Using Copulas as a Risk Management Tool
Independent thesis Basic level (degree of Bachelor), 10 credits / 15 HE creditsStudent thesis
Traditionally, when analyzing a portfolio with a multidimensional approach, a multivariate Normal distribution is used. This distribution is determined by the marginal distributions of the individual assets and the linear correlation between them. This thesis questions this model. Returns on assets often has a stochastic distribution with considerable heavier tails than the Normal. Moreover the linear correlation is not able to explain non linear dependence. The thesis propose the alternative dependence measures rank correlation and tail dependence. Further it explains how advanced dependence structures can be modeled using copulas. The theory is is applied on a portfolio of mutual fund shares. The estimated Value-at-Risk and Expected Shortfall for the portfolio is, respectively, 11% and 33% larger when using a copula approach instead of a multivariate Normal distribution.
Place, publisher, year, edition, pages
IdentifiersURN: urn:nbn:se:su:diva-5074OAI: oai:DiVA.org:su-5074DiVA: diva2:194800