Erik Svensson presenterar sitt examensarbete

Risk Aggregation and Dependence Modelling with Copulas
 
Abstract:
 
This thesis addresses the problem of aggregating Economic Capital
for banks and financial institutions exposed to multiple risk
sources. Analytical expressions for the aggregated Economic Capital
are hard to find when loss distributions corresponding to the
different risks differ from normality. Besides the ordinary
monte-carlo simulation approach, the possibility to use copulas for
simulation and computation of Economic Capital is investigated in
this thesis. Copulas enable construction of multivariate
distributions where complex dependence structures between risks can be modelled
separately from the marginal behaviors.
Specifically, tail dependence and asymmetric dependence structures
can be constructed, being of particular interest in risk management. The
investigation of dependence between equity and bond indices
performed in this thesis shows the great advantages of using
copulas, but also some shortcomings that limit their usefulness for
risk aggregation purposes.