Default probabilities, CDS premiums and downgrades : A probit-MIDAS analysis

This paper examines the relationship between sovereign credit default swaps CDS and sovereign rating changes of European countries. To this aim, a new estimator is introduced which merges mixed data sampling MIDAS with probit regression. Simulations show that the estimator has good properties in finite sample. Also, I investigate a bootstrap procedure introduced by Ghysels et al. 2007, which should be able to handle significance testing in a MIDAS setting. The bootstrap hasgood size but low power. For the empirical analysis I use sovereign CDS data for 22 EU countries trying to correlate sovereign downgrades with sovereign CDS premiums. Overall the CDS data and the ratings are in most cases significantly positively correlated. Therefore, Credit Rating Agencies CRA and financial markets are generally agreeing on the implied default probability of sovereign nations. Also, CDS prices are expecting downgrades in advance in the majority of investigateddatasets. However, this does not mean that a default probability can be extracted from raw CDS prices. Instead, by using a MIDAS estimator, I significantly reduce the amount of noise in the data. Therefore, CRAs are still providing important information to financial markets.