Partial orderings of default predictions

Walter Krämer (Technische Fakultät Dortmund)
Monday, July 6, 2015 - 2:00pm
Spandauer Straße 1, Room 23

We compare and generalize various partial orderings of probability forecasters according to the quality of their predictions. It appears that the calibration requirement is quite at odds with the possibility of some such ordering. However, if the requirements of calibration and identical right marginals are relaxed, comparability obtains more easily. Taking default predictions in the credit rating industry as an example, we show that for a data base of 5333 (Moodys) and 6178 ten-year default predictions (S+P), Moodys and S+P cannot be ranked according to most partial orderings, but that S+P dominate Moodys in the Gini-sense.