Principles of the Chain-Ladder “Method” Selecting and Updating Claims Development Factors

There has been significant discussion recently regarding the roles of “models” and “methods” in actuarial practice. I believe that much of this discussion is misguided as it is based on an imprecise and arbitrary distinction. I believe that “methods” are more appropriately considered to be a subclass of “models,” rather than a wholly different class of estimation procedures. More specifically, as with “models”, I believe that there are statistical assumptions underlying “methods.”

If we accept this conclusion, then it becomes incumbent on actuaries to apply statistical theory when using methods. The most common method is the chain-ladder method. In this paper, as an example, I re-examine the process of selecting and updating claim development factors under this new paradigm.