Victor Lang och Richard Setterwall presenterar sitt examensarbete

Credit Risk Diversification using Default Correlation

Abstract

The new Basel II methodology encourage banks to improve their capital
models by reducing the minimum capital requirement according to how
advanced the risk management system is. Furthermore, credit risk
models referred to as Internal Rating Based models (IRB) are developed
within the banks and the sophistication level of the IRB model will
have an impact on the banks spectra of capital required per risk
tranche.

At the moment, most models calculate Capital-at-Risk (CaR) for
counterparties on a stand-alone basis. A need to compensate accurately
for diversification and concentration risk has grown and several
forums are conducted on the subject. This thesis will elaborate
different approaches of how to calculate and monitor diversified
CaR. Banks new exposures in different segments, of for example
geographic location and industry type, should be weighted differently
depending on the fit with the current portfolio, in terms of current
CaR within that segment and correlation between segments. Default
correlation is one of the key components in Capital at Risk. Hence, a
strong emphasis will be on how to calculate Joint Probability of
Default and then Default Correlation.

The resulting default correlation matrix is SEB specific but the
methodology explained in the thesis is universal and any analyst with
access to equity market data and PD estimates can implement the
methodology and calculate diversification multiples applicable to
their portfolio.

We will also develop an interface to facilitate investments
decisions. The interface will present portfolio managers with
risk-adjusted returns for each available bond and derivative within a
selected segment as well as a ranking based on the portfolio, of the
subset.



Keywords

Portfolio Diversification, Probability of Default, Joint Default
Probability, Default Correlation, Tetrachoric Series, Monte Carlo
Simulation.