In most CDS pricing approaches, recovery is assumed to be constant and exogenously given, whereas default probabilities are inferred from market quotes. In this thesis, three different credit models are presented which can be used to bootstrap implied recovery information under the risk neutral measure Q from the term structure of market spreads. None of the models requires to assess default probabilities or the recovery beforehand. The models are applied to pre-default market data of overall seven companies which experienced a default in 2011 or 2012. The risk neutral implied recoveries are then compared to the corresponding CDS auction results. The first approach is based on a Cox-Ingersoll-Ross reduced form model with a stochastic recovery rate that depends on the time of a credit event. In the second model, the stock price follows a default adjusted Cox-Ross-Rubinstein Process as in Das and Hanouna [2009], where default intensities and the recovery are both deterministic functions of the stock price. Both models are shown to suffer from an identification problem between implied default probabilities and implied recoveries, leading to unreliable estimates. Only the third model, which strategically avoids joint modeling of default probabilities and recovery by considering the ratio of two CDS spreads, yields more satisfactory implied recoveries.     «

In most CDS pricing approaches, recovery is assumed to be constant and exogenously given, whereas default probabilities are inferred from market quotes. In this thesis, three different credit models are presented which can be used to bootstrap implied recovery information under the risk neutral measure Q from the term structure of market spreads. None of the models requires to assess default probabilities or the recovery beforehand. The models are applied to pre-default market data o...     »