info:eu-repo/semantics/openAccess
- Language dc:language
- lit
Identifiers
dc:identifier.*- Repository record dc:identifier
- https://repository.vu.lt/VU:ELABAETD192844562&prefLang=en_US