eThesis - Kauppakorkeakoulun sähköiset gradut

The purpose of this study is to investigate the relationship between rating changes of two American credit rating agencies and the European credit default swap (CDS) market. The first objective is to study the information content of rating changes. In other words, it explores whether there is a response to a rating change announcement in the CDS market on the announcement day of an up- or a downgrade. I also investigate how the CDS spreads move before rating change announcements which gives an idea of the amount of information absorbed by CDS spreads before the actual event date. Finally I study whether the CDS spread movements could be used to assess probabilities of upcoming rating changes.

The data consists of rating change announcements from Moody's and Standard and Poor's in years 2008-2012 collected from Thomson Reuters -database. The data on CDS spreads contains European companies and is collected from Datastream covering a time period from December 2007 to May 2012.

The methods used in this study include an event study and logistic regressions. The event study is used for investigating spread movements 90 business days before and after a rating change announcement. An event window of -1, +1 day around the event is used for reporting the announcement effect. Logistic regression is used for studying the effect of spread changes on the probability of an upcoming rating event. Probability sensitivity measure states the increase in the probability of a rating change for every one basis point change at the mean.

The results reveal significant positive spread changes before a downgrade for both rating agencies. Preannouncement spread changes for upgrades are only significantly negative for companies rated by Moody's. The announcement effect is highly significant for S&P and more extensive than for Moody's. The total sample of S&P's ratings shows a significant reaction to both upgrades and downgrades whereas the total sample of Moody's ratings does not reveal a significant reaction to either upgrades or downgrades. In general I do not observe significant spread changes after the events. The logistic regression provides evidence that spread changes can be used for predicting rating changes.