ABSTRACT

The aggregation of different single ratings into a so-called consensus rating in order to get a higher-precision debtor's default probability is an idea that is hardly discussed in the literature. In their 2013 paper "Deriving consensus ratings of the big three rating agencies", Grün et al came up with a method for rating aggregation, whereby the term "consensus rating" was introduced. To sharpen the whole issue of rating aggregation from a theoretical perspective, in their 2016 paper "Consensus information and consensus rating: a note on methodological problems of rating aggregation", Lehmann and Tillich developed a framework in which the terms "consensus rating" and "consensus information" are clearly defined. The paper at hand tries to connect the two aforementioned contributions and applies the theoretical framework of Lehmann and Tillich in connection with some of the practical ideas of Grün et al. In contrast to Grün et al, a simulation approach is chosen in order to have a clear benchmark for assessing the rating aggregation outcomes. Thereby, the following questions should be clarified. Does rating aggregation really lead to a higher precision of the estimated default probabilities? Is there a preferable aggregation method? Does the consensus rating, as defined by Lehmann and Tillich, outperform other aggregation methods? The simulation results show that rating aggregation could be a puzzling issue.