• A
  • A
  • A
  • ABC
  • ABC
  • ABC
  • А
  • А
  • А
  • А
  • А
Regular version of the site

Working paper

Studying Replicability of Aggregated External Credit Assessments by Public Information

Kurbangaleev M. Z., Zinaida V Seleznyova, Lapshin V. A.
In this paper, we examine if the aggregate rating constructed as a consensus of individual credit ratings can be accurately predicted with publicly available non-rating information about company. Adopting approach from computational social choice we construct consensus of ratings assigned by seven credit rating agencies to Russian banks in national scale and compare it with several proxies based on publicly available characteristics of those banks. We measure how much aggregate (consensus) rating and proxies are agreed in terms of ordering banks by their credit quality and discriminatory power in predicting defaults over the one-year horizon. We show that aggregate (consensus) rating is comparable to financial-data-based econometric default model in term of discriminatory power, but as ordering the former have a fairly low agreement with the last. We also found that using models for predicting initial credit ratings allows for building a proxy that has practically high agreement with the original aggregate rating, but original aggregate rating outperforms proxy in terms of discriminatory power. It was also found that greater agreement between original aggregated rating and proxy can be achieved on the subsample of investment grade ratings.