Journal of Risk

Risk.net

Rating migrations of US financial institutions: are different outcomes equivalent?

Huong Dieu Dang

  • This comprehensive study of the rating migration dynamics of U.S. financial institutions (FIs) finds that upgrades can be treated as equivalent within the same analysis while different downgrade outcomes require separate estimation models.
  • Downgrades to investment ratings have hazard rates that increase faster with time than downgrades to speculative ratings.
  • Within-rating heterogeneity and time-heterogeneity exist in the rating process of U.S. FIs. The effect of rating history varies across downgrade outcomes, and persists or becomes stronger after accounting for the current rating, outlook or CreditWatch. Of all migration outcomes, upgrades and downgrades to A- or higher ratings are more vulnerable to adverse macro-economic and political conditions.
  • The estimated dynamic outcome-specific hazard model which includes rating history, macro-economic and political variables exhibited strong forecast accuracy out-of-sample. Rating outlook and CreditWatch were not particularly useful in signalling potential rating changes during the holdout-crisis period 2007-2010.

This study employs a competing risks approach to examine the rating migrations of US financial institutions (FIs) during the period 1984–2006. It finds that downgrades to alternative major rating categories require separate models, while upgrades can be treated as equivalent in the same analysis. Different downgrade routes exhibit different within-rating heterogeneity and time heterogeneity. The effect of rating history persists, and for downgrades to speculative ratings it becomes stronger as controls for outlooks/CreditWatch are added. The rating history, macroeconomic and political cycles jointly exhibit discrimination accuracy in predicting the outcomes of rating changes during the holdout-crisis period (2007–10). In most cases, adding the current rating, outlook or CreditWatch does not substantially improve the forecast performance of the models out-of-sample.

To continue reading...

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here: