PDA

View Full Version : How to reconcile asset share and GLM for determining relativities?


asile
03-20-2012, 11:08 PM
Both the asset share model, considering persistencies and costs by class, and the GLM approach, analyzing multiple variables at once and adjusting for "statistical credibility" seem like valid approaches to determining rate relativities.

What are the advantages of one method over another? An obvious one seems that persistentcy rates are difficult to estimate in practice. When would you use each method? Is there a way to combine them (maybe use year of policy, 1 for new, 2 for first renewal, etc as a variable in Glm analysis)?

Seems like this type of question could be valid under bloom taxonomy and looking forward to your inputs.

asile
03-20-2012, 11:30 PM
Or, do you use GLM to estimate the true loss for each cell in your class plan and asset share to determine a debit or credit for each class? Seems like even in rate factor selection, analyzing all classes simultaneously would be advantageous.

Vorian Atreides
03-21-2012, 12:17 PM
Agree that the question could be asked on an Exam. And that it'd be a good question to see if you understand the two methods.