Rating philosophy and dynamic properties of internal rating systems: A general framework and an application to backtesting

The paper draws a general framework for asset and default dynamics, separating the influence of the economic cycle into a component which is embedded in the rating system and an unobservable risk factor that determines the movements of defaults around the ex ante estimated PDs. The two components – the sensitivity of ratings to credit cycle and conditional asset correlation - can be quantified through a Maximum Likelihood approach, giving a measure of the cyclicality of the rating system, and allowing for a number of applications: among those the modified binomial test proposed here.