Modeling Consensus and (Dis)agreement in Rating Processes

Publication: ThesisDoctoral thesis


This dissertation introduces a general framework modeling common rating processes in order to aggregate rating information stemming from a variety of raters or rating sources. Ratings play an increasingly important role in our life. They are used to evaluate a variety of objects and activities all over the world. Here we apply our model framework to two different ratings, the credit ratings and the bookmakers odds. Whereas credit ratings represent the evaluation of credit customers or firms by banks or external rating agencies, bookmakers odds are prospective ratings of the performance of the participating players or teams in a sports competition. Despite the fact that these ratings are used in different kind of areas, both rating systems have a very similar underlying rating process. In both rating processes each rater estimates an underlying numerical variable which represent a probability or is directly related to a probability. In the case of credit ratings this probability is the probability of default (PD) of a credit customer or a firm and in the case of bookmakers odds this probability is the probability of winning a specifc sports competition. The proposed model framework is then used to solve the aggregation problem of the two rating processes for different applications yielding different model specifcations. Finally, the model results are used to validate the different underlying rating systems as well as for forecasting.
Original languageEnglish
Awarding Institution
  • WU Vienna
Publication statusPublished - 1 Oct 2011

Cite this