Risk management using the Bernstein copula: modeling and goodness-of-fit
Type
applied research project
Start Date
January 1, 2010
End Date
December 31, 2012
Status
completed
Keywords
Simulation
Dynamic Financial Analysis
Description
The aim of this project is to illustrate the modeling of non-life insurance risks
using the Bernstein copula. Therefore we first conduct a goodness-of-fit analysis
and compare the Bernstein copula with other widely used copulas. Further,
the use of the Bernstein copula in risk modeling is illustrated in a Valueat-
Risk context. In both analyses we utilize German claims data on storm,
flood and water damage insurance for calibration. Our results highlight advantages
of the Bernstein copula compared to other widely used approaches,
among which are its flexibility in mapping inhomogeneous dependence structures
and its easy use in a simulation context due to its representation as
mixture of Beta densities. Both practitioners and regulators working toward
an appropriate modeling of dependencies in a risk management and solvency
context can benefit from our findings.
using the Bernstein copula. Therefore we first conduct a goodness-of-fit analysis
and compare the Bernstein copula with other widely used copulas. Further,
the use of the Bernstein copula in risk modeling is illustrated in a Valueat-
Risk context. In both analyses we utilize German claims data on storm,
flood and water damage insurance for calibration. Our results highlight advantages
of the Bernstein copula compared to other widely used approaches,
among which are its flexibility in mapping inhomogeneous dependence structures
and its easy use in a simulation context due to its representation as
mixture of Beta densities. Both practitioners and regulators working toward
an appropriate modeling of dependencies in a risk management and solvency
context can benefit from our findings.
Leader contributor(s)
Funder
Topic(s)
Copulas
internal risk models
Method(s)
Simulation
Dynamic Financial Analysis
Range
Institute/School
Range (De)
Institut/School
Division(s)
Eprints ID
206697
results