What do we know about rating action report?

Moody’s releases rating action report concurrently with its rating actions. In the rating process to assess an issuer’s credit profile, credit rating agencies (CRAs) emphasize the use of both industry- and firm-specific information. To justify their rating actions, CRAs concurrently release to the public rating reports with detailed default-related information and rating rationales. A typical rating report provides specific rating actions, detailed reasons that lead to a rating action, and scenarios that could lead to future rating changes. Prior literature uses an issuer’s current rating as a proxy for default likelihood (e.g., Cheng and Neamtiu, 2009; Becker and Milbourn, 2011; Cheng and Neamtiu, 2009; Bonsall, 2014; and deHaan, 2017) but largely ignores rating reports. However, rating reports could contain value-relevant information for investors.

To quantify different factors discussed in the rating reports, we exploit LDA, a topic modeling approach to quantify the different weights of topics that credit rating analysts discuss in rating reports. LDA is an unsupervised natural language processing technique to analyze the thematic content of texts (Blei, Ng, and Jordan, 2003). LDA helps researchers reduce the dimensions of the corpus from words to a limited number of topics without much initial analysis of the text. As a result, LDA is easier to scale up and replicable when applying to large-scale texts.

Discussion of the metrics topic om negative tone is associated with with significant positive market reaction and lower default likelihood within one year, which is a result of increased future debt-paying ability due to aggressive financial policy.

Discussion of the liquidity constraint topic in negative tones generates significant negative market reaction and predicts higher future default likelihood