A random effects ordered probit model for rating migrations

Employing a random effects ordered probit model, this paper examines the sources of heterogeneity in sovereign credit ratings in emerging economies. The analysis uses data from six rating agencies for 90 countries. The model highlights the importance of considering the cross-section error, which captures country-specific heterogeneity, in modelling rating upgrades. Watchlist status is a powerful tool in predicting future rating upgrades/downgrades, and dominates rating momentum in some cases. Rating duration and existing rating are important determinants of rating migrations. Evidence of inter-agency differences and dissimilar behaviour of upgrades and downgrades is presented.