dc.contributor.author | Skretting, Per Kristian | |
dc.date.accessioned | 2013-10-29T11:58:01Z | |
dc.date.available | 2013-10-29T11:58:01Z | |
dc.date.issued | 2013 | |
dc.identifier.uri | http://hdl.handle.net/11250/182222 | |
dc.description | Master's thesis in Industrial economics | no_NO |
dc.description.abstract | Value at Risk analysis is a widespread measure in the banking sector and other financial institutions for controlling risk exposure in the market. The historic simulation approach is the market leader in the banking sector, because of the simplicity of the model and the accuracy. Although there exist many more sophisticated models for estimating Value at Risk, most of them fail appropriate test criteria. When even the historical simulation approach fails the criteria, no better alternatives existed, until just recently.
In this thesis we investigate and seek to validate this new Value at Risk simulation approach, named ROM simulation. We differ from 2 models and they are termed historical ROM and deterministic ROM, and we test whether or not they pass the test given by Christoffersen (1998). Furthermore we test for different significance levels and on different portfolios. We also include the standard historical simulation which serves as a benchmark. The result from the backtesting favors the deterministic method over both the historical ROM and the historical simulation. The deterministic approach also proves to be more flexible than the other examined models. | no_NO |
dc.language.iso | eng | no_NO |
dc.publisher | University of Stavanger, Norway | no_NO |
dc.relation.ispartofseries | Masteroppgave/UIS-TN-IØRP/2013; | |
dc.subject | industriell økonomi | no_NO |
dc.subject | risk management | no_NO |
dc.subject | risikostyring | no_NO |
dc.title | Value at risk analysis by means of random orthogonal matrix simulation | no_NO |
dc.type | Master thesis | no_NO |
dc.subject.nsi | VDP::Technology: 500 | no_NO |
dc.subject.nsi | VDP::Social science: 200::Economics: 210 | no_NO |
dc.source.pagenumber | 49 | no_NO |