Credibility as Data Augmentation

Credibility theory is central to the actuarial tasks as a method for "borrowing strength" from external data or expert judgment. This paper will look at the interpretation of credibility using data augmentation; this allows for an intuitive meaning for the "K" parameter in the Bayesian approximation, and an alternative method for implementing credibility in insurance application such as ratemaking with Generalized Linear Models (GLM) and severity curve-fitting.