Emperical bayes credibility theory

Discussion in 'CT6' started by DevonMatthews, Jul 26, 2010.

  1. DevonMatthews

    DevonMatthews Member

    I'm having some trouble understanding this horribly incoherent load of jargon published institute in the core reading. I'd like to know what the exact assumptional differences between EBCT model I and II are,

    Under model II, since the Xj's given theta are NOT identically distributed, then why would E(xj|theta) and Pjvar(xj|thera) not depend on j? Why would assumption 2.2 be a weekend form of it's model I equivalent only to go on and make 2 more assumptions that are heavier?

    Appreciate any explanation between the difference in assumptions between model I and II, read chapter atleast 5 times, do not understand. Thank you.
     
    Last edited by a moderator: Jul 26, 2010
    DevonMatthews, Jul 26, 2010
    #1
  2. John Lee

    John Lee ActEd Tutor Staff Member

    It is a bit grim.

    Bottom line is Model 2 is no longer saying they are identically distributed but it still wants them to have the same mean and variance (over the different years). So together the assumptions are weaker than Model 1.

    Exam questions do ask for you to list the assumptions. But most are of the number crunching kind (ie stick values in the formulae given on pages 29 and 30 of the Tables.
     
    John Lee, Aug 2, 2010
    #2

Share This Page