Lisa Goldberg (MSCIBARRA)

Credit derivatives constitute an important and enormous asset class, which is currently assessed at $24 trillion in outstanding notional value. The risk profile of a typical credit derivative depends on future losses to a portfolio of default sensitive securities. To evaluate and manage credit derivatives, we develop a top model of portfolio loss that is based on a self-affecting affine point process. In our model, risk is driven by economy-wide factors including the portfolio loss process itself, so past defaults influence future loss dynamics. Our specification takes account of the interdependence between the random loss at default and default timing.

In this lecture, we look at the empirical results of calibrating a Hawkes type affine point process to market data. We examine the quality of the model fit as well as the time evolution of model parameters, which are identified with different aspects of financial risk. The material for the talk was developed in collaboration with Eymen Errais, Kay Giesecke and Kao-Chih Syao.