On the information content of sovereign credit rating reports: Improving the predictability of rating transitions☆

In order to identify novel qualitative determinants of transitions in sovereign credit ratings, we construct six different textual sentiment and subjectivity measures using dictionary-based, machine learning approaches on rating reports issued by Moody’s Fitch the period from 2002 2017. After controlling for macroeconomic fiscal strength, soft information, as well known sources proximity biases, find that, average, these text-based improve classification accuracy downgrades upgrades. The improvement is more notable than measures, compared Next, evidence that agencies seem follow through-the-cycle philosophy taking a longer horizon into account. Finally, best our knowledge, offer most comprehensive analysis their effect ratings thus far.