Speaker: |
Håkan Andersson |
Title: |
Applications of large deviations techniques in credit risk modeling |
Abstract: |
Credit risk is the risk that an institution incurs financial losses because some of its counterparties have difficulties in fulfilling their debt obligations. An important task for the risk management of the institution is to estimate the probability of very large portfolio losses. Since closed analytic formulas are not available in general, the quantification of portfolio credit risk typically involves problems relating to rare event simulation.
Using traditional Monte Carlo simulation in this context may not be feasible because of the sometimes very slow convergence of the estimator. This makes variance reduction techniques such as importance sampling potentially attractive. However, realistic portfolio credit risk models are designed to somehow capture the inherent dependence between obligors (e.g., default events of obligors belonging to the same industry sector tend to be positively correlated), and importance sampling does not immediately lend itself to such a setting.
In this talk we give an overview of recent results in this area and we also present some important open problems.
|
|
Organizing Committee
Anna Amirdjanova,
Department of Statistics,
University of Michigan Charlie Doering,
Departments of Mathematics and
Physics & MCTP
University of Michigan
Len Sander,
Department of Physics
& MCTP
University of Michigan
|