Synthetic CDOs - An introduction, evaluation and risk estimation

Abstract

In the last decade we have seen a tremendous growth in the issuance of derivatives linked to the credit risk of an individual, or a pool of obligors. While some have been around for a while, some are relatively new. One such derivative is called Synthetic CDO which is the topic of this thesis. The investors in these derivatives are protection buyers, or protection sellers of credit risk. Being a market instrument, the premium demanded for taking such risks is subject to market expectation about the present and future "credit health" of one or more companies. In the light of history, we seek to quantify the risk of actually experiencing the events triggering the contingent payment in such a derivative. This history, and the corresponding empirical event probabilities, are inherent in credit ratings issued by specialized rating companies. We use these probabilities in a multivariate rating transition model to capture these risks in the setting of synthetic CDOs. Also, investors are not only exposed to the risk of losses following such credit events, but also to the risk that the value of their position will change due to market expectation and general supply/demand factors. We therfore use market history on these instruments along with a standard pricing model to forcast a future distribution of market prices that are subject to the occurrence of such credit events. Aggregating all these factors, we calculate a \textit{Profit \& Loss} distribution for such an investment on a one-year horizon.