The expected equity premium is key economic parameter and part of many forecasting models. Berg and Kaserer (2009) derived a simple and robust method for estimating the equity premium using CDS spreads and structural models of default. The calibration of several input parameters like the proxy of default probability and recovery rates is still vital. This thesis deals with calibration of the default probability based on the ratings of the rating agencies. We consider two models to predict multi-period default rates for corporates rated by a rating agency: the Poisson regression and the linear regression. Outof- sample analysis of default events reveals that the Poisson model outperforms in one to three-year prediction horizon. We also introduce an ex-ante forecasting model for multiperiod aggregate recovery rates. Last but not least, for estimating the equity premium we implement each approach in the model of Berg. The estimated equity premium based on 5-year CDS spreads from 2004-2005 exceeds the results of Berg by merely 2 percent.
«